>>13648 I'd be fine with sk8o dressed as Senko San and basic conversation. Though I do want her to say "kon"

I plan on literally starting with an action figure. I'm not really good with my hands and don't really know what I'm doing. And with how long it takes me to do anything, I'm probably decades away from something that'd probably make me happy. Not to mention that I just can't learn to code no matter how many times I've tried, so I'd have to get money before I could pay to have an AI made. The best I could do on my own would mostly just amount to a glorified mannequin/puppet. I guess on the Human-Mecha scale, I'd be starting at 4, then gradually working up to 1. Anything that I can't get looking and moving realistically (mechanically, not because of the programming) within realistic human body proportions, will instead look overtly mechanical. And as I mentioned in another thread, the brain will be a server rack and the body will be remotely controlled. As far as the action figure goes, right now I'm just trying to figure out how to make something with a realistic or somewhat idealized range of motion with as few parts as possible. I keep trying to figure out how to make a near 180° bend at the knee without a double-hinge.

>>13654 >I keep trying to figure out how to make a near 180° bend at the knee without a double-hinge. use a single hinge?

>>13654 Action figure joints: >>7634

>>9555 shows a method to describe the minimum requirements in different areas and then how to level it up. Here some examples: - Body movement RL00 - moving head, legs and arms, but allowing them to be moved RL01 - walking on all four RL02 - dancing on one spot with guidance from a wall or pole - Facial expressions and abilities RL00 - nice looking smile RL01 - general cute facing - Endurance RL00 - movement in bed for one hour, some time talking and min. movement RL01 - ... - Comfort - internal self cleaning by drinking water and cleaning fluid - internal storage of cleaning fluid and also lubricant for sexual usage - release of internal fluids in bathtub, shower or on toilet Generally the early trend of the initial discussion back on 8chan was: Having a "sexbot" (animated doll) which can move around in bed, and being able to move to the bathroom and clean itself up. Follow up wishes were "making a sandwich" and "washing my car".

>>13655 >use a single hinge? You don't get nearly as much range of motion >>13659 The pics include a figure with no hinge at the knee, and one with the exact kind of double-hinge that I was trying to avoid.

>>13654 Prosthetics are your friend. Human legs only bend to about 140 to 160 degrees at most.

>>13675 I should probably look into human kinesiology more in general, but the flexibility of the arm or legs is mostly limited by the thickness of the limbs. It's easier to make limbs thicker than you need to, but harder to make them smaller. Making a slim, waifish contortionist waifu is a good goal to aim towards, because you can always add to it, but if you design something clunkier, you might not be able to take off of it.

Minimum? Hardware to store and run her, a screen to see her, the ability to talk to her through keyboard or microphone, and of course a motorized onahole.

>>18248 motorization optional

I'd be fine with a smart doll that can watch and read things together with me and remind me when I'm forgetting something. To break that down into steps I'd pick: >basic conversation >text commentary >video transcript commentary >some simple to-do list feature >visual commentary >actual memory I'd be pretty satisfied with that. A full-size robowaifu that can move around would be a luxury.

>>18252 Certainly reminds me of Sumomo-chan's (>>14409) project goals Anon.

>>18252 >text commentary >video transcript commentary >visual commentary I don't understand what you mean by those.

>>18260 Like reading a book or transcript of a video and giving an opinion on it, and noticing things visually like you dropping your keys and saying something about it.

>>18264 You're right though: this seems to be something AI *should* be capable of but just isn't or hasn't been worked on. It should be simple to create systems to "appraise" music, art, writing based on finding self similar patterns (beauty, order) relevance to other works or concepts of importance, and also comparing the qualities of said to the reviews of others. Results would be interesting: "Waifu, rate my writing/music/artwork". Right now chat apps can only give you lip service or tell you "yes its great". But it would be nice to watch a movie or listen to music w/ your waifu and be able to discuss it too.

>>18266 lol ok not simple, but knowing how to proceed should be simple the execution will still take a lot of work

>>18266 This is where preference models become handy because they do exactly that, rate things. You don't want a model to generate the most likely code, art, music or writing. It needs to be the best or it won't work. Preferences and values are what create a personality. Something I've been working on is making a generalized preference model so users can define what they want in natural language and it will perform as well on their preferences as it does on mine even if we disagree.

>>18271 >JUST INSTALL GENTOO muh sides

Will Cogley came out with a snap fit eye mechanism (no screws needed). I put it into the vision thread: >>31472 Personally I wouldn't use this in my robowaifu design, but I think this could be good base to have a module of eyes to carry around, maybe put cameras into a improved eye design, add something like a Rock 5B with Whisper voice recognition, and add a computer that connects to other computers on the network which then can do object recognition, AI responses from an LLM or more into the device in form of voice. Obviously microphone and small speakers would also be necessary. Add some hook to carry this simple design around when you're at home, and use this to think about how the AI and whole system should work.

>>31473 Very nice improvement to his design. Assembly complexity is a deep and important topic for those of us anons who want to devise our own for-sale robowaifu kits targeted at laymen anons. Thanks for bringing this to everyone's attention, NoidoDev. Cheers.

The other day I was thinking about what exactly or roundabouts it would cost to make a robowaifu. Here's what came up with. I think I read the present processor in a Tesla Model 3 is $35. Let's say you add ten of these for more power. $350. Maybe $600 for all wiring, other semiconductors and power supply. 2,500W/h per day for batteries. At $132/kWh we have $330. Maybe $120 of plastic and aluminum. Comes to $1,400. You could surely build one for even less than this. The key, the major big ass key is being able to assemble all these parts without some major fiddling with a ton of strings or wires or soldiering all of them. This is the ONLY reason I have backtracked on the idea of using hydraulics. With hydraulics you could make a large bladder and cast the whole circulatory, muscle and maybe even the skin with sensors system in one operation. It would save a huge amount of labor. There's a rule of thumb that for mass prodution you can get the cost of an item to material cost plus 10%. Obviously this doesn't always work but it's a general rule. So thinking about material cost should not be ignored. I wonder if you made a robowaifu that could walk independently and had a little verbal ability if you could get $3,000 USD for them? Maybe it respond to simple, go here, go there follow me sort of commands. I think you could do this with a mid range desktop processor.

>>31586 >I wonder if you made a robowaifu that could walk independently and had a little verbal ability if you could get $3,000 USD for them? It may be a bit foolish, but I'm honestly hoping to fashion bare-bones-essentials-only robowaifu kits that can sell with a smol profit at US$2'500 . It originally was Dollfan's estimate which, admittedly at the time I considered slightly wishful thinking -- but have since agreed, but as I've developed my ultra-low-cost strategies I also concur with. This is counting on all the software being entirely free and open. It will be by far the largest cost factor in devising successful and pleasant robowaifus IMO (so it's important for Anons to learn robowaifu systems-software development -- we need at least a dozen devs 'bootin' to the cause! :^) >=== -minor edit

>>31592 I bet if you had a kit for $2,000 USD that could walk, talk a little and maybe understand a little direction, with personal training, you could sell the hell out of them. The kit would have not be too difficult. It shouldn't take 500 hours to build. Maybe 50-60. You could have a working mouth but leave the genital area empty or covered with cloth to thwart the feminazis. Just say you are training them to cook and clean, HAHAA. Send the kits with maid dresses to really piss them off. Of course you could add whatever genitals you wish. Are we ever going to be able to post pictures here again?

>>31593 >>31593 >you could sell the hell out of them. Indeed. I plan to be one of several billionaires selling robowaifus in the future. Also, hopefully every.single.OG. Anon here will do the same. After all everything will be free, unfettered, & open (at least insofar as my Model A contributions to this domain are concerned). The more the merrier! :^) >The kit would have not be too difficult. It shouldn't take 500 hours to build. Maybe 50-60. Yes. Simple assembly approaches will be vital for initial success. Thankfully individuals like Will Cogley are thinking along similar lines : (>>31472). I recommend every Anon here do the same with their design-thinking. >Send the kits with maid dresses and catgrill ears to really piss them off.* FTFY Anon. :DDD >=== -prose edit

>>31595 I decided to do some sex doll searches. It looks like they run about $1,500-$2,500 with a few at $3,000. The majority of higher quality seem to hover around $2,000 or a few hundred more, so as I , and many others speculated, the sweet spot is likely $2,000. I think you could sell a walking talking, uhh... "action oriented", but limited waifu with open source AI for $2,500 but if you could profit and make them for $2,000 I doubt you could keep up with demand. I suggest a way around the law would be to sell the muscles, skeleton, processor and skin in a kit BUT ship the parts separately and do not sell any sort of orifices except maybe a mouth. But there should be adjustable areas for insertion of unmentionables. Possibly you could sell these but on another site with different accounts, companies, etc. To sell you will probably have to have a very reduced motherboard,memory and process, BUT a fast built in wifi would be necessary and software to link securely with the clients computer. They could then upgrade either the onboard and/or their own personal computer to amplify interaction. Wifi should be fine for voice and general direction, moving about. I think it would be plenty of bandwidth if you used "gross" movement commands from the main computer while allowing the onboard motherboard to handle moving the body around.

>>31799 Thanks for the research, Anon. >"action oriented" As I've mentioned several times through the years here, I won't be selling any snu-snu-enabled robowaifu kits (nor will my eventual company interest), and I still maintain that every Anon here should follow that same model for their own robowaifu systems for sale. 3rd-party 'ladybits' kit manufacturers will surely pop up by the dozens once the robowaifu industry effects kick in. At most, simply accomodate their likely appearances is my advice to anons here. >home server Yes, I've already begun laying the foundations for all this with, well, RW Foundations : (>>14409) . There's a Dollhouse that fully-assembled versions will be shipped-in/stored-at-home-in. It has associated proxy/firewall/DMZ Dollnet electronics to allow for offlined/nearlined Internet access -- with physical lockout -- built right in. >"Sumomo, find good bathing software please." This will be a great start for privacy/safety/security (>>10000) for Model A robowaifus (even for completely-networking-inexperienced anons), and will also easily provide the secured interface for Anon's other, offlined server rack(s). Great ideas, Grommet. Thanks for the guidance! Cheers. :^) >=== -prose edit

>>31800 There is a surprising amount of prefab, powered fun bits already available. Given that one would need some electronic experience to do maintenance on their on robowaifu it wouldn't be too far of a stretch to figure out how to integrate accessories of a lewd nature.

>>31803 Yes, I think you're right Anon. Given that a typical robowaifu kit will likely require 50h - 100h of assembly time (including basic soldering), then such mods shouldn't be too challenging for a robowaifuist anon IMO. Cheers. :^)