nocomment

9mo

Check out comfy ui.

Havent seen the movie but going off what you said i imagine they use facial detection of some sort

Is there a clip i can watch to see?

awawh

9mo

unfortunately it’s not really shown in the trailers,
https://m.youtube.com/watch?v=CpLTGFgKWE0&pp=ygULYWdncm8gZHIxZnQ%3D
the best way I can explain it is that if you see the gear-like/mechanical patterning under some of the characters skin or faces e.g. on the woman’s body’s around 1:30- those patterns were moving and fractalizing actively, just coming in and out of clarity, self multiplying, moving around

Will Defs check our comfy ui tysm for the rec

nocomment

9mo

Ahh ok it reminds me of google deepdream. Im not sure what thats called specifically but im sure theres a term for it.

awawh

9mo

fireee
another friend of mine told me to check out runway ML so adding that to the list too

neon

9mo

Cloud Services:

RunwayML Gen-2
PikaLabs
OpenAI Sora (closed Beta, hard to access long waitlist)
Domo AI

Local/Open Source Services:

Stable Diffusion Video
Deforum
EasyAnimate
T2V-Turbo
VideoCrafter2

Local stuff usually requires good hardware, 10GB+ RTX20+ GPU, Deforum only needs a 6GB GPU but the other ones need a lot more VRAM from my testing. You can always use a Cloud GPU with these if you don't have the hardware but you got to pay. You can find free trials though. Some of these models run in ComfyUI or Auto1111. You can watch YouTube videos to see the workflow of both, they are both Stable Diffusion installs just different UI layouts. Some of these have standalone UI also.

If I had to guess what Harmony used by looking at the clips online, still haven't seen the movie. It looks like EasyAnimate or Deforum, using Video2Video, maybe some Image2Video. It looks like he's using semi-transparent alpha channels, You'd need something like After Effects to do this. You'd have the video on top and then have a alpha channel with the AI movement giving it this trippy look were only part of the scene is animated/moving. You'd need to play with the opacity and diffusion noise you're injecting but doesn't seem too hard, Honestly a pretty cool idea.

neon

9mo

Cloud Services:

RunwayML Gen-2
PikaLabs
OpenAI Sora (closed Beta, hard to access long waitlist)
Domo AI

Local/Open Source Services:

Stable Diffusion Video
Deforum
EasyAnimate
T2V-Turbo
VideoCrafter2

Local stuff usually requires good hardware, 10GB+ RTX20+ GPU, Deforum only needs a 6GB GPU but the other ones need a lot more VRAM from my testing. You can always use a Cloud GPU with these if you don't have the hardware but you got to pay. You can find free trials though. Some of these models run in ComfyUI or Auto1111. You can watch YouTube videos to see the workflow of both, they are both Stable Diffusion installs just different UI layouts. Some of these have standalone UI also.

If I had to guess what Harmony used by looking at the clips online, still haven't seen the movie. It looks like EasyAnimate or Deforum, using Video2Video, maybe some Image2Video. It looks like he's using semi-transparent alpha channels, You'd need something like After Effects to do this. You'd have the video on top and then have a alpha channel with the AI movement giving it this trippy look were only part of the scene is animated/moving. You'd need to play with the opacity and diffusion noise you're injecting but doesn't seem too hard, Honestly a pretty cool idea.

awawh

9mo

and def check out the film; see it in theaters if you can. Intense sensory experience, the use of tech in the film in general is incredible