Regarding full body motion capture, Move AI might be worth a look. It used to be iPhone only, but they just announced an experimental version for HD cameras and Android devices.
I’ve been playing with this. I used all the ipads in the house for set up 🤣 as could get hold of 3 Phones 😅 Thank though ! What were your results like?
There will be something for android i’m sure of it. Some kind of bridge. I quoted the tech lead on that bit as I didn’t want to be the bearer of bad news haha
Big issue with metahuman is that it creates EGGHEADS. I have 0,025mm accuracy 3D scanner (metrology grade scanner for reverse engineering) that makes acurate 3D models. And i tried to use that model with metahuman. ANd metahuman fucked it up as it created EGGHEAD. Washes out many medium and large details. Using metahuman sliders to make model look right is also dumb. They basically transfared zbrush "move topology" and "mov elastic" brushes in to sliders. And what is the point of scanning when metahuman fucks it up? long road a head with metahuman.
Yeah totally agree with you. Tbh I think metahumans are only good for the head….you are also limited with their damn neck thickness. It’s a huge give away and gives the meta human “look”
big problem with Android phone is that they used to have ToF camera, but for some reason they have been getting rid of that (damn Samsung....) and you need ToF censor to do this
A bunch of the recently new possibilities with AI art piqued my interest, and I expect many opportunities with full 3D animations to be just around the corner. Prompt based images from stable diffusion are nice, but you may end up with different faces every time. and you need to fix wonky problems, like hands and other weirdness. With additions like controlnet, esp. openpose and masking you really get more control. With LoRa you can train your own models and may combine them. But it's all just not ideal. One of your videos, getting an image into blender, either with KeenTools FaceBuilder plugin (or DAZ FaceGen / Face Transfer), and then get it into metahumans unreal could be a huge step forward. But with how fast development is going, maybe there will be huge breakthroughs that will directly link a set of images (front/45 angle plus face mesh) to either a 3D asset or a pose or face for a 2D artwork directly. Maybe the tech will go so crazy as to confer add-ons like clothes and hairdoes, first from image to image (in different poses/angles) then to a 3D asset. Now you do loose some consistency, with different angles and higher resolution introducing small changes. Perhaps metahuman can produce output an AI model can be trained on and similar altered files could then be read back into metahuman? You never know, every week there seems to be something new!
I know a lot of dev is going behind the scenes using image to mesh…. It’s coming. It’s a lot to keep up with 🤣 But yes I also believe 3D AI will work together in harmony very well. It’s exciting to be in these early stages and seeing all the development almost in real-time. I agree prompt based images were okay… but we want to do something much bigger ✌️✨
@@promptmuse yes, I'm such a beginner, but I have a start now with sd with controlnet in a google collab now and it works. I haven't made an image I'm please with yet, but will try again this weekend 😃.
Been always a big fan of you, and i think you would be a great candidate to cast Cleopatra if you ever decide to make an unreal short based on your own look.
for a company that develops games 99% for windows you'd think they would release their apps on android as most windows users have android phones... oh well what a shame.
I’m sure they will develop something. Someone ( A kind open source dev developed) Developed a Live link android version for the current pipeline. So I can image in someone will do they for the release. 👍
Hello Prompt muse. You are very pretty but I was just curious why is one of your eyes different to the other one. Sorry if it is rude to ask ☹ I am just curious. Thanks.
Yeah I hear ya! Try the forum for latest announce. Also, on this page there is the latest documentation on the top horizontal tool Bar forums.unrealengine.com/categories?tag=metahuman
I cannot stand Apple. Will never buy another. There's almost double of android users out there over apple. These devs making apple users first or exclusively apple must had a boat load of money thrown at them. It makes no sense to make a app that's exclusive to a smaller market using unreal engine which anyone can use. The devs must be some serious fan boys or something. I enjoy android but I wouldn't exclude a huge market like that either.
This is what they have stated. Hopefully this isn’t the case, as they said it’s been made for that….which could translate as it maybe unstable on other devices but still work. That’s an assumption on my behalf.
Always wondered how is it that while animations are being captured from a real person they still look so weird/uncanny, now i got the answer, the person's face while doing said expressions looks extremely weird/uncanny
Visuals are good for games indeed, but AI is more important. I do not mind fighting stick figures as long as they do not act like DOOM\Diablo NPC's. Hope some chatGPT AI can be implemented for such task.
Thank you for all of your videos! You explain so efficiently. Learning here is a relief, compared to a lot of other tutorials are 4x longer and don't communicate nearly the same amount of info. Loving all of the Unreal, character, animation and AI content! Seriously, thanks.
Meta human is super annoying. CC is better I think but when I imported the characters from that software I kept getting deformed hands and legs in the trial version. Did that happen to anyone else? I really wanna buy it for its flexibility.
Thank you. There's one more thing I have to add to the list-buy an iPhone:) I assume that in a couple of months there will be various extensions in stable diffusion that will already be very good at generating video and improving masks.I assume that in a couple of months there will be various extensions in stable diffusion that will already be very good at generating video and improving masks. Good luck and success to all
Thanks for you described our feelings about Metahuman. I wish the next update becomes to make such a tecnology more useful and quick access. As if we have a cake but it is fixed at the other side of the table and we can pick up just a piece of choclatte on it each time
@@promptmuse What tool do you use? I am making only a low-poly stuff for an older game, but I am still curious. Good luck with your channel, you are doing great job ;)
I didn't even know you covered UE5 topics. I like Unreal, but I am writing a new, lightweight engine that's faster, better graphics and integrated with machine-learning at the hardware level 😅