Possible to Add A.I. StyleGAN(Generative Adversarial Network) to TVPaint? Updated 6-18-2020

[TheQuestionMark]

I know this is crazy to request something this big. But is possible to add AI to TVPaint? It might be beyond knowledge and understanding.

Here are the videos for StyleGAN
https://twitter.com/roadrunning01/statu ... 8034757633
https://twitter.com/_Ryobot/status/1096565388165300225

It seems great potential that it can probably do in-between maybe even more that can allow you to skip a few key frames.

https://www.dropbox.com/s/z0iqlkj4e0nxk ... e.mp4?dl=0 This one seems convincing that it can potential do in between. Even doing 60FPS Animation for Anime. It's still learning I think it's only 5 days?

It's seems open-source AI?
https://github.com/NVlabs/stylegan

There's another GAN too, to do anime backgrounds.
https://medium.com/syncedreview/reprodu ... 30d583736e

[TheQuestionMark]

There's an update on A.I.! They can do moving paintings! If they can they do that, they can do animation! I saw that they can do it with the Mona Lisa! It feels like doing traditionally feels finally dead. TVPaint please take this seriously it's not a joke! I think it's the future. TVPaint could be left in the dust if it's not researching or experimenting it. I think Disney could be adding it to Meander, and I think there's no way TVPaint can compete against big corporation like Disney in R&D. What is TVPaint going to do?



There's also a pose estimation A.I. For those anime crowd scenes.

[Jet]

[...] It feels like doing traditionally feels finally dead. [...] TVPaint could be left in the dust if it's not researching or experimenting it. I think Disney could be adding it to Meander, and I think there's no way TVPaint can compete against big corporation like Disney in R&D.

What problem do you think this solves exactly?

If it's for production cost/speed, TVPaint can speed up the tweening process by building vector layers that support shape tweening like in Adobe Animate (Flash), job done. There's one guy on GitHub adding this to OpenToonz, so we'll see what happens. This is what was used to produce Long Way North and someone could correct me if I remember this wrong, but the producer said it allowed them to keep everything in house instead of outsourcing.

Dofus Livre 1: Julith was produced using the same technique (TVPaint was used for animating, Animate/Flash used for cleanup, tweening and color).



Using shape tweening for something like the image below, is much less expensive to do and also not possible with this AI unicorn stuff, you'll never get this level of nuance from machines, at least not in my lifetime.

[TheQuestionMark]

There's new update! TVPaint have you been trying to program with A.I.? This one seems compelling.

Also this when doing backgrounds. Some of it pretty compelling doing backgrounds the "interpetation". But not sure how it looks doing 8K.
https://github.com/FilipAndersson245/cartoon-gan

[TheQuestionMark]

[...] It feels like doing traditionally feels finally dead. [...] TVPaint could be left in the dust if it's not researching or experimenting it. I think Disney could be adding it to Meander, and I think there's no way TVPaint can compete against big corporation like Disney in R&D.

What problem do you think this solves exactly?

If it's for production cost/speed, TVPaint can speed up the tweening process by building vector layers that support shape tweening like in Adobe Animate (Flash), job done. There's one guy on GitHub adding this to OpenToonz, so we'll see what happens. This is what was used to produce Long Way North and someone could correct me if I remember this wrong, but the producer said it allowed them to keep everything in house instead of outsourcing.

Using shape tweening for something like the image below, is much less expensive to do and also not possible with this AI unicorn stuff, you'll never get this level of nuance from machines, at least not in my lifetime.

[TheQuestionMark]

There's new update! TVPaint have you been trying to program with A.I.? This one seems compelling.

Also this when doing backgrounds. Some of it pretty compelling doing backgrounds the "interpetation". But not sure how it looks doing 8K.
https://github.com/FilipAndersson245/cartoon-gan

Opps I didn't post right sorry.