stylegan2-pytorch VS first-order-model

Generative networks like StyleGAN

It's interesting that AI generated pictures of people can get most things right but seemingly always struggles with teeth & fingers. If you check out thispersondoesnotexist.com the teeth are almost always goofy looking...

Teeth as well, check out thispersondoesnotexist.com you'll come across a lot of convincing looking people with odd teeth

https://thispersondoesnotexist.com/ has been around for 4 years now...

I remember there was https://thispersondoesnotexist.com/ Did it also work on StableDiffusion beneath?

Another great place to see AI generated faces: https://thispersondoesnotexist.com/

Check out https://thispersondoesnotexist.com/

Similarly check out This Person Does not Exist . com

There is this website called This Person Does Not Exist that shows you a random AI generated photo each time you refresh.

I think the novel and interesting tech is still happening, its just that without the colorful ads for it on TV, and without the software being packaged up and sold with pretty box art that you can physically hold, it doesn't feel as much like a capital-E Experience. It's probably the Internet's fault that we don't do things like that anymore, but the upside is that we now have access to so many ideas and applications from all over, even ones that aren't commercially viable.

Some that look exciting to me are: an AI that lets you animate still photos realistically [1], a simple website that guides you to discover new parks, eateries, and other places near you [2], an AI that colorizes old black-and-white photos/video [3], a Street View style map of the game world from "The Legend of Zelda: Breath of the Wild", with some 1st person 360 degree photos [4], and a tiny game engine that lets you distribute your whole game physically via printed QR codes [5].

If marketing and graphic design people ever felt like getting together to do some 'side projects', I vote that they should make print ads for apps/websites that they like :)

[1] https://github.com/AliaksandrSiarohin/first-order-model

[2] https://randomlocation.xyz (https://randomlocation.xyz/help.txt for customization)

[3] https://github.com/jantic/DeOldify

[4] https://nassimsoftware.github.io/zeldabotwstreetview/

[5] https://github.com/kesiev/rewtro

First Order Model

First Order Motion Model (Easy): You can give it an input image and a source video, and the movements from the video will be copied to the image. It has some janky-ness and I would recommend using an ai upscaller afterwards because the resolution isn't that high https://github.com/AliaksandrSiarohin/first-order-model

This project was originally released 2 years ago, it allows for fast motion transfer from a video to a still image

First of the model we are going to be making use of is a Deepfake model by the name of First Order Motion. Deepfakes you allow you create an artificial version of a person saying or doing an action, I first found about this particular model on two minute papers (an awesome YT channel for lovers of AI ⚡) and wanted to try it for myself. The video below talks more about the model.

NB: The best thing to do would be to edit the dockerfile and requirements.txt(can be seen above) then add the app.py to a forked version of this repo so the container image can be built successfully

You can use this repo with pytorch https://github.com/AliaksandrSiarohin/first-order-model I tried it locally with the voxceleb dataset which seems to be good with faces and got some fun results though it's restricted to 256 x 256 driving videos and source images