>>28152 trying to build high DoF face.

Here is my idea for the mouth

For the flaps we could use 4 of these

>>28152 >>28153 Yeah, it's a personal interest for me too Anon. I'm very concerned about mass, so I'm pursuing non-'traditional' manufacturing approaches & designs. Good luck with your work! >>28158 >>28159 Looking forward to your facial design progress, Anon!

>>28198 thank you.

>>28205 Y/w Anon. Just keep focused on what you can achieve, and guide that energy towards your goals! :^)

>>28198 >I'm very concerned about mass same, but in this case, I think the concern is minimal. Take the neck for example. I took it from Eva. It's already extremely strong, if using the six 9g mg90s which are only 1.8kgf cm @ 5V, it moves around 19 mg90s + silicone + plastic framework + other parts. on top of this, i found these metal gear ds power 21g servos, they do about 4.2kgf cm @ 6V and are barely larger than a sg90/mg90s, (it's like mm difference). Way stronger, much quieter, tiny bit faster and far smoother running than real tower pro mg90s. the 21g are way too expensive however at 15 usd a servo.... but they are worth it on sale, i've seen em low as 4 usd. it's not so bad for the dof's that need strong forces I started originally converting the Eva head from Colombia university to use all 21g's in the same locations, and keeping the 14 line pulling facial mechanism but to basically make it silent, stronger, and faster. but after some rudimentary testing i decided against it as the string system limits constructability far too much instead im modifying Eva to use slotted or tracked pads connected to servo linkages. the pad contains a pattern of magnets to interface with internal skin magnets which are littered around the movement areas (cheek, lips, eyes, eyebrows). changing the pads shape and which internal magnets it interfaces with, allows for facial expression refinement and/or separate silicone faces using the same internal head. the constrained magnetic pad system is how the best quality/life-like animatronics in film are done in my opinion. since the skin isn't attached directly to anything but by magnetic fields, the skin has a real physical compliance you can't get anywhere else. >so I'm pursuing non-'traditional' manufacturing approaches & designs. i am excited to see how it goes. the more experimentation the better imo. >Good luck with your work! you too

>>28293 I only like the top-left one but I don't think it'd look great in a robot. When designing my robowaifu's model I zoomed into her face 1:1 on my screen and looked up close to get a sense for how she will feel in real life and it completely changed how I designed her. Things that seem cute on the screen or at a small size, like dolls, can feel really uncanny up close at a more lifelike scale. Some anime figures would look cute scaled up, like pic related, but I'm not really a fan of them either because it's not just about how they look but the whole experience of being with them for me. I really want to nuzzle with my robowaifu so I gave her a nose. It might be silly but I put a lot of thought into what daily life will be like together. I want her face to be lovely in all situations, from across the room, up close, kissing, looking down at her in my arms, waking up next to her, sitting side-by-side resting our heads together and more. I'm considering resculpting her face in clay because working in 3D leaves too much wanting. I have no idea how her face will feel once I hold it in my hands without burning a ton of money on prototypes I can't even adjust. Clay would be much easier to work with. When I'm finished I want to be able to squeeze her silicone cheeks in my hands and feel that they not only look good but feel good too and fit in my hands perfectly.

Doing dev work on a basic head. My head is the bottom one next to the human reference. Any advice or suggestions? Which head do you prefer?

Yours is quite fine but hard to compare since it's less finished, the middle-top has too small eyes and the forehead is too big. The nose of the one on the right isn't good. The one with hair and the male one are only confusing. Overall, I like the middle-top the most, but I don't really like it. I think it's because the mouth and nose look better and the ears aren't too big. I made a similar comparison here: >>18868

>>28356 Middle top looks like a baby head, not really a good thing imo

>>28354 The one to the right is 90% of anime girls

>>28357 Kinda true, that's one reason why I suggested improvements. The better nose compared to the others is relevant to me, but the eyes are too human-like and the forehead too big. Just look at some of the 3D pictures I post. I'm not gonna go with a "anime nose". That said, I don't think it matters necessarily for the first iteration. >>28358 Yes, but they're 2D.

>>28348 You get it. Her head is dearly important for many interactions. We need something we can look at as we fall asleep. Here is my new design, going for a mix of real and Aigis. I gave it hair to help

>>28364 Nice moodeling. The only thing that jumps out at me is the bridge of the nose: either too narrow, or the edges (sides) are a bit too sharp for my sensibilities :)

At this point, I think it's mostly about getting the eyes right. Thanks for the feedback, it was helpful.

>>28354 >>28364 Sorry I haven't responded earlier Kiwi, been a bit distracted. (From the first post:) I think the two most-appealing forms are the one on the far right, and the one on the bottom left. The Aigis one is fine and so is the girl one. Obviously, the male(-looking, at least) one is out. >>28397 Is that a single print? I'm presuming those are cutout eyes, but it makes me think of screen eyes like Mechnomancer's. Do you think you'll go that route? Cheers! :^) >=== -minor edit

>>28252 >i am excited to see how it goes. the more experimentation the better imo. This. Me too, Anon! Cheers. :^)

>>28397 Generally it's best to leave about one eye width of space between the eyes, going by the width of the eyelashes. Also keep in mind that once you do the faceup, the sides of the eyelashes will extend beyond the edge of the eye socket. I'm looking forward to seeing your progress!

>>28398 Thanks for the feedback! Yes, it is all one print. Those indeed eyes cutout from a photopaper. Using standoffs for the follow me effect. So far, it actually works, it just feels like she's constantly glaring. Her eyes need to be bigger and the spacing needs adjustment for her eyes to follow rather than star. I'm honestly excited how well it works. See picrel, she haunts me wherever I go in my lab. >>28407 This is true. Her eyes being spheres is also a problem. We have almond eyes, the spherical look is cute in anime but, doesn't work IRL. Thanks for reminding me how important eyebrows and lashes are.

>>28411 >This is true. Her eyes being spheres is also a problem. We have almond eyes, the spherical look is cute in anime but, doesn't work IRL. Why?!

>>28411 Maybe you could try making the irises bigger and a slightly darker blue.

>>28416 They're more human like when they're more elliptical. It didn't capture well but, this face is almost cute at the right angle. >>28441 Will try that tomorrow. Biggest update is her head is near the average human female in terms of dimensions. This means she can wear hats and wigs normally. Men's hats are still a tad big on her.

>>28455 Nice. Do you think you'll do any articulation on her head, Kiwi? Her mouth for instance. Regardless, relly good to see your progress! Cheers. :^)

>>28455 why does this look so familiar swear theres a statue that looks exactly like that, maybe an ishtar or moai

>>24464 I've had the idea of screenfaces like the one in >>24450 for a hot minute now, and I've had one fanciful thought about it, specifically the possibility of having said screen be touchscreen. This would allow for settings/configs to be managed directly on the girl herself rather than having to boot something up on your computer (as well as possibly other things, who knows). Regarding this, thoughever, I have no clue if it's economically/technologically feasible to have such a large curved touchscreen.

Moved to faces because reasons >>29394 >>29414 >>29422 To an extent they already make heads this way. They mold hollow rubber heads to go on fiberglass skulls. Unfortunately for us, the skull face is usually just a hollow void to allow the head to be used for, uh, other activities, so we would need to build a structure to support facial articulation and expression. See picrel. Even so we wouldn't be the first to do some modification for electronics- see picrels. And skull design seems to vary between manufacturers. We may want to do an open source animatronic skull for dolls project that anons could print and then fit their chosen doll's face onto by carefully removing material from the inside until it fits. Some adjustability in the skull would also help. So the doll makers are already familiar with the concept of hollow rubber parts, just not beyond making heads yet. btw these pics are from this thread on the doll forum inventor's page- NSFW: https://dollforum.com/forum/viewtopic.php?t=91728 I almost renamed the first pic "flaccid face", but considering the subject of the thread it would have been too easy.

>>29423 Thanks for the great information, Robophiliac. Cheers. :^)

If you are into a more robotic look you could get full face sunglasses, they are a couple dollars on places like Temu and Aliexpress. If you want expressiveness you can put LED grids behind it to have eyes and a mouth shining through. This setup would allow you to use any number of cameras or LiDAR hidden behind the visor.

Because I'm lazy, I figured a screen-type face would be a better option if you simply wanted to get a product out the door. Modern OLED's allow for curved screens, so you could affix it to a skull shape without it turning into a "TV head". With eye tracking tech, you could show where the bot was focusing on to allow it to look at things. As with all screen-type faces, you don't have to worry about mechanical eyes and mouths so less moving parts to break or make noises. >But wait, there's more! When it comes to customization you could have all sorts of options; eye color, mouth shape, nose shape, all of those options are scaleable and interchangeable. If you 3-D print something and realize it's the wrong size, you have to reprint it or shave it down or something. If everything is based on .pngs and assorted digital assets, it becomes much easier to fix designs on the fly with a screen face. My main issue is the combination of digital eyes with a face plate. As much as I love those maid bots with the illuminated eyes, they're kinda odd when I look at their face. I'd rather they have a fully digital face. See pic related. I won't say it's uncanny valley territory, but it's off-putting for me at least.

>>30632 Do you know about Spud? >>30227? The little mess around one eye comes from damage that happened to the screen.

>>30632 Like NoidoDev mentioned it has been done. I will add to this that circular screens are on the market now. I had also suggested using eink displays to prevent glowing eye effect for if using other materials for the face. Not sure how it will look till someone does that.

>>30645 The refresh rate for an hobbyist e-ink screen is too long for any realtime video, and takes like 30 seconds of the screen flashing to get a pic.

>>30653 Ah, that's too bad. Not all eink does the flashing thing all the time though, that's just to maintain the image quality to prevent a burn in. I've seen full animations done on eink. I just dont know where to find the screens by themselves that are of this higher quality able to display video.

>>30640 Not explictily, I've only skimmed a few threads as of late. This is the only place on the net to dig deep when it comes to robot waifu development. Looks pretty rad. >>30645 >Like NoidoDev mentioned it has been done. Neat. I figured someone might have. If anyone had, it'd be Robowaifu no doubt.

>>30655 I don't think they're quite yet available for the DIY market yet.

>>>30693 There is some it seems though still slow for frame rate they do update fast enough to display animation. Apparently you have to look up "fast update" and "partial update" to find them. There is a difference between the two but im reading one claim fast update isnt faster than partial update when it comes to small screens though I am unsure how true this is. You just have to be sure it has the ability to run in a mode where it doesnt do an inversion image. You should check if there is some alternative way of running the eink displays you have or not. Not full color but these are multi color available a three or four colour would be enough for eyes especially if dithering is used to produce shades. Though just a monochrome can work technically though all black eyes is a sort of dead eye'd look lol https://www.pervasivedisplays.com/products/?_sft_etc_itc=pu https://github.com/PervasiveDisplays/PDLS_EXT3_Basic_Fast Would have to keep digging in order to find more. There is a video here I also came across. Skip to 4:13 for seeing partial update. https://www.youtube.com/watch?v=cUylSiuoLHc&t=64s

Some animatronic channel by Gary Willet, I post it here, since it's mostly about (monkey) head design. >My Art of 3D printed Animatronic Designs, I document everything, processes and techniques, Instructional How to do it. https://www.youtube.com/@garywillett4146

Hannah-dev Anon has a quite interesting recent facial articulation system. https://www.youtube.com/watch?v=UoSK_w19kUI

Some videos I had on my list, some have already been posted in this or other threads: Hannah mouth test (David Browne): https://youtu.be/2li8Sq9Un38 Will Cogley's robotic head: https://youtu.be/eUMVtoO_fS8 Hannah's 7 motor jaw: https://youtu.be/UoSK_w19kUI silicone doll: https://youtu.be/WxQgEL4LlN0 3D printed teeth (Garry Willlet): https://youtu.be/VUxryzJJevo mouth movement speaking: https://youtu.be/4pK5x_vOXTk animatronic eye mechanism without fasteners (Will Cogley): https://youtu.be/uzPisRAmo2s facial skin animatronics (Garry Willlet): https://youtu.be/WtIdlhdrI2c animatronic head (Garry Willlet): https://youtu.be/vt2uMbHBNw4 Lower lip linkage (Hannah / Browne): https://youtu.be/vocoKKmszUc some human-like head: https://youtu.be/uqxhR49N3ws Hannah smile (David Browne): https://youtu.be/Esw5gjrFL-w

>>34192 Great information NoidoDev, thanks! Cheers. :^)

Hey everyone. Have you guys seen Milky Highway on Youtube? There is a really good candidate for a robowaifu I would like to discuss. Makina is a robot that has a screen for her face and she shows the cutest things like a loading icon when shes calculating driving distances, autopilot mode when shes doing something but not busy and then she has full blown Facial expressions. Also shes a gamer robot girl and thats good. What I'm getting at is that once Teslabot is released I think it would be much easier to just use and LLM like Llama 4 (once it comes out) or maybe even whatever version of Grok the Teslabot is likely to use and have the robot show dynamic things on its screen. This is just my idea for what I am likely to do for a robotwaifu's face.

>>34231 Hello Anon, welcome! Please have a good look around the board (via the catalog) while you're here. >Robowaifu screen face Yep it's a great idea, Anon! You might specifically see what Mechnomancer's adventures with them during his continuing development of his dear SPUD robowaifu : ( >>26356, >>28131, >>29420, et al ). Also, there are some similar ideas floated in our Visual Waifu thread : ( >>240 ). Good luck with your work, Anon! Cheers. :^)

>>34311 Intredasting, thanks Anon! Still uncanny af, but it's pretty good progress if that's the route the devs want to go with their waifu headpieces. Also, it's clearly a phenomenon that otherwise seemingly normal adults will revert to childish antics when testing controlling a waifu -- even destructively so. We should expect this type of juvenile behavior and plan for it accordingly when displaying our prototypes to the public at conventions, etc., in the future.

A fellow is working on a pretty nice silicone face setup. Got some Will Smith "I Robot" vibes lol. https://www.youtube.com/watch?v=yWrldOS6xBw

>>34311 That's amazing. I think we could do something like this or have that level of feel with some sort of AI. You can do face recognition with a ESP32 microcontroller now. So I'm thinking the compute is not so high for this.