/robowaifu/ - DIY Robot Wives

Advancing robotics to a point where anime catgrill meidos in tiny miniskirts are a reality.

LynxChan updated to 2.5.7, let me know whether there are any issues (admin at j dot w).

Reports of my death have been greatly overestimiste.

Still trying to get done with some IRL work, but should be able to update some stuff soon.


Max message length: 6144

Drag files to upload or
click here to select them

Maximum 5 files / Maximum size: 20.00 MB


(used to delete files and postings)

Welcome to /robowaifu/, the exotic AI tavern where intrepid adventurers gather to swap loot & old war stories...

Open file (477.54 KB 975x974 knjdhu2bi2131dsa.png)
Open file (821.22 KB 850x605 mb2dh12y7v812.png)
Madoka.mi prototype thread AllieDev 09/17/2021 (Fri) 19:58:39 No.13288
As was suggested in another thread after I am making a model of this image and will attempt to 3D print it into a working prototype and am following the suggestion of Kiwi as to not derail the Actuator thread. >>13205 >>13216 The goals so far will be to test whether or not the average robowaifu technician can simply design and 3D print their own parts. At the end I hope to be able to share what I've made. This is also a good thread for any Anons to share their Digital 2D/3D or 3D printed creations of robowaifus or robowaifu parts. >=== -minor edit of subject
Edited last time by Chobitsu on 11/29/2021 (Mon) 06:55:58.
We certainly look forward with eager anticipation to your project's progress OP. Godspeed. Also, I am going to modify your OP's subject slightly to let everyone else know this thread will also serve as a 3D modelling/printing project-progress general, as you indicated. This has been an ongoing need for the board, and you have our thanks for filling it.
Open file (420.41 KB 386x1092 MadoHeight.jpg)
>>13288 Nice work on her guard plate for her side rocket booster. Her mechamusume design does not lean into walking with everything on her back. I'm thinking wheels on the ends of her wings would be a good way to get her moving. Open to your ideas AllieDev. Madoka's official height according to show notes is 152 cm (5'0").
Open file (73.88 KB 900x675 8b9ey1hv9212.jpg)
Open file (48.61 KB 740x416 124231rdcr32e1.jpg)
>>13329 I have looked into other mecha and feel that a design where the ends of the "dress" wings could pop out and reveal a wheel arm that could fold out through an edited design like on the bottom right. Though at that point small metal hinge parts would be a necessity to keep it together. Or if we're going the practical route I could make a simple door that looks like generic paneling that can be opened to reveal the wheel arm. I'm a fan of magnets but I have seen other designs where you can hide the wheels. Mecha designs are surprisingly extremely flexible if you are creative enough. Have to finish modeling madoka herself first since making human bodies from scratch is the most time consuming part. It helps that I don't have to spend any time making toes or fingernails in her mecha form. https://www.gundamkitscollection.com/2015/05/custom-build-hguc-1144-gundam-rx.html?m=1 I Also searched up what the RX-78GP03D Booster system looks like after I was stuck on the small details like the inside of the engine turbine thinking it went deeper and had at least 2 layers under the guard plates like some other mecha designs do. The comically huge size comparison to the RX-78 was hilarious too and the real challenge will be to balance the wings so they are movable and bendable but wont fall off or break apart. Looks like a simple hinge joint would work half way on the wings with some kind of pivot being worked into it. I am no expert on Gunpla so further research into the many many variants and designs will probably already have a solution to that like with Gundam Wing. The best picture I could find of the connector area between the missilepods/wings and the booster rockets/dress was the second picture and the artist's design leaving some parts to the imagination. It doesn't look like wings are connected directly to her back here so that leaves room to make a simple connector with some details added in.
Open file (55.09 KB 400x600 foldOutWheel.jpg)
>>13333 >double dubs naisu get A generic paneling door where her motors could pop out of like landing gear would blend in with the rest of the aesthetic. I think for the sake of simplicity, it would be best if her "landing drive" was like a simplified version of pic related. Her owner would fold out her wheels when she'd need to move. Really liking your design thoughts, I've been looking into fishing line and other ways of securing her wings as well.
>>13333 Portentous quad digits OP. Good luck with Madoka AllieDev, Kiwi, & everyone else!
After some digging online, I found the original image file for the mecha-Mado and the artist created a mecha-Homu. Hope these higher quality images can help you AllieDev. From what I can tell, it seems her wings and rocket "dress" are all one unified backpack that's connected to her. I'd imagine making it a press fit ball joint would be ideal. There will definitely need to be point of articulation for her rocket "dress" unless you'd want her to appear as if she's floating.
Open file (2.22 MB 1920x1080 hubcap lol .png)
>>13374 That was the plan, did not want to double post yesterday so I went ahead and make some progress on the model and did not have anything else to share at the time. Was thinking of an alternate connector where you have to attach magnets to the insides of the hip region and the back of her skirt to hold her in the booster frame. While also having a version of the parts where it connects traditionally with a press fit ball joint hidden by the sides of the dress. The artist's design leaves some things to the imagination and most if not the entirety of the booster frame can simply connect to each other from what I am seeing here. Especially with the connector going most of the way up her back. Would probably have to require clear plastic for the wing glows and modify the design so it can be opened up to put LED's in there since its an unspoken requirement of robowaifu law to put LED's in your robowaifu somewhere. I modeled most of Madokami's body today, the only really hard part, currently its in a state like in this picture. >>12354 When I got to a sufficient stopping point I made a few greebles for her side booster rockets. Could use a madokami reference sheet where she is standing facing the camera. For now, I am currently using the magia record L2D model as a reference and starting point, but the only problem is that her feet are cut off and I need that size reference to be able to accurately translate it into the robot feet in the mecha design.
>>13374 Thank you for finding these Kiwi. It helps to be able to see more details in the design of the official version. That Mecha Devil homura design looks cool too.
Open file (378.17 KB 936x1286 MadokamiReference.jpg)
Open file (83.49 KB 2100x1192 SchoolMadokaReference.png)
Open file (103.29 KB 2100x1192 MagiMadoReference.png)
>>13376 >magnets That is is a very interesting idea and I really do think it deserves investigation. I'm not sure how well magnets would work unless she was small. >entirety of booster frame connecting Definitely would be the ideal way of going about it. We should develop a standard connector for ease of prototyping. I'd recommend ball and socket joints for ease of integration into designs while allowing for some "give" in her assembly. >lightpiped glowing wings That would look awesome AllieDev >References Glad to provide official references of whatever you need. >>13389 I'm back, I've been distracted by family matters. For now I just realized our ideas are somewhat nebulous. We should create a spec list for Madoka.mi. My current idea for her: Height: 1:1 scale 152 CM Head: 3 DOF with a display for her eyes. Arms: 3 DOF with basic grasping hands each. Torso: 2 DOF Legs: 3 DOF each with fixed position feet. Her wings, rocket "dress", side I-field generator, side mega beam cannon (we should probably replace with another I-field generator for symmetry), and the mechanism to pop-out her drive wheels should all be manually positioned to save on her weight and costs. For her drive system, both front wings of her rocket "dress" will have a motor enabling her to drive tank style. The back wings that are parallel to her outermost rockets will have casters, a roller ball, or some other method of enabling her to have four points of contact with the ground for high stability. Her wings will of course have lightpiped LEDs.
Open file (616.58 KB 1005x656 Madoka.mi eyes.png)
Open file (66.15 KB 620x414 cartreads.jpg)
>>13390 Its understandable Kiwi. It does help to have some basic measurements and specs so it does not drag out because of loss of momentum with no direction. Like how much power would she need to be operational for up to 16-24 hours on a single charge? I believe that since rechargeable giant electric battery powered robots actually work In real life, that the dress should be like a portable charging station and the dress doubling as a charger she can be detached from. Though the dress reminds me more and more of the shagohod every day with the absolute extra mass being dragged behind her with a tiny fraction being the heart of the whole thing. Was going to suggest a deployable small tank tread instead of airplane wheels for the ground since the frame is large enough to house them and would be simpler to implement. These are the size I was thinking of since if we are going by her 152 cm height, then the booster frame would be roughly about the size and length of a small car. The DOF make sense and I just needed the feet measurements as the hands and feet are the only things missing on the basemesh. Then comes Breaking it down into the various robowaifu sections. No Hair yet as that comes later in the process. It also appears her ears were replaced with robotic sensors in the design. Will also rework these to be more detailed but I felt they would look more round than being just a flat gundam panel since hearing is important for all robowaifus and it resembles a chobit's ears with functional utility vaguely. Would like a second opinion on this as I can also just make a boring flat gundam fin.
Open file (48.26 KB 800x715 Ascento.jpg)
>>13391 >How many watt hours would she need for 16-24 hrs? It would honestly depend a ton on her mass. This was a major reason why her current specifications have minimal DOF to lower actuator mass. Given that her whole Dendrobium dress is about half a sedan, she'll likely need a hacked hoverboard to move her around. Given that a hoverboard can last for around an hour of continuous use, you'd need 12 to have 12 hours of use. >Tank treads being simpler then wheels. I don't understand how. >hair A wig would work fine >ears The ones you've designed seem just fine. Making them open in the middle to reveal her ports like Chobit ears would be cute and functional. Also, good job modeling in her tareme, really gives her the cute and sweet look. Given the sheer volume (and implied mass) of her Dendrobium dress, making the dress a charging station makes sense. The motors needed to propel Madokami would then be significantly smaller and lighter weight, her whole power system would be simpler and longer lasting. We could hide her drive wheels in her heels, she'd function like ascento (picrel). Her dress would still have wheels for mobility but, they could have generators in them. This way, as you walk around holding your waifu's hand, she'd be recharged. All this being said, I think it would be best if we focused on getting her main body functioning first, we can then build her Dendrobium dress after, if you do think they should be separated.
Open file (2.35 MB 1920x1080 Basemesh.png)
>>13392 >good job modeling in her tareme Thanks, Thought I overdid the eye shape when I was making the outlines. This is not to scale yet beyond the basemesh and the hoverboards would make more sense than implementing several wheels like Ascento's. I made some of the outfit and the hands for the basemesh today. Attempting to bridge the upper chest area together was frustrating with blender just not wanting to rotate the curve to make the center crystal area of the dress and will end up remaking that part. Was happy to get the hard part of the hands out of the way and The Shoulderpads look separate from the body's dress and seeing how gundam does their shoulderpads, is likely the case here. Wish there was good sculpting tutorials since most of them are not that great and I've been using the proportional editing tool more often than even attempting to sculpt again. Though I could benefit from using those tools. I also did not make the feet yet since it may be simpler to just make robotic shoes that fold a wheel down than model feet that wont be practical to every day movement. Most real world bipedal humanoid designs that can actually walk and stand upright have some kind of flat ski design to them to distribute the weight properly. The mecha madokami design also gives me some concerns around the ankle area with that large spike shape sticking into the foot when standing flat on the ground and I do not see any of the heel areas so I assume its either similar to the shape of a horse's hoof, but stretched out, or built around the basic shape of madokami's foot inside a shoe. Either way the feet here do not look organic at all with the screw ankle joint. I also stretched the legs to be slightly longer to match the design here. Though they are not twice the length of the arms like they drew her. Could you find a madokami version of that madoka production sheet? Would help to have a consistent palette.
Open file (249.41 KB 1280x756 MadokamiProductionSheet.jpg)
Open file (179.00 KB 792x1087 UltimateMadoka.jpg)
>>13433 >Overdid the eyes It truly is hard to say until she's made real. Adding eyebrows would be helpful for determining how they should look. >Ascento Has only two wheel and two servos to contract the legs. It's remarkably agile for its simplicity. >Basemesh Looking good so far. Definitely looks like Madokami, actually prefer her looking more accurate to her animated look compared to the Mecha Madokami illustration. I feel your pain with Blender, I usually just make a base in VRoid Studio then save it as a .glb file and import into Blender. >Shoulder pads Her shoulder pads and other accessories would only make sense as separate pieces. It would be better for printing to make them slot in. >Feet Walking is best avoided at this stage. Rolling is significantly more efficient and easier to implement. Also, her ankle wings should be simplified so she doesn't stab your legs. Actually, cutting them off where the gem is and putting wheels there would be ideal. I did some math on her energy consumption today. I had realized I would need to take her on a plane. She would thus need to have a battery of 100 w/hr or less. If her main bodies mass was 5 kg, with 6 cm wheels, her drive motors would need a speed of 955 rpm with .42 kgf cm of torque to match human walking speed (3 m/s) on mostly level ground. This would consume 4 watts in an ideal world but, 6 is more likely given motor losses. If her OS runs on a smartphone, she would be able to run for 8+ hours continuously. I think that would be a good goal for her main body. Naturally, I'm very interested in your thoughts on the matter. I'd like to think >5 kg is acheivable with her main body but, I'm more a math and electronics specialist, you seem to have a better grasp on modelling. Of course, we could make a spot on her back to quick exchange batteries as needed with capacitors to smooth out the swap. >Madokami production sheet I found a production sheet and official art that would be helpful if you used the Color Picker in GIMP.
Open file (170.98 KB 981x426 Shoulderpads .png)
Open file (332.61 KB 738x904 basemesh 2 .png)
Open file (101.59 KB 645x599 Madokami dress sketch.png)
>>13436 I made some progress on the shoulder pads and skirt today. Will remake the chestpiece tommorrow. I also made a rough draft of what I believe would be a decent connector for the dress on her back with areas for both magnets and ball & socket joints since that artist only made one isometric drawing of this design. Though the spinal cross section is optional and I thought it looked great with a small back part that extrudes so the ball and socket connector can lift her up with that and the two lower ones keep her firmly in place. The circular lights are supposed to be where the magnets go along with the strips along the spinal cross section. Would be scaled down to fit along the middle of her back of course. Skirt was a pain and there were probably better methods than the flat modeling technique I was using. Happy to have gotten the basic shape of it for now. I feel that using something like Vroid would be "cheating" and I re-thought the ankle situation. What if when the foot is bent upwards or lands on the ground, that diamond section lowers slightly to fit the design and not get scratched/stabbed by the pointy lower legpiece? Some gundam and mecha designs actually do have that kind of mechanical design functionality for them when they transform and its far simpler to implement than that.
Open file (25.55 KB 369x400 SkaterMadoka.jpg)
>>13443 Those shoulder pads and skirt look pretty good. She really is starting to look like the picture. Her back makes a lot of sense. Using three ball sockets surrounded by slots for magnets makes sense given her dress will be designed for either ball joint or magnetic connection. Overall, good design for her back, though it is good to remember she needs to be as simple as possible for feasibility. Her ankle wings are an engineering nightmare honestly. The tiny connective point, being as long as her legs, looking stabby in general, I think it looks cool but not nonsensical. There should be two designs for her ankle wings, one that's accurate and one that's miniaturized without sharp edges. Both should be on a pivot that can be tightened to position as her Anon sees fit. I'd recommend using a screw for simplicities sake. I understand the mechanism you're talking about, it just would add to her complexity when a simpler solution exists. I'm still working on her internal hardware. I was thinking to connect her dress via I2C over a cable but, having three ball joints/magnets with metal connectors would be perfect for a seamless connection. So far, I'm planning on using a smartphone to run her personality and for various sensors (her "brain" or "Soul Gem". Using an ESP32 to control her body and react to stimulus so she doesn't hit things, knows when her hand or other things is touched, controls her motors and servos (her "brain stem and nervous system"). I originally wanted to integrate batteries and a USB C charger but, having replaceable USB battery packs is both simpler and would make sure she always could get a new battery when the old one eventually stops working. Need to have a good idea on her standalone bodies mass before deciding on actuators for her. I guessed 5 kg based on her being mostly empty plastic but, if you have a better idea on her mass, it would be helpful. Her dress is simple, just a hacked hoverboard integrated into the dress with an I2C that'll allow her main body to control it.
>>13445 I understand the need for simplicity. Though I have modeled and 3D printed more complex models before. The "new" back design I made is still relatively simple and fitting with the original design. Though I was trying to avoid drilling a hole into her back directly at all costs and there are better solutions to that problem than wasting time on a quirk in the design that no one will notice. We both had the same idea of using her soul gem section as an external memory unit. Though I still believe an onboard backup in the head would be necessary as those externals would be potentially exposed to inclement weather. I had also drawn another sketch of some places that may make sense for external battery storage with the huge sections on her hands that could also be used for item/spare part storage. Would be similar to replacing a phone battery but with the same connectivity as the soul gems. I assume connection ports would be necessary for some part of them? As cool as wireless charging can be, its just not as effective as traditional plugged in connection. For weight, you would have to get the average weight of a hollow shell of given skin material her size, then add the weight of each internal and external component you plan to add, including the skeleton. Depending on the 3D printed material, I assume the prototype is not going to use very much CNC cut stainless steel and will be miniature as a proof of concept. You would have to weigh the amount rolls of filament you used instead of what is on that list. Would make sense to have a Doc of some kind to keep track of the specs. With her dress, locomotion would be very easy being pushed along with her rear wheel drive. Perhaps taking a page out of the knightmares from code geass and Ascento, it can be possible to salvage that nightmare of pointy and thin ankle wings even on her official merch by making them thicker and at least connected by a metal universal joint with wheels at the end of the wings and a deployable wheel under the soles of her feet like the ones you were suggesting. Would be very rough but it would make her stable like a small car with 6 points on the ground to start with and go towards walking from there.
Open file (304.02 KB 1600x1145 GundamMadoka.jpg)
>>13448 Backups can be made via flash drive, we'll add a USB port to her so she can back up her memories nightly. Putting this in the back of her head would be sensible. I was thinking a hatch on her back with a USB C cable to plug in her battery. This would keep the mass localized to her torso. Using her hands for storage is an interesting idea. Making them magnetized would be nice when having her help in the shop. I'll look into manufacturing low mass materials. I print in PLA which can work but may not have an ideal mass at full scale. I'll investigate materials. Actually, using her ankle wings with wheels like knightmare frames is an ingenious way to give her balance while retaining accuracy to the picture.
>>13449 I print in PLA+ which is a stronger version of PLA and have to replace my printer's PTFE tubing. Still looking for a reliable brand for the tubing and couplings. Have not replaced a ptfe tube before and there is much mixed info out there and very confusing sorting through all the shills. Would have to take some creative liberties for her ankle areas as even her official merch sculptors have a hard time making durable enough footwings that don't break off easily. Would be just reinforcing that thin area with a detachable area like the Sutherland and making it fit in with the already huge shoes compared to her ankles. It's also why I believe a deployable wheel would make sense around the soles of her feet and would be difficult to see normally.
Open file (258.41 KB 1470x645 testMass.png)
>>13453 >Replacing PTFE tubing As long as the internal and external diameter are in spec, you're fine. Capricorn tubing is supposedly superior but I've used generic PTFE with success for years. Definitely agree that her deployable wheels should be in her soles. As for the ankle wings, I think it's best if we continue with the knightmare frame idea. After finding a similar model online and forcing cura to render it at almost 152 cm, her frames mass could be kept near 4 kg if she were hollow PLA. Still looking into ways to lower this, 3 walls is needed at these scales for a minimum acceptable level of strength.
>>13454 Large hollow parts will be easily pressed into. Been stook on that for a long time.
Open file (429.41 KB 529x963 foot wheel.png)
Open file (1.42 MB 1477x1080 Shoes.png)
>>13454 I will probably go with the Capricorn pack then that comes with replacement connectors with metal teeth. >nepgear model in slicer. Its a rough estimate material-wise since its a different character. the Wig could be an actual wig and not need to be printed at all, but then that would leave the hair ties needing to be be held on somehow. I made a rough draft of the shoes and knightmare wheel mechanism today. I also modeled a more accurate side wing but then remembered that ascento has belt chains that feed each other and are required to connect to the side of the foot where the main drive motor is. though could probably be condensed down to one big rubber belt since its only one joint and on a interchangeable copy of the foot modified to be like the Sutherland's to house the extra wheel in the feet so she doesn't get dragged.
Open file (359.31 KB 1024x504 Example mechanism.png)
>>13457 >Nepgear Was chosen as a worst case comparable model. Definitely going to need a wig and after some thought, her backup USB port should be near the nape of her neck so you won't have to fiddle with her hair to access it. After much resource, I've come to the conclusion that rigid polyurethane foam would be the idea material for her frame. It can be had with densities as low as a kg for 30,000 cm3. Her frame could realistically be made between 2 to 3 kg. She would be more durable too. Good job so far on her shoe and her pop out wheel. To clarify, I was planning on hub motors. Brushless motors integrated into her foot wheels would be the simplest and most efficient way of integrating them. Ascento's parallel mechanism in its upper leg that lets it legs fold up straight and wheels for feet were what I wanted to integrate into Madoka. Picrel is another robot with the same mechanisms but, quad.
Open file (35.51 KB 1080x1080 New Project.jpg)
>>13458 Added a bad drawing of the parallel mechanism. it uses a 2 to 1 difference so that it folds up straight.
>>13459 That is actually how the design was engineered to work in the anime. That material would also work for lining the internals of the chasis. Was about to post this in the walking thread but felt it would be better organized to post it here. This design looked interesting for keeping joints slender and simple with double joints. https://youtu.be/ENMZsPwCUcA Hollow “bones” would make more sense than driving a pipe in the middle of the articulation points. Most robot designs don’t do acrobatics. Looking at atlas again, maybe attached “veins” that protect the connecting wires would be easier unless someone happened to model and make a custom printed Double joint made for wires to be able to go straight through to the feet without being caught up in the joint itself with smaller capstans. Though I am no expert when it comes to the shoulder and hip joints and feel that the torso should be segmented but still connected as the mecha medoka design itself supports leaning. Will elaborate on that later. The main issue will always be weight distribution. You can have a great leg design, but it does not matter if the robot is top heavy and falls over even when standing still. Its always been the problem for bipedal designs. Managing weight distribution in the torso would be important with some kind of gyroscope. Though that argument may be redundant when her knightmare wheels attached for fast movement and not super tall. With a larger surface distribution on the floor.
Open file (61.95 KB 678x960 NepNepMado.jpg)
>>13461 >Capstan drives Very interesting, I do prefer mechanisms that do not rely on gears as they tend to generate less noise and heat while being more resistant to wear. A properly engineered capstan drive should never need replacing or maintenance. Not sure if it was you without your name but, an Anon mentioned capstan drives are inherently capable of having both drums remotely connected to localize mass. Though this is true, it would require careful engineering to ensure low friction on the strings. >Double joints I do not use those so outside of gunpla. I was thinking she would use normal joints. If you have a good reason for her to use double joints though, please share. >External veins We are planning on making her hollow for many reasons. Keeping her wires hidden within is a yet another advantage to her being hollow. As for pipes in her articulation points, do you mean bushings? Brass bushings can be made very thin with good friction properties. I thought we were planning on using normal bearings. Segmenting her design is indeed necessary for her to have proper articulation, glad you are already thinking about how to segment her design. >Weight distribution Ideally her mass would be mostly localized within her feet and hips. Feet to keep her balanced, the knightmare wheels on her ankle wings will help with this. As you mentioned, spreading out her points of contact with the ground is also a benefit of that design. Hips for balancing as an inverted pendulum. Many self balancing machines use this principle. The higher her mass is, the more torque is required for her to fall over. Since we're planning for her to move her upper torso, having the mass in her hips would be high enough for balancing without adding torsional forces to her waist servo. We're definitely getting a lot of her design figured out. An editable online design document would be good to organize and update her specifications over time. Any idea for a good place to host such?
Open file (352.20 KB 656x588 Chestplate redone.png)
>>13463 >>13463 >Capstan Drives I only found these capstan drives yesterday when trying to research into possible joints for Madoka.mi. Thought it would fit the elbow joints very nicely to look into something like that which could hook into the hollow skeleton idea and be form fitting if worked with. >external veins Its an idea I threw out there since I assumed by the end the skeleton will be covered by both internals, the shell/chasis, and some kind of flexible stretchy skin material covering the surface of her exposed skin areas with the chasis shielding everything. The skin material would also help with things like rain or liquid not damaging her by acting as some form of external waterproofing. There have been Japanese robots that have successfully been created like that and Im sure the board has seen similar creations before: https://www.youtube.com/watch?v=7F43R8ghTiU >Weight distribution At least we have the general idea for her to be fairly balanced. Better to have her be firmly on the ground and a good plan to solve one of the first hurdles of any robot design. I redid the general shape of the breastplate today like I said I would and also modeled some of the gems. Tried to bend the shape into the right position but I did not use a curve for this one and cut it out of a plain after fixing all the Ngons. Chiseling away at completing the rough draft before all of the technical details need to be modeled in like screwholes and plate sections. I do know AutoCAD and Solidworks, but do not have a personal license for those. Would think it would be best to leave an indent on the plate where they should be and put a USB 3.0 or C Slot in the center. The gems themselves would be part translucent plastic like a fake bead for cost effectiveness, and part solid on the bottom center with enough room for all a LED in the middle to give the nice pinkish glowing effect and all the internals of a standard USB for at least 16 GB per crystal as a glorified custom thumbdrive like we brought up before. Storage space is getting smaller and more efficient every year. I have not seriously dabbled in AI but I have skimmed the thread here on it. How much onboard storage would be needed for her OS? Her head would be definitely large enough for some memory storage and some sensory motors like Sight but It may be better to store the onboard memory SSDs where her heart would be since the upper torso would be more fortified than the head and be more accessible than taking off the hair or completely undressing her via unscrewing the connector panels in the "new" back and working inside from there. The Gems would also be a good alternative to memory wiping her over long periods time as they would be where all of her temporary and new memories go for each day save the few ones she chooses to store internally just like a human brain. >Editable Online Design Document Maybe Google Docs? so the Document can be edited in real time but only allowing actual project members to edit the document. There are permission levels for those. Let me know if you have any other ideas.
Open file (282.28 KB 2048x1152 wp-content.jpg)
>>13470 For the sake of clarity, I thought she'd just be hollow. There would be no skeleton, her servos would just move her exoskeleton. >Skin That's an interesting question. Will look into it, immediate thought would be a soft polyurea but, there's many options. >Flash drive gems Good idea to keep things functional while matching the aesthetic. >Her OS and the hardware related to it. Was thinking we could just use a phone. Android phones with 64gb of storage and good enough RAM and processirs are remarkably low cost and energy efficient compared to other options. The screen can be used for things as well. For reference, we could use Intel's Openbot as a base. That application can navigate and recognize things using just the phones built in camera and only uses 23.9 mb in the latest release. https://github.com/isl-org/OpenBot I made a new google doc for us. Published link with a link to editable version. https://docs.google.com/document/d/e/2PACX-1vQjJm3G2H2yGCJhGv6XsPUXaMMQz1Pmc4v81iNVm2iFvOTufvHc4Qje0FgV_NRu6LVij8k8QgCwvRZV/pub
>>13472 Its alright. I appreciate the enthusiasm and effort though. Am setting it to be only editable by contributors to avoid trolling. Link to Design Document: https://docs.google.com/document/d/1p3O5FutFCu1yGnNHEZ0Xfkbvdt7VyeOMsi3u6i0BatU/edit?usp=sharing
Open file (143.72 KB 850x975 MaiWaifu.jpg)
>>13474 Yeah, that was me without my name.
>>13473 Your design document looks pretty good so far. I'm kiwidev so enable access for that request.
>>13476 Alright I accepted the request and sent it out. Its somewhat disorganized but I tried to at least make a skeleton to it. There are obviously areas I did not add yet. But its a living document to be changed over time.
Open file (1.99 MB 1920x1080 1hb8d1v212.png)
>>13475 Do you have any ideas on how we could get her to talk on her own? Was thinking we could get her to do simple things like state observations such as the the weather and wave hello as a starting point. After not being happy with trying to move the breastplate into the right shape was turning out, I practiced some new techniques today by modeling a door with hinges and a handle and a doorframe as to better learn how to model joints. Also doubled as an array exercise. I now know what I was doing wrong when trying to model the sharper edges on the hard surface parts of her design and it was a really simple fix. As for Implementing the DOF, that is a matter of rigging and armature bone constraints in Blender while using actual thought in the actual modeling/design process after the model is completed. Also was looking at smoothing out 3D prints with sandpaper to make them look better and prevent accidental damage to robowaifus from sharp edges on some prints. https://www.youtube.com/watch?v=ZTE9bJyUO_8
Open file (232.97 KB 1000x1023 UltimateMadokaBirthday.jpg)
>>13497 >Speech Implementing a simple chatbot with text to speech would work well enough. She'd be limited but, that's fine for now. We can add to her as needed. Having her ping a weather server on demand wouldn't be too dificult for example. Good to see your progress in modeling. Madoka will surely benefit from it. Sanding and painting really does make a huge difference. If you use fine layer lines on a decent printer, just painting is often enough for a smooth surface too. Looking forward to smoothing out her first prototypes. Currently I'm debating on sensored vs sensorless brushless drive motors. Sensorless is easier and cheaper but, her battery life would be a little lower. I want to do sensorless with blheli s for ease of production. That's the cheapest and easiest and most accessible. Do you think she'd ever need to slow at high torque? Aside from using slightly less current, sensored motors can provide significantly more torque at lower speeds. It would be a design trade off, and I'm interested in your opinion on the matter.
Open file (257.60 KB 640x972 9xc1tw8.png)
>>13498 Happy Birthday Madokami! Today is her character's Birthday. Did not think the 10th anniversary celebration was still going on with more new official artwork on her birthday >Do you think she'd ever need to slow at high torque? Unless you are doing something dumb like pushing her down a hill or ramp, probably not. Would still make sense to have some kind of break mechanism just in case and design her for a max speed fast enough to outrun most people while on her wheels. We do live in interesting times. >sensored vs sensorless brushless drive motors? I am no mechanic, but smooth acceleration looks great with sensored motors going from slow speeds since she will likely not be out in public for a while and She also needs to be able to use them daily. Will also not likely ever need to move at the local speed limit in your home. Can always revise the design when needed. The goal is to get her working in the first place and then go from there. https://www.radiocontrolinfo.com/sensored-vs-sensorless-brushless-motor-applications/ Had also saw a neat open source wrist design that we could modify to fit the frame. Someone even managed to attach bowden tubes to it and a shoulder for a hypothetical lightweight robot arm. They found those pnuematics to not be a great solution mostly due to friction. This guy did not include a link to the original wrist design he printed a copy of. But I did find the source of it before finding this video. https://www.youtube.com/watch?v=ahSS5HUylT8 Also thought that as a future goal, a Self preservation matrix of sorts for the AI part would be helpful since its not like she is supposed to be a generic factory arm. I had also ordered the replacement parts I needed for my printer. I heard mixed things about Capricorn so I went with the TL series since most PLA/PLA+ filaments and the one I use are a 1.75mm filament.
Open file (868.22 KB 1180x1160 HappyBirthdayMadoka.png)
>>13504 Happy Birthday to a wonderful waifu. >She isn't going to need low speed torque. Figured as much but, second opinions are good to have. For the sake of simplicity, cost, and development time, the first version will used sensorless brushless motors with BLHeli_S software. (At walking speeds the difference in smoothness between sensored and not won't be noticiable with this software.) Drone motors with low KV ratings (around 1000) would only need a simple 4 or 5:1 reduction for effective low speed control. Field Oriented Control (FOC) can be implemented down the line. Making a prototype as soon as possible should be a goal. >Open source wrist That wrist design is really good for compact two axis joint. Definitely a possibility Madoka.mi will use it. Skyentific has a wealth of great information as well. If she didn't need to operate on battery power for hours, steppers would have been the way to go too. >Self preservation When we have the ability to prograam her AI with such abstract concepts as self preservation, some form of the other laws of robotics would be good. We could advertise her as being "Asimov compliant." Good luck with your new tube, hope we'll have things to print soon. On the note of speed, Producing a half scale prototype would be useful for working out kinks in her design while helping to accelerate development on the full scale model.
Open file (65.65 KB 623x960 YunoMado.jpg)
>>13504 Created a parts/cost/mass list for the current estimates related to Madoka.mi. https://docs.google.com/spreadsheets/d/1PYzP5fISoMx79wDM1jODuUP9uznD8W3u_7-4boE-WX0/edit?usp=sharing
Open file (236.79 KB 483x536 Dress_Details.png)
>>13517 I had requested access. Though you can also merge it with the master document to save time later. I made some of the details on her dress using the techniques I've learned over the weekend and will end up remaking the skirt entirely at some point since it looks very bad. Was thinking of using some of Allie's design with Madoka.mi's high heel frame but with a rounded square that fits her heel while hiding everything under her foot plating for now. Might just be overthinking it. Was also looking into slicing the frames into different connectable pieces for ease of printing. Thinking Mostly on the horizontal side since she should be able to lean forward and to the side like we discussed earlier by segmenting her torso.
Open file (30.33 KB 350x491 StudentMado.jpg)
>>13522 Access to you has been granted. Dress looks pretty accurate to the picture. You even included her crotch light which feels both pervy but useful. A crotch light helps with low light navigation. As for using Allie's design elements, recycling what you have already made is a great way to save time. I'm personally a big fan of kitbashing, both with model kits and digital files. Slicing her frame is absolutely necessary on many levels. First, horizontal prints have superior strength since layer adhesion is less of a problem. Second, Print times for flat objects are substantially faster compared to tall objects. Third, they are easier to post process since the ridges of print layers are less of an issue. Making as many things as possible out of flat interlocking panels is ideal. Also, not sure what printer you own but, I have an Ender 3 with a print volume of 220x220x250 build frame.
Open file (203.38 KB 920x1300 534687865749.jpg)
>>13523 Yeah I had worried about that. I can always remove it and change the design slightly. May be useful later on in the process, but it seemed very redundant and cosmetic in the artwork. Had also just thought her torso light would be great for determining battery level. Its not that hard to program the RGB value of an LED to change when the battery is low. Printing horizontally always produced the best looking prints as well. Ender 5's are fairly close in build size but with a 300mm z axis. I think that each of the 3 torso segments may have be made out of 6 or so interconnected segments due to the limited build volume and each segment of her arms/legs in 2-4 pieces since they don't outright bend. They will also need to be slightly thicker to have a strong connection but it can still easily accommodate what we need in a hollow design.
Open file (20.54 KB 416x350 KamiMadoka.jpg)
>>13525 I honestly would be more comfortable without her crotch light. Having her belly button light functioning as a battery indicator would be easy to implement. I'll add that to her electronic design. High Speed TPU is mostly PLA with TPU alloyed in. This material would allow her to be folded from flat panels, like papercraft. This TPU also has really good impact resistance. Thickness would need to be around a mm from my experience with it.
Open file (614.30 KB 675x946 Fixed_lights.png)
>>13526 I have removed the redundant light. Also did some research on TPU and High Speed TPU. Could not find any good videos or resources on high speed TPU beyond the average 3D printing shill not showing anything they were talking about. Like an example print to show off the elasticity for a skin-like material. It might be good for making a robowaifu huggable, but it seems too flexible and may not be strong enough to avoid accidentally damaging her internals when she collides with anything. https://www.youtube.com/watch?v=ACRh51hdBxo Though I have never printed with flexible filaments before and would feel more comfortable combining something flexible like TPU with a PLA+ shell underneath to protect her internals.
Open file (785.78 KB 1046x860 Shoe_protoype.png)
Open file (680.53 KB 675x1042 shoe_prototype_front.png)
I got around to fleshing out the shoes more with some of Allie's design for the heels that can be retracted upwards when moving on wheels and angling the wheels slightly to the sides so they don't scratch up her legs when retracting. Can be made shorter if needed. Also had to make the shoes slightly wider and thicker as to avoid 2 problems: 1. Fragility 2. Allowing room to fit her internals. All while I somehow fixed the shape to be more perfectly rounded than it was before on the outer foot plating in the process. Will of course break it down later so the outer plating comes off after removing some screws for easy maintenance Should look more into a proper ankle joint since her connector is hidden by the lower part of her legs and the two giant screws in the design and did not model it yet though it seems like a simple 1, maybe 2 DOF joint so the feet can move up/down from the drawing and actually turn rather than having some odd backwards driving design when attached to her dress.
Open file (168.85 KB 3200x1800 FoamReinforcedPrint.jpg)
>>13534 Thanks for changing the design. I will print some things in high speed TPU tomorrow to demonstrate its properties. I wanted to do so earlier but was unable to. Any suggestions are welcome. Could be useful for skin in select areas. >>13547 Good job incorporating the landspinner wheels into the design. They do look a tad long but should be fine. Widening the shoes made them look more natural. Her ankle joint only needs to fold up and down with 1 dof. That's enough to deal with rolling around and they would naturally hang down when she's floating on her dress. I've been doing more research into lowering her mass while speeding up production time. Folding TPU is not the best option. Printing with one perimeter/vase mode then filling with foam would provide the fastest production times with good strength. Printing an outer and inner shell will provide the best results. This thread is interesting to read. https://www.rcgroups.com/forums/showthread.php?2964377-Injecting-Polyurethane-Foam-into-single-walled-3d-printed-parts
Open file (1.10 MB 2172x1183 MultipleWalls.png)
A good video on the importance of walls and thickness. I personally recommend a mix of nozzle width * 1.5 and 2 to 4 shells depending on needed strength. From personal experience, infill provides very little strength compared to adding more walls. I don't print with infill unless it's needed for internal support of overhanging features like holes.
Open file (642.06 KB 1298x555 hands.png)
>>13566 He is a great resource. I use his bed leveling program every time I start up my machine. >print multiple hollow walls as a mold Had not thought to do that with cura. Typically youtubers I have found would stress slower speeds for higher quality and stronger prints which actually works in practice but is time consuming. Been looking into breaking apart the basemesh into mechanical parts like in some Japanese modelers of an older thread I found on here and SophieDev did. Was interesting how one of the Japanese modelers replaced the whites of the anime eyes with a bright light behind the iris that was able to track movement. Though I would imagine that would hurt to look at. Can't find the thread it was in on here again. Also stuck on adding the 5 detachable segments to the hands in a mechanical way that makes sense and getting that angled purple metal part to fit with the hand as a joint covering. Also learned how to make a glove effect today by experimenting. Have not tried to fix the replacement Breastplate for a few days. I did make a swappable hand with modeled generic anime fingernails underneath though I assumed a white glove that slides past the wrist guard may make more sense for a more mechanical design to look like the picture. Though probably have to make them thinner at the risk of more fragile fingers.
Open file (123.62 KB 1220x2039 CuteMado.jpg)
>>13571 Gunpla could help with getting ideas for breaking her basemesh apart into pieces. I would also recommend using hollow cylinders to connect the parts. >Replacing the whites of eyes with LEDs Fascinating idea given optical qualities of light emitting sources but, I do worry about power consumption and staring into LEDs could hurt like you said. If you find a convincing example, it may be worthwhile for further investigation. I'm personally a fan of follow me eyes. See: https://www.youtube.com/watch?v=3BwAM_V2Jhg >Stuck on the hand flare For now, focusing on her "nude" mesh would be most important. Adding flare to a minimum viable Madokami would sensible. The breastplate can plug into her, once we have her base printed, it will help in figuring out how the flare would fit. You've done a good job with her hand. I really like the gloved hand, Still having the nails being vaguely visible makes her appear more alive. Making them thinner will be fine with TPU, They would just flex.
Open file (80.20 KB 1100x1100 possible LEDs.jpeg)
>>13572 I remembered that these generic Diodes exist today and can be easily programmed for her lights. Should not be that hard to find some pink/purple Diodes. since they are cheap. Though there are probably better flat LED's that would be better though most the LED strips are all way too long and would need to be cut since we don't need several feet of LED lights. One strip is also about as cheap as the diodes and more than the length that is needed. With the LED Strips, we would just need to set the remote to change the light to pink with some form of either very rough purple plastic in the center, or something else that is not fragile and won't be removed easily on the torso and then use the rest on the wings. https://www.youtube.com/watch?v=uP1YC7XiZRs
>>13573 I found cheap pink LEDs on amazon. https://www.amazon.com/EDGELEC-100pcs-Resistors-Included-Emitting/dp/B077XDNCN2/ref=sxin_14_ac_d_rm?ac_md=0-0-cGluayBsZWQ%3D-ac_d_rm_rm_rm&cv_ct_cx=pink+led&dchild=1&keywords=pink+leds&pd_rd_i=B077XDNCN2&pd_rd_r=09fabcf6-36f1-49dd-9856-9a3aa69315da&pd_rd_w=m4JoZ&pd_rd_wg=jKTXY&pf_rd_p=c41d1f6c-956c-4fe2-8019-1663b7e1dd23&pf_rd_r=AE4WGCZMGWCRNRDEBXDF&psc=1&qid=1634073317&sr=1-1-12d4272d-8adb-4121-8624-135149aa9081 A hundred LEDs and the necessary resistors for around 7 cents each. LED strips cost more but, being addressable makes it easier to add light effects. I may have missed it but, I don't think the video touched on how it is best to provide power to both ends of the strip to ensure even brightness. I've been researching ariless tires and think I found a way to make her semi-flexible using TPU, will update later.
Open file (1.39 MB 1549x984 hand flair.png)
>>13588 The video only explained how to connect them. I've used them before and equal lighting is not a problem with these lights. The Diodes are more programmable but at the expense of not being able to program rave lights. Started on the hand flair today with parts that plug into that wrist piece like in the drawing. Will have to adjust the center to be more hollow to fit the wrist and add more utility later. The blocking out of the basemesh has been completed besides the rounded end near her ankles. The ankles are at an angle so its been concerning me that attempts to make a decent shape like in the drawing have not come out great despite my attempts to find better techniques for it. Though I recall agreeing to have that part of the design changed for technical reasons.
Open file (74.39 KB 480x655 MeganeDoka.jpg)
>>13593 Good job thus far with the hand flair. If the basemesh is completed, it would be helpful for you to upload it somewhere I could access it. Then I could experiment with electronics integration and figuring out how to make her as light as possible.
>>13594 Sorry for the Confusion. Basemesh is not rigged and its just a static model for the moment. She is also not broken up into the various printable parts yet. Also found another open source robot very similar to Sophie. https://medium.com/pollen-robotics/we-are-releasing-reachys-models-for-creative-people-befe45b437fb
Open file (4.85 KB 188x188 Poppy.jpg)
>>13595 Ah, very well, good luck with modelling further AllieDev. That reminds me of Poppy https://www.poppy-project.org/en/
Open file (284.68 KB 4000x3000 Test1.jpg)
Open file (300.54 KB 4000x3000 Test2.jpg)
>>13595 After testing various designs, foam filled PLA has the lowest density for a part of usable strength. I overfilled it to, in practice we should be able to get even better results. Both parts are 6 cm cylinders with a height of 1 cm.
>>13611 That is an excellent idea managing the overall weight of the body. Would help so see a cross section of the part under the top with what brand of polyurethane foam you are using. and for me to get around to cleaning out my printer again. PLA was coming out from behind the hot end after I'd cleaned it out and was in a state of "fuck that printer." for a week. Cleaned out most of the junk and it peeled right off the hot end.
>>13615 The polyurethane filled one just has a hole in the center for the filament to bridge to. I'm just using the cheapest Great Stuff spray foam. There are far better polyurethane foams for our purpose, I just wanted to use something cheap fast and easy for the sake of testing. As for your printer, when I had a similar problem it was because the nozzle was broken. It looked completely fine but, replacing it solved all the problems.
>>13616 >the nozzle was broken Good news, I cleaned it out thoroughly and tightened the coupling and hot end to be tighter in this time and it works again. Had to use the provided needle and melt most of it through. What was on the outside peeled off surprisingly easy up through the threading after being cooled down. The test print I had came out great. Though I've only ever successfully printed in raft because the prints just don't stick to the bed as well even after I've cleaned the bed with isopropyl alcohol. >I'm just using the cheapest Great Stuff spray foam That works. Have not looked that much into foams.
Open file (191.15 KB 1040x1093 FutureDream.jpg)
>>13616 Glad to hear your printer is fixed now. I personally smother my print bed in glue stick. When the bed is hot, the glue melts and bonds to the plastic. It comes off with a little wiggle of the bed. (I use removable beds) Any broad flat scraper will pop it off too if you don't use removable beds. Hope we both can start printing parts for her soon. I already did the research, low density marine grade closed cell polyurethane foam is ideal for our purpose. It only needs the thinnest of shells to be surprisingly durable. We print out Mado's parts, fill them with slight amounts of foam, let them expand and cure for a day, and she's both light and durable. Also, could you please provide the total volume of current basemesh? With a volume I could get a better idea of her actual mass rather then relying on the estimates from a Neptunia model we are currently using.
>>13618 Basemesh volume is 20790.806 cubic centimeters when scaled to the correct height. Really weird how almost perfectly exact that number is given some rounding.
Open file (172.72 KB 1555x2048 5kgCrab.jpg)
>>13620 She's incredibly lithe then. A human woman of healthy weight generally has a volume over 60,000 cubic centimetres. Which works out great, I found marine foam that's 9.8 kg per cubic metre. This would put her at less than 600 grams for her foam assuming she was completely filled. This could lead to her being similar to if not lighter than Senjougahara afflicted with crab.
>>13617 Check out this video on G10, it may be the material you need for a working bed. https://youtu.be/g0PK4oXbJT8
Open file (81.17 KB 324x288 -crab-404785.png)
>>13622 >Lithe? That character is physically bigger than she is. So of course the volume would be lower and she would be lighter. A layer of inner foam for impact resistance would work for the limb segments (including the hands and feet), and torso. Though would have to keep in mind printing the inner wall so the foam could have a mold in the first place. Would also depend on how thick the foam layer is since that would also take away from internal storage space. For the head, would be my only concern due to how complex a shape it is. Most of the time a standard infill works just fine from what I've seen in Sophie and the Japanese dolls in the abandoned projects thread on here, so It would seem like a lot of extra work for no reason since she has to have blinking eyes and a talking mouth and have a removable skull section for easy matinence. Since she has relatively short hair we should be fine, but the wig should be trimmed down to around her waist or mid torso so it never gets caught up in the wheels or joints. The shoulder pads would prevent it from getting caught in the arms. For the clips I will modify what I have made to either be tied together somehow or have a thin connector section with a hole in the middle for the hair to go through with small magnets in the sides to keep firmly attached to the head. They should be very light since they'd be hollow like ribbons anyway. Maybe could be reworked into external flashlights beyond being cosmetic. Also have to take into account the maximum tolerance for her actuators being the weight of lifting an object + combined the weight of her appendages she is lifting. The stress of a shoulder is more than a wrist joint and so on. The next issue would be installing actuators that have a sufficient tolerance of 40-50 lbs for the arms and 120 lbs for the legs while remaining stiff enough to stand up on her own without a stick. The lightweight hollow arms would be sufficient to not contribute as much to the strain on the actuators. >>13624 I will try to use adhesive again since I assume you also have a stock Ender bed. I got a sample of Magigoo with my PTFE kit which looks like overpriced glue sticks from what I've researched. Costs more than a roll of PLA+ for a single stick. Raft is a very wasteful way to print in both time and material and It's something I'd like to not to do much of going forward.
Open file (89.38 KB 600x840 NekoMado.jpg)
>>13628 Naturally she won't have foam reinforcing her entire form. Her low volume just means that foam will account for barely any mass. It's surprising how light she actually could be, especially compared to anything of a similar height. The head should not see any physical strain. 2 or 3 wall layers and 20 or 30 percent infill should be just fine. A face plate and a scalp plate to access her internals is definitely a good idea as well. Though the picture we're working from depicts hair long enough to reach her heels, only going to mid torso is far superior for many reasons you probably already realize given your suggestion. Her hair clips are essentially mecha nekomimi so, a cat ear hairband with optional lights would be appropriate. >120 pounds for her legs We're making her ultralight, why would she need such high torque? I use a steel print bed because my first one wore out. Those specialty adhesives rarely ever work better then normal glue stick but, getting a free sample is nice. Good luck, hope those rafts are no longer needed with your Magigo.
Open file (329.82 KB 1182x736 ouch.png)
Open file (72.42 KB 850x698 47315.jpg)
>>13632 >why would she need such high torque? So she can stand up on her own and not collapse on her face. I am not an engineer. However, I get she's not a powerloader. Was just throwing out a higher number than the arms since the legs have to support the entire weight of everything above them and be stiff enough to do it. Also looking around for ideas on breaking the basemesh into parts that make sense. Plan was to cover it all in a skin material at the end anyway and also wanted to have fingers that work. Would make more sense to print sections of the limbs since they do not have a radius bigger than the print area of an Ender 3/5 and I can break the arms down into something like this or reachy's. SophieDev even had a similar idea with some success. https://twitter.com/61laboratory/status/1446835094518317057 I do have concerns about the torso segments fitting in the build space horizontally. The average width of the torso is 4-6 inches and the length is around 8 inches at most around the hips and seems too small but then I measure Reachy and he is smaller than her. Her arm is around the same size as this but thinner. https://www.youtube.com/watch?v=L4fB0rVJI8g Though working Fingers are another Problem even with Hollow hands along with cooling. Just showing some of the topology of reachy's joints. Upon further inspection of his Eyes I could probably make a miniaturized version that fits inside an eyeball like the picture and have an LED's shining through yellow tinted plastic like with the lights on her torso.
Open file (60.66 KB 500x500 Charli.jpg)
Open file (2.64 MB 3203x2019 TendonHand.png)
>>13637 Ah, similar robots of similar heights used similarly strong actuators. Like CHARLI (picrel) http://www.romela.org/charli-cognitive-humanoid-autonomous-robot-with-learning-intelligence/ I would recommend breaking her mesh down into halves and quarters. We can figure out how her servos integrate into her mesh when those are finalized. Hands are easy, see other picrel. 8 inch wide hips are incredibly small. I never would have imagined Madoka only had 20 cm wide hips but, she is tiny. This is good to know, her internal mechanisms will need to be streamlined. Reachy is another interesting robot for sure.
Open file (61.81 KB 1024x684 LinkageArm.jpg)
>>13639 This was from Kiwi, my PC updated and my name went away. Also, I was planning on localizing her servos into her torso in a similar way as this linkage driven arm.
Open file (123.25 KB 494x782 basemesh.png)
Open file (110.32 KB 845x448 arm test.png)
Open file (108.16 KB 891x315 Arm test 2.png)
>>13639 That makes the hand a lot simpler to break down. Should not be a problem with hollowed out arms. >I would recommend breaking her mesh down into halves and quarters I broke them down to where the natural joints show up, which is basically what you suggested. Did not attempt to break down the hands yet. The joints would obviously be replaced by the mechanical parts. Also broke the torso down into a simple three parts so she can lean. may end up just attaching the ankle screws as some form of cosmetic or cut them entirely. Used some of the simple design Reachy had for an elbow as a reference so the arm or joint is not damaged when pulled closer with a simple bevel. figured the wrist joint would be hidden by the bracers anyway. >>13641 That was also what I had in mind. Have been double checking her measurements and apparently she is 1.52 m according to blender. I'd scaled her to a block of the same size and then imported her PSP game asset to make sure she is correct. Only her limbs are super thin compared to it and I began to adjust the proportions to fit. Her torso was surprisingly accurate and more detailed than the game asset. The Game asset's legs were also a lot straighter and thicker than the workflow I had been following which encouraged a slight bend from the knee down to the ankles. Will have to remake them for better stability and durability.
Open file (282.33 KB 1349x1920 BallJoints.png)
>>13645 Big fan of the move to a more ball joint-esque aesthetic. Overall, good job for splitting her body up in ways that are sensible and congruent with desired degrees of freedom. Her hips should be rounder where they connect to her pelvis though I'm sure this is an iterative design and you are aware. Good idea to check against an official game asset. That is as close to her real world physical measurements as we can attain. Her legs are probably straighter because the Japanese like "chopstick legs" i.e. long straight and slender legs. Making them thicker will help with stability and should not add much to her mass.
>>13647 Was messing around with some ball joint hinge ideas. This was just something I know would not fit, but the concept was simple enough. It would conflict with the hollow part design we already have. Don't recall ever going over what specific motors we were going to use.
Open file (19.73 KB 392x423 Revoltech.png)
>>13661 That looks kind of like a revoltech joint. We definitely can use those with push and pull rods for it.
Open file (1.08 MB 3000x1774 LegDesign.jpg)
Open file (1.00 MB 2280x2584 WaistDesign.jpg)
Open file (630.28 KB 1609x1870 HeadDesign.jpg)
Open file (47.51 KB 480x640 DrGuero4Servo.jpg)
>>13661 Drew some designs in a notebook for her basic mechanisms. She'd walk by shifting her mass using her waist mechanism then moving her legs. Similar to Dr.Guero's 4 servo biped. http://ai2001.ifdef.jp/mr1/mr1.html
A Design like that would work modified to fit inside the chasis. Though her torso would be lighter than her bottom half.
Open file (293.10 KB 1191x870 example.png)
For people looking to use vroid models, know that it's difficult to prepare them and they have complex geometry inherently. This slightly undersized head shows what I mean. Even without hair, print times don't go down much and cleaning the model for printing is still a pain.
>>13721 Too bad, but thanks for sharing.
Open file (379.83 KB 1151x715 Example.png)
>>13721 I've had no issues like that when I sliced prints from models I made from scratch. Was significantly less frustrating too. Was surprisingly manifold and small compared to the measurements blender's measure tool said it was. Had to scale the tiny model up 7500% from the size of a marble. Still feels too small l tried a slower printing speed of 45 mm/s which works for my printer and goes down to 2-3 days when around 80mm/s or higher with similar results. Have to fix stringing issues with my printer and the scaling of the model before I attempt to print anything. Found vroids to be more of a pain in the ass than they were worth in my experience and had none of the utilities that people claim they have. Like being able to toggle hair on/off and not have to manually separate that part of the mesh from the whole model. That program was designed for Vtubers and VR game engines. Not 3D printing.
Open file (109.86 KB 1492x1275 LargeServo.jpg)
Open file (97.72 KB 1500x1073 SmallServo.jpg)
Open file (44.32 KB 679x632 DriveMotor.jpg)
>>13661 For big servos: Standard sized brushless/coreless motors capable of 35 kgcm For the smaller servos: Metal geared micro servos would be fine. For her drive motors: 4108 brushless motors provide high torque and would require minimal gearing.
Open file (435.88 KB 1000x1200 MadokaModel.png)
Good to see your design is printable. Though it would probably be faster to print as two halves that did not need support. That's a shame vroid is such a pain, really looked interesting. Modeling from scratch does seem to be the most effective way of getting real progress.
Open file (16.40 KB 474x355 madoka doll.png)
>>13751 was looking into dolls for ways on how to make a better eyelid and ball joint mechanisms while being able to stand up as a psuedo carbot with what we already have for now. They don't use skeletons, are mostly hollow like the basemesh, and most of the time do not stand up on their own at all without a stand. Sophiedev had the right idea for working eyelids but there are simpler ways to go about what he did with mechanical methods. we also can fit a normal shaped eyeball behind the opening depending on the design and it will still look about right from what I've researched on anime dolls with different shaped eyes. Madoka has a non-standard eye but printing something that looks right would be a non-issue. Would just need some yellow translucent filament for a tiny part and a generic white PLA+ eye that can fit a small camera and LED. >>13752 Was just seeing how well it would look first. I would need to make the parts thicker and it would take much longer to print 2 halves along with making sections to connect to each other to where it looks flush before putting some kind of sealant over the thin gap in the parts.
Open file (73.61 KB 664x332 FlappingMechanism.png)
Open file (58.07 KB 640x480 SimplestBlink.jpg)
>>13774 That Madoka doll is one of the cutest things ever made. I truly hope our project approaches this level of cuteness. It's worth thinking about her design like a dolls. They are very low density and if you increase the friction in their joints (drops of ca glue and powder can do the trick) can stand just fine. In our case, servos have plenty of natural detent torque to prevent unwanted movement. For eyelids, a simple flapping mechanism can be made by placing a magnet in a coil. This mechanism has low torque but, that shouldn't be a problem. Its greatest strength is that it's dirt cheap to make and super fast. This is important as slowly blinking robots give me the heeby jeebies. There is the other option of just painting an eyelid and flipping the whole eye. https://www.youtube.com/watch?v=gDrjgVRJONA
>>13774 Sorry, but it's not possible to understand that ext. What are normal shaped eyeballs? >>13775 What is 'ca glue'? >just painting an eyelid and flipping the whole eye. Ah, interesting.
>>13775 Dang, I was hoping that tutorial would have one more axis.
Open file (104.46 KB 1136x640 TsurineTareme.jpg)
Open file (194.42 KB 628x472 EyeMechanism.jpg)
>>13782 Normal eyes refers to human shaped eyes, as opposed to tsurine and tareme eyes which are common in anime. Madoka has exaggerated tareme eyes which denote her exaggerated soft and cute personality. Tsurine eyes would be representative of a more fiery and confrontational personality. CA stands for cyanoacrylate glue, which are also commonly called superglues. >>13789 If you mean a left and right axis, that could easily be remedied by adding another servo and control rod.
>>13795 Well, yeah. But the pic you posted shows exactly the problem I'm expecting. For a full 3-axis control needed for 2 eyes, it usually goes really far back into the skull.
>>13795 It looks like we have new people who are interested in the project.
Open file (121.80 KB 567x600 KawaiiMado.jpg)
>>13808 Speaking of real interest, I believe making a scaled down version that is legally distinct from Madoka would make a good product to sell to help fund the full scale version and production of necessary tooling to create better Madoka's. 3D printing will take too long to make multiple full scale Madoka's at any reasonable timescale. Thus, having the resources for actual tooling to create molds is necessary. Smaller Mado's could be made with 3D printing at reasonable timescales and sold for reasonable costs to generate capital. I was thinking of a 3rd scale Mado to sell, any bigger and shipping becomes problematic for small scale business.
>>13808 I think you posted a response to my suggestion for selling mini legally distinct Madoka's but, it seems to have disappeared.
>>13808 Hello? Is this project still on?
>Crosspost: Flying humanoid robot which can also walk >>13863
>>13840 Yes its still on. I do not know where my post went.
>>13886 Did you post something more and it disappeared?
>>13888 I think I saw him post something about agreeing to make mini madokas but with a lot more text. He should repost something
>>13721 Are their similar problems with programs like Make Humans?
>>13891 Been practicing by making more complex characters and attempted to print a helmet I made piece by piece to accomadate the build space as a test since the head does cover most of the build plate of my printer as one object and felt too small. I hate this printer so fucking much. On better news, I also modeled the empty shell to be functional and be able to mechanically lean in 5 directions since humans dont bend their spine backwards like the shaft head tilt. Also Found out how to properly rig a mechanical robot model. I do not remember my post so am mostly paraphrasing. Im sure banpresto and Aniplex would not be happy with Custom Garage kits of something resembling Madoka floating around and I hate sculpting with a mouse. All of the models I’ve shown are not sculpted at all. Mini Legally distinct madoka.mi’s would make sense.
Open file (30.98 KB 700x525 FullTiltHomu.jpg)
Open file (50.85 KB 659x659 Kuroe.jpg)
>>13934 Good to hear you've been practicing. I have faith your models will yield quite the qt. My Ender 3 works fine and I can do test prints for our Madoka. >No Shaft head tilt My disappointment is immeasurable and my day is ruined. Looking forward to seeing the updated model that is capable of greater articulation IRL. Once we start printing her, we can make some great improvements. I've pretty much mapped out her internal mechanical structure, once I get my hands on her model I can start modifying her to integrate it. It's essentially what I already doodled and posted. It is true that they wouldn't want garage kits on one of their more profitable characters. Though it is really nice that she is still selling so well. I do wonder how distinct she needs to be though. There's a few characters that look like her. Maybe if we replaced her pigtails with buns like Kuroe and kept her black and white, she'd be distinct enough.
>>13935 Trying to make legally distinct characters like Kurou or Suzune would also probably go over well enough considering its okay to make similar characters in that genre. Something along the lines of “Build a meguca” would be funny with some imagination and I did learn new techniques for a faster workflow while trying to learn sculpting with a mouse. Modularity would still be a concern since obviously not all orders would be the same.
Open file (440.54 KB 1200x981 MechaMami.jpg)
>>13953 I have a good amount of capital saved up from being a wagie living in the basement. I'd be happy to follow in the steps of General Products / Gainax and invest everything into making waifu garage kits. I think "Mecha Mahou Shoujo" would be a great name for our first product line of demure, diminutive, mecha maidens. For scale, I will say that a meter is the tallest we could build at an affordable price of manufacture. Of course, we will need to sell them for a profit to help investment into Madoka.mi and her tooling. From research, 50 cm would be the smallest we could make a robot that had reasonable capabilities related to movement. 80 cm is the smallest that would be clangable as far as I can tell from research.
>>13934 >made piece by piece to accomadate the build space as That's a good thing, imo.. That way little changes in the design won't require a complete reprint. Also, if you had a big printer but others not then your design wouldn't be useful to them. Learning how to design, print and build assemblies will pay off down the road.
Open file (170.93 KB 667x1000 Allie Figure Official.jpg)
>>13965 1/8th scale is around 8-10 inches tall typically and are printed out of ABS PVC. (Whatever the hell "ABS PVC" is as a filament, I have no idea. Cannot find any info on it but PLA and ABS are interchangeable as we already know and PLA is non-toxic.) Upon further digging it is in fact, a resin called "PVC/ABS" which explains how smooth the figures are with seemingly immaculate paintjobs. Painting them is a whole different process and the eyes are just small stickers on most anime merch. Though never touched an airbrush before so I can't make the insanely subtle shading all over every print yet like our Japanese Brothers swear by with dual action airbrush painting in all things involving figures like gunpla. along with sculpting in exclusively Z-brush while saying a whole lot of nothing for over an hour in their conferences. I do not think I am that great but I want to be one day once I am more familiar with sculpting tools and the masochistic experience of sculpting with a mouse.
Open file (27.69 KB 450x426 Femisapien.jpg)
Open file (431.62 KB 607x852 She'sTooSmall.png)
>>13979 10 inches is really small. Even Femisapien is 15 inches tall. We don't need to paint them, just use the colors of the plastics. Her owner can then customize her to his liking. 80 cm which is about 2 and a half freedoms tall would be better. I want to be able to hold her at night. I do hold a Madoka plush at night sometimes and she's too small.
>>13983 Was referring to that size and painting for the garage kit figures. Not the prototypes. I always had cuddling in mind when designing the basemesh. Hence detachable KMF wheels to be reinstalled when moving again. Thats an adorable plushie. Reminds me of the neat figma madokami I have. The 4th movie might be interesting.
Open file (238.74 KB 580x800 FigmaMadokami.jpg)
>>13985 Ah, wait, so non-robotic garage kit figures? I thought we would make minimally robotic garage kits that were just big enough to snuggle. All the figma Mado's are cute as heck. I'm looking forward to the movie too. The second season of MagiReco showed they can still make Madoka compelling as a character.
>>13988 Was planning on both since it helps to have more to offer. Always had specifically cuddling in mind so the material would need to be thicker and stringer than 2 hollow walls and some foam so they dont easily get crushed. We did discuss it before. I believe thicker walls with foam would work fine in my experience printing lightsaber hilts I modeled and Painted. The thicker parts were undamaged while the thinner ones chipped off easily as soon as it was accidentally dropped from a table. Fortunately was going for the “ancient tool of the jedi” look with the mediocre paintjob I did and mixing black/white with metallic silver and red to give a worn look. I had a nice long thesis as to why I dont like specifically the magireco anime but the site deleted it as soon as I posted it. Urobuchi had nothing to do with writing Magireco at all by choice and with all the stakes removed like grief corruption, It had the emotional impact of a wet noodle for me with its rushed pace and asspulls like getting stuck in doppel form. Book 2 was far more interesting than Book 1, which the anime adapted and it was a glorified 2 season advertisement for the game with a breakneck pace.
Open file (29.80 KB 391x750 UsagiMadoka.jpg)
>>13997 >Making both That makes a lot of sense honestly. Having multiple products at different price points is good for sales. The normal figure would need to be simple and easily mass produced. Something like the bunny Madoka figure. This would allow us to rapidly implement her tooling so that she could be produced at scales that would permit her having a low price. >More then 2 walls That is fair. 4 would be the limit though. 3 would likely be fine for anyone. The foam filling provides surprisingly high rigidity and strength. Nice lightsaber bro I didn't like either season of MagiReco either. Felt like a soulless nonsensical cash grab. I did enjoy the scenes with Madoka in the second season. Which is likely because of my infatuation with her rather then an endorsement of quality. I also thought the fight scenes that weren't a clustertruck of lasers and bullshit at an asspul carnival to be enjoyable as well. Kuroe was also cute but, I have a thing for sad loners.
>>14001 I should probably start printing some of the shell parts and doing stress tests. Though I haven't modeled any of her internals yet since the list is not finalized. I have no idea how to proceed with actually filling the walls of the limbs with foam besides a cap of some kind and altering the design slightly. She will need some kind of inner frame mechanism so she can actually stand up and move her extremities while being physically in one piece.
>>14007 Strong joints will be fine for keeping her stable upright. Think of her as having an exoskeleton. As I've stated, I'm working on her electronics so, I will work out her internal structure when I get a chance to print her.
Open file (195.15 KB 772x639 segmented 3.png)
Open file (122.76 KB 708x836 segmented 2.png)
Open file (112.92 KB 616x804 segmented 1.png)
>>14009 >I will work out her internal structure when I get a chance to print her. I need some internal specs so I can model in all the screw holes in the joints....
>>14044 Her leg, upper arm, and waist servo: https://www.uctronics.com/uctronics-30kg-high-torque-digital-rc-servo-with-bracket-kit-full-metal-gear-steering-servo.html Her servo for her hands, forearm, and head: Generic 9g servoshttps://www.dfrobot.com/product-1579.html See my previous sketches on how they are meant to be mounted.
>>13673 You should add that to the document so we don't forget it later on. She will have a higher center of mass for the final iteration but a small prototype would work fine with this concept.
Open file (258.56 KB 1500x1367 MicroWaifuMind.jpg)
>>14103 I've added some updates. Unfortunately, I'm in college and recently had a family event which took my mind off of this. I'm back to focus on this though. I've been thinking that our first product should be similar to a gunpla model kit. Having runners will help with keeping her parts safe in shipping. A silicon mold with holes for ejection sprues and a hard exterior for dimensional stability can be easily made. We would have to use resin just like old school garage kits but, that allows for dirt cheap tooling at the cost of each kit taking a long time to finish curing. We would still need a gimmick for her to stand out. I was thinking we could integrate an ESP32 CAM powered by 4 AA batteries to cut costs would allow her to have many functions with a companion app. I'm thinking a 20 dollar micro waifu would be a good first product.
Open file (48.87 KB 700x726 limb servos.jpg)
Open file (39.63 KB 909x346 head servos.jpg)
>>14106 I found the schematics conveniently on the product page and were helpful in lining up the holes correctly. >We would have to use resin Resin printers are on average, far more expensive than the ones we have and the resin bottles cost 200% more than a single roll of PLA+ on average for a 1000g bottle. Though there are smaller cheaper ones and had been thinking of getting this one since getting something with around the print space of the ender 5 with 200x200x300mm is very expensive and this could print smaller garage kits like cheaper manufacturers do: https://www.youtube.com/watch?v=g9ZyygByLVQ No idea what "PVC & ABS" means for most figures these days since Allie sure doesn't have any PVC pipe parts.
>>14149 just stick an ipad onto a roomba, instant waifu and it cleans the house too
Open file (359.41 KB 4093x2894 g-1.jpg)
>>14150 ...that's genius
Open file (247.60 KB 1588x1226 MoldedGirl.jpg)
>>14149 I meant making her using epoxy and polyurethane resins so she'll be easily produced for very low costs in all steps of manufacture. Similar to the garage kits in Japan
>>14149 PVC and ABS likely means that the figure uses different resins for different parts. It could be a PVC and ABS alloy as that could have good mechanical properties though they would be difficult to alloy.
Open file (115.86 KB 1200x628 madoka-magica-1265791.jpeg)
>>14156 I will be troubleshooting and printing test arms this weekend after learning some bolt threading techniques and getting it to a point where I am satisfied with it. Had considered threading the holes themselves but after looking at reachy and other designs a few times, figured that a few tight bolts is all that is needed to hold the limbs of the shell together. It was also not that difficult to just Boolean a mold for the parts. Blender just wanted to crash a ton after the latest update to where I just took a break out of out of frustration.
Open file (149.71 KB 500x281 HeatSetThreadedInsert.png)
>>14178 If you find you need stronger threads, using a soldering iron to melt in a threaded insert can help. Good to hear molds can be made easily. Once kits are being sold, we will build brand recognition and actually start the robowaifu industry. I still think "Mecha Mahou Shoujo" or (MMS) is a good product name for us. We could be "Robowaifu Productions" though other suggestions are welcome.
>>14178 >frustration For me it works best to do something else on the side, like listening to music or to a podcast that doesn't need to much attention. Also, some calming tea from time to time helps.
Open file (1.52 MB 2576x1932 20210318_215552.jpg)
Open file (1.31 MB 2576x1932 20210318_220619.jpg)
Open file (292.53 KB 385x800 RiCO_full_body.png)
>>14106 I was working on a project very similar to this. However, instead of molded resin I was just using SLA 3d prints because I have no plans on sharing/selling the thing and it's easier/cheaper. I've got other stuff I want to finish first, so she's on haitus for now. I made control code that lets you control, potentially, up to 32 servos over HTTP, and view the camera as well (however doing so locks up the servo side; moving the camera server to it's own FreeRTOS task may fix this). I don't have any server-side code however. Code: https://pastebin.com/A7NYWggP
>>14188 Very impressive Anon. We look forward here to your future creations!
Open file (1.03 MB 1000x1725 Cinnanya.png)
>>14188 >RiCO Based choice in reploid waifu. Thanks for sharing your code, I'm definitely going to reference it with our ESP-32 based projects. How'd you get that model for her? Looks amazing and I look forward to seeing her completed. If you need any help, don't be afraid to ask. (Cinnamon is my reploid waifu)
Open file (223.27 KB 909x699 updated slicer.png)
>>14209 I found out the problem with my failed prints was cura 4.9.1 all along. Anything I printed that was sliced in 4.8 worked perfectly. Anything in 4.9 would quickly create weird burns in the print and generally not look very good if it didn't fail badly within the first 20 minutes. So far since updating to 4.12, the scaling was actually correct all along and it was just some programming fuckery that screwed up the scaling in 4.9 when I moved the STL files from blender. I like Mecha Mahou Shoujo as a name and should probably merge the dress with the basemesh. I started with the limbs first because they have the fewest parts. Was also odd that Cura has Arduino support now ever since october apparently as a printing option in the 4.12 installer and found some further reading: https://support.ultimaker.com/hc/en-us/articles/4407742254610-Ultimaker-Cura-interacts-with-microcontrollers
Open file (252.68 KB 453x704 Ganbare.png)
>>14291 It really is best to keep things as up to date as possible. Good luck with printing. Mecha Mahou Shoujo it is then, hopefully we can make the first prototype before end of year.
>>14291 >cura 4.9.1 ... programming fuckery What bothers me, is that I've read about that a while ago. I wanted to warn anyone here on the board about this problem, but I think I forgot about it. Darn, what a waste of time.
Open file (382.61 KB 543x641 helmet.png)
>>14292 I learned some new techniques. Ornaments will be useful later since some anime designs use them. The upper arm prints also came out decent and figured from now on it would be easier for the printer to just cut limb segments into two parts and screw them on together to improve the quality and look of the prints overall with only the connecting point being noticeable.
Open file (2.86 MB 4608x3456 GunplaExample.jpg)
>>14310 You could always use cylinders to avoid having holes for screws. Gunpla commonly use nesting cylinders to connect pieces together. These cylinders would also help to act as structural supports for better strength and printability.
>>14311 That is a good idea, but it would take longer to print and waste more support material. since those gunpla parts are made with molds, not 3D printers explicitly. Its why they have trace amounts of oils on them when they are brand new to get the parts out of the molds and more experienced gunpla builders wash their parts of they are going to paint them.
Open file (175.93 KB 748x664 ProofOfConcept.jpg)
Open file (247.91 KB 1049x863 ProofOfConcept2.jpg)
Open file (170.11 KB 726x672 RaftFilter.jpg)
>>14315 Came up with a better idea, using rings to clasp together parts. See picrel for an example. Took me multiple tries to actually print it because of printer problems. I can only get prints to work with rafts now. I now use rafts with no top layers as it is a waste of plastic and makes getting the parts off a pain. I set the raft extra margin to 1 mm. The default settings are a huge waste of time and plastic so I recommend changing them. The rafts could be used as filters for other things too.
Open file (304.34 KB 1194x770 Reachy re-examined.png)
>>14330 That is actually what I was thinking of after re-examining Reachy and realizing I forgot to add a way for the wrist to turn on the forearms. They used some Bearings with rings and simply screwed in the servos. >I can only get prints to work with rafts now. Lol welcome to my hell. >I now use rafts with no top layers as it is a waste of plastic and makes getting the parts off a pain. I set the raft extra margin to 1 mm. Had not thought of that. Thanks for saving me a lot of material and potential heartache because whatever gets printed on the face touching this support type is marred by ugly lines that aren't even hidden by paint and barely hidden by modeling putty. Was planning on printing prototype arm parts tonight but was waiting until I finished modeling parts like this and during a random study of Reachy, I noticed that they didn't model in threading on any of the screw holes and all the threading was inside the servos. Will have to model the servos quickly to accurately seat them inside the models and maybe modify the elbow so it looks more natural like this design. Had an idea to simulate the anime eyes where the iris was the only part that moved with a small camera attached to it and the eyelids close with small eyelids similar to Kibo. similar to what I mentioned a few months ago but with less High beam eye whites and some transparent plastic covering the opening shaped to look like her eye in the anime.
Open file (6.16 KB 225x225 DIYKibo.jpeg)
Open file (8.49 KB 225x225 DIYKibo2.jpeg)
Open file (6.85 MB 480x360 PaperDoll.gif)
>>14331 >welcome to my hell Kek, I'm looking into getting a new printer because this Ender 3 has problems all the time. Not modeling threads makes sense. PLA is soft enough that almost any screw works as a self tapping screw and, if you have anything beneath the plastic, its threads will be better. >Kibo Chan She works like Japanese paper dolls/puppets, using a plastic sheet to lower her eyebrows. Similar to Cermnrl's work, though using plastic gives Kibo Chan minor follow me eyes. Cernmrl's works are like living animations, they have no follow me eyes effects but, they could allow for a direct representation of Madoka. https://www.cermrnl.com/gallery-1
This video really shows off Chibo Chans follow me eyes effect. >>14331https://www.youtube.com/watch?v=dJYUOxxDW4k
>>14333 >digits That doll puppetry is adorable tbh. Great character appeal.
>>14330 >>14331 Try to find a way to fix your printers. We could discuss it in the 3D printer thread >>94 or you could find the solution somewhere else. The thread on printers on 4chan in /diy/ board /3DPG/ is actually quite good.
>>14209 She's a wonderful waifu Anon. I hope you can make her physically real soon.
Open file (1.27 MB 1920x1080 Madoka.mi Joints.png)
Open file (15.91 KB 415x379 Wtf is this.jpg)
>>14333 During my overhaul today, I decided to model the missing parts from the schematics just like they did with reachy to better align the screws before I start with the prototypes. No idea how they attach the other side but that's "DIY" for you. Did not model in the screw holes for the main part yet. Also somehow missed this small detail on the limb joints and will get to it tommorrow. Is that supposed to be how you can tell the Left from right joint?
Open file (797.41 KB 1000x1100 FallMadoka.jpg)
Open file (130.34 KB 649x892 ServoHeadBushing.jpg)
>>14348 The bracket will keep the smooth bore servo head on the nub on the other side, they form a plastic bushing to help balance forces on the bracket. See picrel of a cheap servo that works like that. >WTF is this Chinese engineering does not always make sense. One must be very careful when working with parts from China. Autumnal Madoka presents acorns for encouragement.
Open file (179.26 KB 828x708 bd7.png)
>>14333 For RiCO I created an eye mech similar to the Kibo-chan one but with the eyes on an angle, so you don't have that flat-face look. Red piece has printed eyes attached. It is pushed by the left-right sliding blue rack and pinion. Green frame has an angled slide rail and the pins on the rack + socket on the red piece move in and out like a cylinder. Up/down motion is provided by moving the entire mech vertically (green rack and pinion). You were asking where I got my model, I got the face model from somewhere, don't have the link on me. The rest of the body was all drawn in fusion360. It's not 100% accurate to the original but trying to convert a vidya mesh into a complex assembly like the third image would have been just as much trouble (and about 32% as fun) as re-drawing it myself.
Open file (187.81 KB 472x850 KawaiiFlat.jpg)
>>14359 >Angled eyes Interesting idea. It will be good to see how that effects how cute she is. I was thinking of keeping the face rather flat as anime characters have flat faces. Picrel illustrates how flat Madoka is. It is best to try different things, diversity of thought is valuable. I do worry that the back and forth motion may look strange when the eyes are at opposing angles. Have you considered making the eyes arcs? >converting vidya is a pita Heh, been there. I tried with Neptunia rips and it was crazy trying to makes those models workable. Building our own meshes and assemblies is looking to be the idea until we figure out generic parts everyone can use.
Open file (273.37 KB 1222x804 screws.png)
Open file (51.31 KB 1001x907 official servo.jpg)
>>14357 I successfully modeled the joint parts and learned how to make proper screws in blender today. I will never understand the autism known as Chinese Engineering. Now to install them in the model.
Open file (226.55 KB 801x964 knees.png)
Open file (752.07 KB 1566x607 arm joints.png)
>>14366 I have begun experimenting with internal structures and added some joints to madoka.mi today. Blender was having a stroke whenever I tried to boolean and will end up just taking a block and booleaning that to fit the mesh later. Since it gives you 2-4 holes a side to work with, it's not as complicated as I believed it would be. The real question is how the hell am I going to rework the waist to accommodate this these joints. Also made my version of the wrist turning mechanism idea that I borrowed from Reachy. Did not model the bearings yet and will innovate on it as I go.
>>14381 Looks good, you're making some visible progress.
>>14381 Hey there AllieDev, I had a request for you: >>14378
Open file (12.97 KB 300x550 Tecfoot.jpg)
>>14381 Looking decent though, there seems to be some misunderstanding regarding her walking mechanism. I was thinking of going for a Tecfoot style configuration for her upper leg with a locked lower leg. The big difference will be her upper body swaying instead of having her ankle sway her entire body. There shouldn't be a servo in her knees. http://davidbuckley.net/DB/A_Minimalist_Approach_to_Biped_Walking.htm#TecFoot Her arm looks perfect, hiding her forearm swivel with black makes is a good idea. Would be extra cute with ribbons. Also, PLA can be welded as a possible way of adhering parts. https://www.youtube.com/watch?v=Q-PgjTYsAFY
>>14384 I would prefer bendable knees for aesthetic purposes. Oddly did not have this design in mind at all when designing her but its similar to how she came out with the extra mass on the feet. He's using a soldering iron to weld it and interesting how he takes the spool to hold a piece of filament he melts manually. Looks similar to how 3D printing pens work. I was looking into some hinges and buttons to hold parts like the head hatch in place with what I learned from the support structures in the arms. >>14383 I do. It is the same one used in the Madoka.mi Design Document.
>>14384 >PLA welding Yeah, but if the whole build is made out of PLA, keep in mind that it also fails at about 60°C. I could imagine it might reach these temps close to the servos, or just standing in sunlight. ABS, ASA or some not printed material might be better there. PLA also has force creep, it slowly gives in to any permanent applied force over time.
>>14416 If we angle her lower leg slightly, she will look fine when walking. Made a quick edit of a pic to show what I mean. Of course, her lower leg should be posable by her Anon. >Hinges and buttons Definitely needs hinges but, clips are easier to make and replace.Buttons are good if we are using pre-fabricated parts. >>14427 PLA is for prototyping and for Anons to make their own versions rapidly and affordably when possible. I fully plan on making resin parts that will be durable and somewhat resilient to heat. I've been printing for years. ABS is shit for anything large. It also has horrendous dimensional stability unless it is in an ideal environment and dry. I can not recommend it unless you have an enclosed printer, dry filament, and are willing to machine/file the parts. But, everyone is free to do things as they like. I stick with PLA for its structural stability.
>>14429 ABS/ASA was meant to refer to small parts close to the servos, bc temps. Maybe PETG or ASA might be better for the shell if she's exposed to the sun. I mean after prototyping.
>>14429 Made some improved versions of her soul gems today. Found that changing the color of LED's in flashdrives that come with them would be easier than making it more complicated than it should be. They light up when connected and could probably have the lights turned off with some programming.
>>14468 To clarify: I meant an on/off toggle. Have not thought about how she is going to speak or how the ai is going to work yet beyond movement. How would she be able to dynamically generate relevant conversations beyond picking and choosing pre made topics and responses from a text file?
>>14432 I've had good luck with PETG and it does have good thermal resistance. Definitely worth using. I use it for parts that need to last for years. >>14472 Good to hear you're getting development done on her soul gem. As for dynamic conversations, I've been thinking of using sentence fragments that would be dynamically used as needed. So she would say "It is _ in the afternoon" or " It is _ afternoon". The second part will be different depending on a random number generator so she appears more lifelike. Another example would be "My favorite fruit are apples!" could change to "My favorite fruit are apples, they're great in pie!" (Though her favorite fruit is never stated, she has stated that she likes apple pie.) This is still choosing appropriate responses from a database but, have multiple ways of saying something will go a long way towards making her seem dynamic.
Open file (113.63 KB 895x600 leaning mechanism.png)
>>14477 >Discovered the MaSiRo project Found more info on that maid project I liked in the other thread. https://www.youtube.com/watch?v=texDnpLJUMs They did in fact use some small display screens to simulate the eyes and blinking and used motion tracking cameras to do eye movement. The robowaifus in the video are essentially glorified larger Ascentos with a waifu upper torso shown in that video. They have a small following and shill patreon a lot. The list takes up at least 1/3rd of the video near the end. >Madoka.mi news: I did modify her arm segments to fit firmly and finally wrapped my head around how they installed the bearings in the wrist and connected each segment with the various attachment points for screws. also slightly changed the shoulder so she could bend her arm without damaging her torso like in pic related. >need another servo We will need a different simple turning servo for the base of the hand shoulder, hip, and maybe front wheel joints so she can turn her extremities normally and turn while moving. Did not model the small internal bearings for the wrist yet. Would also recommend looking into some simple spinning servos for the shoulders and hips since that is what you recommended for them. Reachy has some but I would rather use my own work and different parts and thought of a simple leaning mechanism while typing this in pic related. Blender was being a pain in the ass today when I went to model the interior of the forearms and making the notches for the screws. Had also come up with an even simpler one but it doesn't support leaning back/forwards. I have some mechanical experience so I assume we will need to house the motors in the torso and shoes as well with the arduino board going in the head for easy access to the LCD's and maintenance and the power supply also going in the torso. As for recharging? Put the USB-C port on her upper back would suffice with a removable cap to prevent any damage that stays attached to the torso, and I already have a functional design for that. >possible working eyes Could implement something like those maid robowaifus into madoka.mi's design with the clips housing the cameras since there is more than enough room in them for it, and also save some time and trouble with an actual eye mechanism and just have to make a simple eyelid mechanism while I wait for some test prints to finish for her arm. Would just need to see how they got the eyes to display. I assume it was the custom Live2D model method that is popular in asia. Can probably find some cheap LCD's and come up with something similar. Could share some eye textures I made after I started on madoka.mi when I followed some tutorials and feel that may be the quickest and easiest route to getting the eyes to look like in the picture with some transparent plastic covering the eye sockets as a step to be waterproofed. >hiding joints. Was also looking into methods of hiding the joints like in Allie's design. At risk of reinventing the wheel I also included a hand design. There's probably better ways to do it.
Edited last time by AllieDev on 11/26/2021 (Fri) 19:47:16.
Open file (575.27 KB 2500x2500 FollowMeEyes.jpg)
>>14492 Though I really like the MaSiRo project, their LCD eyes creep me out. I prefer glass follow me eyes that rely on natural optics to always look you in the eye. >Servos I do like your body tilting mechanism. I'm unsure of why you'd want to use different servos for her hips, shoulders, and her wheels are going to have drive motors but, are otherwise popped in and out by her Anon. The servo we are currently using also comes in a 35 kg/cm variant if you feel we'll need more power. >Simple spinning servos Do you mean continuous rotation servos? Not sure how those would benefit her design compared to standard servos. You are correct that motors need to be as localized to her hips and feet as possible to lower their torque loads while providing inherent stability. We already agreed on the USB C port, adding a removable cap is a good design choice to add. Hiding her camera in her hair clip is a decent idea. I also like the idea of her soul gem in front of her chest having her camera.
Open file (146.94 KB 1195x715 Generic robot arm pivot.png)
>>14498 > continuous rotation servos I just mean we need to have the ability to rotate the arms and legs forward/backward. That is also incorporated into that design. >>14384 With the open source Reachy, they used a 3 different servos. One kind was a servo for the rotation attached to the central slot of the bending servos in the shoulders and they used another one for the wrist. This version that I'd been using for study didn't have really any internals and was literally just the upper torso and arms on a stick. The neck mechanism wasn't even there. Was planning to hide most of the connections internally and for the rapid prototypes I will remove segments of the outer wall so I can actually get in there and attach the screws. I did not order any of the parts on the list yet. I know there is plenty of insightful info in the robowaifu archives pinned to the front of the board that can be of use for neck mechanisms worth looking into. I also could just not overthink it and miniaturize and adapt the same tilt mechanism for the head and seat it in the neck or upper chest using the smaller servos. I am no expert on motors and have not installed any of that into the chassis yet. >Screw sizes. Was also wondering about screw sizes specifically for the non standard parts like the wrist and for simplicity wanted to stick between one or two of them when it comes to attaching parts. My gut feeling is a 1/4 inch as a standard would be the safest and most reliable with longer ones for connecting the joint bends to the parts. I had also kept the insides of the limbs hollow to allow for wiring to easily get through. while incorporating some of what I learned with the wrist for better connections to the shell and seeing how they seated the servos inside the arms. Just missing a servo to rotate them. >LCD eyes Would make for a good horror movie gimmick but if that isn't a good idea then I would need a yellow LED inside a printable version of her Irises Which I already have made. Could probably put it on a stick to simulate motion tracking and have a white background to look more like anime eyes and a translucent covering for the eyelids to roll over. The only downside is that it wouldn't change size but could also be held up on a mini arm since the iris and would be light and allow her to simulate some expressions like eye rolling. Just a theory of mine. Now I am going to stop talking ad nauseum and actually attempt to make some real progress this weekend because its almost December and I need to have something decent to show for it.
Edited last time by AllieDev on 11/26/2021 (Fri) 23:09:22.
>>14501 >Servos I thought we already agreed on servos. >Head tilt mechanism There's no reason to over think it, just miniaturize her waste tilt mechanism. Upper chest is the ideal place to put it. >Screw sizes I agree that designs which use similar screws for everything make the most sense. 1/4 inch is honestly a good size. They are ubiquitous and easy to design around. (From my experience using them.) >Led back-lit eyes Huh, that's a clever idea for simulating motion. Good luck, you can get good things done.
I just wanted to stop by to say you guys are doing a nice job, great progress. BTW, should we dropt the 'general 3D' from the subject? My apologies if I was out of order on that one Anons.
Open file (204.00 KB 609x728 wrist mechanism.png)
Open file (889.36 KB 1517x673 Update.png)
>>14505 >I'm back Did not get as much done as I wanted to. Mostly because I used an array modifier to make accurate points for the wrist mechanism because fuck having to mess with that constantly. Also figured out a good shape for her shoulder mechanism without further butchering the arms any more than I have to fit the internals and manually deleting the screws that I merged with the joints was tedious. Will probably just make a copy and replace those models with ones without screws in the near future because all that matters is lining up the holes in a straight manner. Also had an idea to make troubleshooting gaps in her arms for better setup and maintenance or at least something with a cap similar to the USB-C port. >Wrist mechanism I did model some printable cylindrical bearings that should fit in the slots of the internal parts just like with Reachy. This is not final but don't think the smaller servos would have enough torque to turn the entirety of the lower forearm + hand + whatever they are holding or carrying and had also considered making another turning mechanism above the elbow so she can fully turn her arm like a human. Missed that on my first passthrough and also have plans to use miniature versions of a similar mechanism for the eyes and torso segments to simulate leaning. It's modeled but the holes are not cut out yet and this is part of why I think a third stronger servo type would be needed to turn the arms than the tiny ones that were specifically for the head. For the hell of it I just tried to fit one of the smaller motors at the end of the wrist and it's not centered without busting a hole in the mesh. I'd prefer to have some form of immersion with her internals mostly hidden unlike some designs such as reachy. I used 32 face/standard circles for the internals of so there only needs to be 18 bearings for both the elbow and shoulder. The hand can be a bit more independent since it would only need to rotate up/down. >Hair I got sick of looking at bald Madoka.mi, so I modeled some hair quickly and some basic ball joints for that part of the hairpiece along with that can be connected like a bigger anime figure. Can also go the wig route but figured I would try this since it had been successful enough when it came to anime figures and is also how her ponytails are attached on the official figma. Can always go back to the wig idea but I thought it would be easier to have a hollow piece of plastic attached to a peg to for easier ponytails would be one solution that won't get in the way of anything else like hair can. Though if the robomaid cafe can get it working like they do, There should not be much concern here. Also made the chobit-type ears functional but the hinge is hidden. Was looking into something like a car trunk mechanism for that along with the big head flap I posted screenshots of earlier. >>14536 > BTW, should we drop the 'general 3D' from the subject? I'd appreciate it if you did remove that as a formality, but I don't mind other anons posting their work here since I started here by posting in one of SophieDev's threads.
Open file (88.04 KB 2320x3408 ComplexKinematicArm.png)
>>14540 Her model is truly taking on a nice shape. After seeing the modeled hair, I'm starting to come around to the idea. >Wrist mechanism Looks good, clearly functional. Would be difficult to provide her wrists with significant torque. Strong servos have a lot of mass to be swinging from the ends of the arm. Moving the wrist rotation servo as close to her elbow as possible would be necessary. A long servo horn could handle that though. I'd prefer a mechanical kinematic chain with all the servos located in her waist similar to picrel but, that mechanical complexity is too much for now. I agree with everything for the most part.
>>14540 Done.
>>14542 >Let there be Light I did find a template program for changing the RGB lights on a robowaifu in arduino. https://create.arduino.cc/projecthub/muhammad-aqib/arduino-rgb-led-tutorial-fc003e >Eye mechanism Also sketched out how I wanted to go about the eyes with the option to turn the LEDs off to not hurt eyes. eyelids would work similar to kibo like I've said before but would slide over the iris plates. Figuring out where to get small cameras for the center of the irises is what I am currently trying to do along with small microphones for hearing and voice recognition in her chobit styled ears that now have a hinge similar to that anime, and a speaker at the back of her mouth for now. Would need to find some translucent colorless filament to get the effect I wanted for the eyes. >Actual RGB LED's https://www.amazon.com/Tricolor-Multicolor-Lighting-Electronics-Components/dp/B01C19ENFK This seems like it would be better suited than the diodes we do have planned as these can be reprogrammed for other eye colors at the press of a button or asking her. someone even left a helpful review for resistors based on voltage. >The minimum voltage is 3.1 V w/o any resistor, use 50 ohms from 3 to 4.5 volts. Use 100 ohms from 5 to less than 6 volts. Use 150 ohm for 6 volts, 300 ohm for 9 volts, 450 ohms for 12 volts, according to formula R (ohms) = (V - 3.1)/0.02. Warning: You have to mind the resistor power dissipation rating. That is, if using discrete, axial resistors, you may use the 1/8 watt until 9V, then you must switch to 1/4 watt resistors (bigger) until 15V. After 15V, you need to use even bigger 1/2 watt resistors. You'd be wasting a lot of power after 9V! Found some basic tutorials to better understand LED systems as this guy goes over some basics. https://www.youtube.com/watch?v=UlM5uvO7sao >Power supply What is a "Baseus USB C 65 watt"? All I can find is info on a wall charger and a car charger of the same name. >>14547 Thanks Chobitsu!
Open file (175.43 KB 917x1200 KawaiiKriemhildGretchen.jpg)
>>14584 Neat code for LED control. It would also work with servos as they both use PWM, which has a similar library on Arduino. If her camera is going to be in her iris, then her eyes must be stationary or else the design will get really complicated and the camera will become a point of failure. >Chobit ears So, Chobit Madokami is the plan now? Placing her mics at the periphery of her ears, slightly forward facing would help to emulate human hearing. I'd recommend a mic in her forehead to help with triangulating sound. Changing eye colors is pretty kino. Those resistor ratings line up with my experience pretty well. It is worth noting the different LED colors technically should have different resistor values for each color as every color has a different forwarding voltage. This can be offset with PWM, so with that Arduino library, they seem to be compensating judging by there use of uniform resistors in the example. Probably don't need to worry about it but, it's good to be aware. For instance, LEDs and other diodes can be used as high efficiency voltage regulators for low current loads with different colors for different values. >Power supply Anything with 65 watt (my estimated maximum power consumption) output will work. A Baseus USB C power bank was the cheapest option that offered that output with a 100 wh of energy dtorage.
>>14590 >Chobit madoka? I updated the documents with what I came up with for some head mechanisms. Would make more sense to me to use some of the empty space in the ears to house some electronics to make room for what I have in mind for the eyes. It is probably wishful thinking and could be dumbed down further. Not an explicit chobit, but something that works. Would need to somehow do motion tracking with the eyes using the cameras and I am not a super programmer. Did not apply eyelids to the sketch but put down the general idea. There are plenty of arduino code resources for her speech and mouth opening while talking floating around the internet with explicit blessing from their creators to innovate on them or use them with our own projects. >Where are the arm prints? Blender keeps having a fucking seizure every time I go to make the bores and boolean whatever joint didn't fit 100% so I came up with this plan instead. Had to use different techniques for the wrists as shown in the last render. This is all before I just take each part into a separate blend file and append them back into the main one. Will find away around this bullshit soon.
Edited last time by AllieDev on 12/02/2021 (Thu) 17:43:12.
Open file (529.57 KB 1458x757 Madoka.mi Ears update.png)
lmao I feel like the cura devs are lurking here. >Cura update Cura got a new update to basically make hollow prints just like you wanted with only 2 walls automatically with minimal infil via the "lightning infil" method: https://www.youtube.com/watch?v=BAtqjN2l65M >Chobit ears Also modeled a cool hinge joint I found and thought about putting the power button in one of her ears. It's Literally just a piece of plastic that pushes down on a button connected to the main board via wire. Though will change it to a safer door-like hinge later. Felt this would make more space to work with without losing the cover and fastening them to the head via inserting them through the back of the head and screwing them in. They don't rotate so wouldn't need to worry about holding them beyond that. I bored the arms and can finally print them after fixing the topology. Will use reachy's method of another servo in the shoulders to rotate the arm forward/backward.
Open file (35.52 KB 600x838 Sealdoka.jpg)
>>14596 >Updates in the document. I mostly agree with what is written. Using earrings to hide her cams and mics is a nice idea. I would add an option for her mouth to move with the audios waveform. Like a singing bass. https://www.youtube.com/watch?v=bgsQR1qECso >Motion tracking OpenCV springs to mind. YOLO is faster, more efficient, and made by someone who definitely seems like they lurk /g/. https://pjreddie.com/darknet/yolo/ >Blender seizures Been there, same thing happens with Meshmixer. Have a cute Sealdoka. >>14600 >Making what I do manually a feature Topkek, maybe they do have a lurker here. >Chobit ears Naisu ear mechanism, I agree with your decision to just use a hinge though. Would make it easier to print. Maybe the mechanism can be used for something else as it is cool. Looking forward to seeing your first prints. Chobit ears implies she can be turned on down there.
>>14601 >singing bass It’s funny how I was going to post a singing bass teardown to elaborate on the simple neck mechanism and how the skin would work. >mouth movement Remember at some point I wrote a reply where I shared a link to some open source code the creators gave explicit blessing to use or innovate on in personal projects. Was some code to have the mouth open according to how loud the audio being played is that was used for skeleton animatronics for now. Don’t know where that went. >chobits Yeah I am getting better with cura as I go. Safer to just use simple hinges held on by small screws for the like 1/8 since it will be relatively infrequent past assembly for them to be open. Would like for them to have more utility in the future where they pull out a swiss computer’s worth of random boards that have functionality from what I remember in some episodes. Had completely forgotten about that part in the first episode. Its been a while since I watched chobits. Got stuck in the slicer as the bulk of the ear just does not want to slice for whatever reason. I don’t understand how something the relative shape of a symmetrical potato with two completely flat sides, is not manifold and instead outright refuses to acknowledge the bulk of the model exists. Will just install some bores and a simple mechanical hinge and move the power button elsewhere with an led status light to see if it solves this slicing problem. Also found some good slicer profiles I am going to try out.
Open file (102.80 KB 900x1200 MoeHomuDoka.jpg)
>>14605 >Mouth movement You likely posted about that code in another thread. I just went through our thread and could not find anything. >Boards in her ears Could just be USB ports in there >Slicer Are the parts big? Cura has a hard time slicing big things. I have an I5 10400 and RTX 2060 and it took over 2 hours to slice a Neptune model with only 3 walls and my computer damn near melted and Cura damn near crashed
Open file (80.66 KB 236x439 madoka.mi rig arm.png)
Open file (223.86 KB 650x526 Curaisgood .png)
Open file (235.22 KB 676x449 curaiscool.png)
>>14608 Everything is scaled to full sized madoka. It appears really tiny in the blend file but cura scales it up to be a little bigger than a water bottle as it was meant to be screwed in from the inside of the head. Oddly the head, which is far more complex, sliced just fine even in 4.9. Looking into some tools to generate proper support for the ends of the ears and better calibrate my printer. Usb ports would make sense but a small hub of sorts would probably be easier to implement. Yolo looks interesting. No idea why there’s linux download command code spam taking up 75% of the page. Also found another source for the madoka app. most places hosting it got DMCA’d even though it is by definition, abandonware. >forgot a elbow pivot So I started rigging the parts together and came to realize that I forgot to add in a 90 degree pivot for her elbow when I tried to pose her. Glad I caught this now. Will also need to find a rotating servo because I don't think it would be that great using only one side of the same type of servo used for her elbow just turned on its side for the rotations in her arms and neck. >shoulderpads In the process I also thought maybe bores in the sides of the arm parts are not the best idea for these servos and would be less wasteful timewise to not just have the joints be already securely held in place with the screws in the bores on each end. There are anywhere from 4-8 bores and I have already edited the mesh to fit them and we are using a different joint than the examples I studied. I feel it would make more sense to add bores for her hollow shoulderpads. >fixed the slicer Turns out I just needed to remove that stopper part I put at the end of the ear. Blender and some modifiers can be retarded like that sometimes. Also found turning this part upside down would make the least likely to fail print with some generated supports touching the buildplate. Edit: Merged posts, formatting
Edited last time by AllieDev on 12/03/2021 (Fri) 23:08:50.
>>14605 >non-manifold This has nothing to do with the complexity, it means your mesh has bad structure. Files from video games, and really anything created in a mesh editor (Blender) are going to be non-manifold. A "manifold" mesh is a single, closed, water-tight form. Any holes in the mesh, overlapping geometry, or badly formed faces will make it non-manifold. Modern slicers try their best to work through non-manifold meshes, but if the issues are too severe you get malformed features or even entire missing sections. In Blender, you can highlight the non-manifold sections of a mesh by going into edit mode, clearing any selection, then going to Select->Select all by trait->Non manifold. They also recently added a 3D printing addon (installed by default but disabled). In it you can use the "make manifold" button to do just that.

Report/Delete/Moderation Forms

no cookies?