/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.

#WEALWAYSWIN

Max message length: 6144

Drag files to upload or
click here to select them

Maximum 5 files / Maximum size: 20.00 MB

More

(used to delete files and postings)


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


Feasibility and thinking about right now Robowaifu Technician 11/21/2019 (Thu) 06:48:16 No.1525
So I believe the endgame here is to get a fully open source and home-buildable robot waifu available to the masses. I'm imagining something along the lines of that Will Smith movie "I, Robot" except they do a bit more (well a lot more if you know what I mean) than get the newspaper for you and help you with Siri-tier "digital assistant" bullshit.

The problem is that in terms of hardware I know of no one better at the moment than Boston Dynamics and unless one of you works there and can build like fully capable bipedal waifus, I think this falls in the realm of "waiting" for that kind of advanced technology to come to us.

None of us are getting younger and none of us want to be 60 by the time we get a robowaifu so I think we should focus on things we can do right now. I think our best bets are going to be neural networks/machine learning AI and VR. Digital modeling and art, interactive VR environments, learning, speech generation, text to speech and speech to text. I think all of this in currentyear is possible.

Do we have anything in the way of real collaboration going on right now? Like on gitgud? Gathering people who are skilled at X and Y and Z and having them work on anything specific? Honestly I think we could get a VR waifu AI sort of thing accomplished if we actually channeled our autism towards it.
You have lots of good points in your post anon. And certainly there are a number of ideals for owning robowaifus portrayed in books and movies that are both motivating and inspiring. I'll be quite pleased if we can reasonably approximate even half of what's out there on this topic.

While I completely agree that we should drive towards open-source with everything we do here (I'm releasing my own software under the MIT license), IMO we don't want to hinder any anons who decide to take their own efforts towards closed, commercial ventures. I don't particularly agree with them, but I also recognize the power money has for motivating people to be inventive. If they contribute even the slightest things towards /robowaifu/s efforts then we can be grateful for that at the least. My own vision for this project is definitely in line with your stated end-game: "...a fully open source and home-buildable robot waifu available to the masses".

The hardware is an issue yes. But tbh, I don't think the problems with bipedalism for robowaifus will remain untenable for amateurs forever. Certainly you won't have to wait until you're 60 anon. Mobile Inverted Pendulum software is out there already and the issues of mechanical and kinematic coordination for bipedal locomotion are becoming more and more well-understood. Surely some autistic mechanical & software engineers will get to the bottom of an effective solution for it before too much longer. There are certainly many motivations for them doing so. My only hope is that the essentials will be open and freely available to us all to create our own robowaifus with.

However, I also entirely agree that we should take an incremental approach and achieve what we can right now, with what we have. I think most so-called engineering & math 'breakthroughs' came through this step-wise method, and most probably took years before fruition. And remember Sir Isaac Newton's famous quote.

Yes, we should be trying to achieve a VR/Visual waifu together now. For digital assets creation, Blender 2.8 is out now and has become a really a powerful tool. An anon mentioned a couple of major ecchi games that could be helpful in this area in the Meta thread. And the list of everything else you mentioned has at least one thread here on /robowaifu/ (often more than one). And it's getting easier and easier everyday to both imagine and to accomplish things like mocap and virtual assistants using cheap, commodity hardware and (often) open software.

And yes, we should be working together on coordinating and collaborating on all these things. /robowaifu/ is a decent enough platform to at least begin to accomplish this. And Torvald's git, and repo sites like gitlab & gitgud, etc., are already pretty well-established at successfully enabling distributed teams to work together on software projects. Trello is a great (botnet) tool for coordinating practically anything in a more freeform way.

So I think there's hope for success. But it will take each individual here to stop lurking and become personally motivated to put in the hard effort of their own to move us all forward. Anons who feel motivated to lead should do so, and help motivate the others forward as well. Personally, I'm learning to program. I'm learning electronics basics. I've learned Maya and I'm dabbling with Blender. I'm trying to gitgud at maths and to dabble into TensorFlow and Deep Learning. All these things took me shittons of effort and time to pick up, and I definitely have a ways to go yet. But I'm trying.

Honestly, each of us will have to be serious about the idea and be willing to put in a lot of hard work. I think we can do it together anon. I'm always hopeful.

BTW thanks for the quality OP, Anon.
>>1528
Thank you for your hospitality. I never browsed this board on 8chan or anything but it was an immediate click after I found the webring. Technology is fun and so are waifus.

>IMO we don't want to hinder any anons who decide to take their own efforts towards closed, commercial ventures. I don't particularly agree with them, but I also recognize the power money has for motivating people to be inventive. If they contribute even the slightest things towards /robowaifu/s efforts then we can be grateful for that at the least.
I agree with this. There is a lot of free (in terms of cost) software that is great at what it does and has been repurposed by the effort of programmers for a multitude of purposes. I could name these for hours, but a lot of software has been used in wrappers to be extraordinarily useful beyond their immediate potential, and a lot of hardware has been reprogrammed and repurposed to accomplish new tasks, etc. and I can even think of examples which could possibly help in robowaifu efforts. In fact, reverse-engineering both hardware and software made by those with more money than us can hopefully lead to some great prospects. For example I mentioned Boston Dynamics, and if you weren't aware, Spot (one of their quadrupedal robots) I believe has been released to the general public if anyone is able to get their hands on it, however it would surely be expensive. There is a lot of technology in the closed-source area that should absolutely be considered, if not for repurposing, for reverse-engineering nonetheless. I'm happy to provide what I know about anything on these subjects if it's even slightly useful.

>Mobile Inverted Pendulum software is out there already and the issues of mechanical and kinematic coordination for bipedal locomotion are becoming more and more well-understood. Surely some autistic mechanical & software engineers will get to the bottom of an effective solution for it before too much longer.
I hope you're right. I'm really not the best at hardware outside of regular computing (meaning no robotics and locomotion experience) other than knowing what currently exists.

>Yes, we should be trying to achieve a VR/Visual waifu together now.
I agree, I think this is absolutely possible right now. There are so many open source (or at least free as in costless) tools available for nearly every one of our needs on this front. This is surely possible.

>Personally, I'm learning to program. I'm learning electronics basics. I've learned Maya and I'm dabbling with Blender. I'm trying to gitgud at maths and to dabble into TensorFlow and Deep Learning. All these things took me shittons of effort and time to pick up, and I definitely have a ways to go yet. But I'm trying.
Keep it up anon!

>BTW thanks for the quality OP, Anon.
Don't make me blush.

Not sure what to do with this thread. I guess we could start coordinating via git repos if anyone has anything they'd like to contribute like this one anon >>691 who made an interesting repurpose of GPT-2.

Perhaps someone has a specific project in mind we could begin coordinating? VR "chat with your waifu" sounds promising to me.
>>1530
YW. One of the literal best things you can do w/o any skills whatsoever in robotics or AI etc., is simply to provide links back here on /robowaifu/ as you run across them. We have more than one thread that are more or less specifically to help consolidate this general anon's task. For example:
>>366
>>374

There are also a wide variety of other thread topics here that have more specific focuses. Just get familiar with the board anon. Many of these empty threads will be populated with dozens or hundreds of posts already made in them on 8ch as we migrate everything over.

As far as VR, I personally think our Visual Waifu thread is probably the best place to coordinate that area.

>Perhaps someone has a specific project in mind we could begin coordinating? VR "chat with your waifu" sounds promising to me.
I did once create a basic control description in C++ and posted it here, where I tried to break down every thing I could think of for the external shell of a robowaifu. This was to do two things specifically:
1) Provide a basic foundation to enumerate all the different graphical items and 3D meshes needed for a visual waifu.
2) In a similar vein, begin to lay the foundational framework for an IRL robotic gynoid, and one (hopefully) already well-tested in advance as a Visual Waifu.

Anyway, welcome aboard anon. I'd just encourage you right up front to be patient and not get over-anxious for immediate results and be demotivated by discouragement. We'll all get there, but it's going to take a lot of hard work and a good bit of time too.
>>1533
>I'd just encourage you right up front to be patient and not get over-anxious for immediate results and be demotivated by discouragement.
I won't, I'm just thinking about what is assuredly possible right now. Like I said, I know we can do things and I think if there is good collaboration we can get them done.

I'm bad with hardware outside of general computing hardware but I'll try to contribute what I know in the field of tensorflow and what is being done with it, speech, available hardware and software, available software in git repos, new code, and whatever else I can offer.

If we ever get git repos up I'll contribute testing and submit PRs and whatever else.
>>1535
>If we ever get git repos up I'll contribute testing and submit PRs and whatever else.
Well, here's the kludgery I hacked together a year or so ago, that I just mentioned.

https://gitlab.com/robowaifu/simple

I suppose we could migrate it all over to gitgud or something?
>>1544
>I suppose we could migrate it all over to gitgud or something?
Gitlab is probably fine but mirroring on gitgud can't hurt. Thank you for your skeleton, I've got it bookmarked. I'm gonna lurkmoar now as it seems I've definitely missed some things in my short time being here.
>>1545
Nah you didn't 'miss' it, the board is more or less empty atm. After the deplatforming, I began to slowly get the board back together, but got the bright idea to automate the whole thing after a couple of weeks and I've been pretty sidetracked by that project, BUMP
>(just did a new subrelease today: >>1523)

By all means, don't lurk anon, talk. This is how we get the creative juices flowing. Trust me, this is one of the friendliest communities on the IB-verse tbh.
Haven't seen this much activity on /robowaifu/ in a long time

>>1528
>Yes, we should be trying to achieve a VR/Visual waifu together now.
Since my first post on this board years ago I've been saying this. I've done some basic VR tests with UPGE and will doing some more with Armory later, both of those are closely associated with Blender. There's also Godot which could be an option for the engine, I've given up on my previous plans posted here >>1507

There's already something called the Viva project that's very similar to what's being discussed;
https://www.youtube.com/user/mrhalholhel1900/videos?disable_polymer=1
I've tried it out awhile back and it really isn't my cup of tea but most 3d jap ecchi games aren't either. Everyone has different things they want to do in a waifu VR sim and different ideas on how they should be done so I don't have much hope for a collaborative effort in making a general purpose one here.

Sharing assets and tools while quickly getting people up and running on building their own seems like an easy enough project that's attainable.

>>1530
>[Spot] I believe has been released to the general public
They only lease it out to companies for testing modules that attach to top of it or for trying out novel use cases, there are similar Chinese robots that are for sale though. I'm also a big believer in salvaging tech.
>>1547
Hey there anon. Ah, so that was you ehh. I've downloaded both the ecchi cracks but haven't had a chance to dig into them yet.

>VivaDev
Yeah, I was following him back on 8/agdg and tried to convince him to open up his code specifically for /robowaifu/s benefit but sadly he declined. Anyway, his project is really pretty remarkable as a one-man-show and he made me realize I had a penchant for vampire loli waifus I never realized before haha.

So, yeah. We already have /robowaifu/. It's obviously not the perfect platform, but frankly I don't know of anything out there that's better for the anons of this world tbh. Let's get something going.
>>1546
>By all means, don't lurk anon, talk.
Just don't wanna state anything previously stated or miss anything pertinent.

>Trust me, this is one of the friendliest communities on the IB-verse tbh.
I believe you.

>>1547
>There's also Godot which could be an option for the engine
I'd definitely give Godot a try if free/open source is important to you, it absolutely seems the best supported. I've never given it a go but I need to.

>They only lease it out to companies
Well, fuck. I could have sworn a couple years ago or so they said they were going to sell it.
>>1549
>Just don't wanna state anything previously stated or miss anything pertinent.
I don't think that's possible anon. In the 3years of /robowaifu/, we have approaching 6K posts, most of them decent or even high quality. If it's part of the gynoid robotics or vr waifu world (from a serious engineering perspective) then we've probably already touched upon it here. Just speak your mind.
Open file (157.50 KB 960x540 VRM_WorldConnect_EN.png)
So I've been looking into VR waifu simulators and stumbled upon 'Waifu Sex Simulator VR'. That program doesn't interest me much but what's powering it does; MocuMocuVRM

I've heard of VRM before but in case anyone here isn't familiar with it
https://vrm.dev/en/
While the VRM format is based on open standards from Vulkan all the tools and development for it are stuck on Windows and the Unity game engine at the moment. As it's a brand new standard there aren't many open game engines that support this. It seems to works with Godot 3.1.2 if you rename it to .glb but it doesn't work in Blender 2.81 .

This will probably be the new standard for VR avatars and 3d anime characters, it's got significant industry backing with wide adoption in Japan thanks to Pixiv. The Waifu Sex Simulator VR mentioned previously has hundreds of models to choose from because of this.

Instead of jumping straight into a full VR capable waifu simulator I'll try making a simple interactive desktop virtual assistant using Godot.
My two cents: incorporate Visual Novels into the mix. I'm one for bringing together previously unrelated communities which can lead to unexpected collaborations and insights. Scripted storylines can be combined with more flexible interactions. In between scenes you can have playful interactions like with VivaDev. Eventually as AI improves even the "scripted" interactions can become less scripted. Perhaps they never play out quite the same way twice. The point is, this offers you a path to incrementalism. Scripts and structure can fill in for the short-comings of AI as this technology gradually improves. I'd say reach out to those communities. There's an active modding community on reddit for the DDLC franchise (if that's the right word). They include mods for choice-driven conversations that are under continuous development by writers. Maybe this could have AI slowly introduced into it as the realism improves. You get the benefit of a fully fleshed out character with back-story that you can focus on emulating, and people respond to a beloved character more than a generic one. Plus you get to take advantage of their viewership and popularize your ideas in a larger community.
>>1857 >Instead of jumping straight into a full VR capable waifu simulator I'll try making a simple interactive desktop virtual assistant using Godot. I sure hope you do that Anon! Good luck. >>4336 >My two cents: incorporate Visual Novels into the mix. Very good idea IMO. In fact, I intend to incorporate sets and scripts copied directly from Chobits once I have the MRS into a workable state. It just seems a natural choice for me since I've marathon'd this so much. I think I'd have well-tuned instincts to check on the quality of the AI solutions as they work towards the 'perfect-script-reading' goal in this type of scenario. >In between scenes you can have playful interactions like with VivaDev. Good thinking. I tried to interest him in /robowaifu/. Maybe he'll look in on us someday heh. >I'm one for bringing together previously unrelated communities which can lead to unexpected collaborations and insights. >I'd say reach out to those communities. Not a bad idea. One of our vols, Ribose, sort of took the reigns with the whole propaganda thread. Maybe he could be encouraged to follow up on your suggestion Anon? >>2705
>>4357 I've considered other possibilities with regard to using AI in combination with pre-scripted interactions. All of the pre-existing porn on the Internet could be retro-actively rendered into a 3D interactive environment. A.I. software could "watch" through a video in order to form a 360 view of each person as well as a panoramic view of the room. Any details not shown in the video could be improvised, just as a professional artist would improvise. Furthermore this could be deep-faked into a scene with your favorite waifu complete with voice conversion and all kinds of other little important details. Combine all this with VR lenses and a waifubot. Even a regular sexdoll. As far as what we could do right now, start by improving on deep-fake technology. Since so much of /robowaifu/ is into anime maybe build something that can convert any porn image into hentai. But do it with an eye to quality. Highly polished hentai, not something with obvious polygons in it.
>>4357 >Maybe he could be encouraged to follow up on your suggestion Anon? I just posted a link in that thread.
>>4373 Good thinking. Thanks Anon.
>>4357 What is Chobits?
>>4593 The anime where /robowaifu/ mascot, Chii comes from.
We don't have an official VR waifu thread, but since this would cut out the body and make it easier and more feasible, I think this here is the right place. Good news again, the haptic feedback hardware for VR games and VR chat seem to make huge progress. Also, even the current popular VR chat software supports it already (nothing sexual, I think): https://youtu.be/wBV9hDNd2ws The downside is, it's still very expensive. So if this is the easiest and most feasible way or not depends on one's priorities, I guess. Also, I'm sure it's going to be much cheaper in five years. The other thing is, that this is not about sensing touch yet, only little vibration motors letting you feel something on parts of your chest, arms and feet. Okay, then there's also brain controlled VR. Fascinating, but not sure if this is usefull, related to waifus: https://youtu.be/xAzdH4fFbxA
Also, the thread about "early business ideas" is inherently linked to early robowaifus as easy and feasible as possible: >>3119
>>5372 >We don't have an official VR waifu thread I think our Visual Waifu thread is about the closest thing atm. >>240

Report/Delete/Moderation Forms
Delete
Report

no cookies?