Suggestions to the next-generation generic restoration models are welcome!

[AIisCool]

Thank you for all the work on these models! I am wondering if there's any chance a new/improved model is being worked on?

[IceClear]

Thank you for the enthusiasm of SeedVR series.

I may begin my work on the next-generation restoration approach in the next few months, and I would like to hear from the community about the expectations of the next open-source super-powerful generic restoration model : )
Feel free to add comments below to provide any suggestions and feedback.
They are really valuable to me >_<

BTW, it is always not easy to open-source large vision models from the company, but I promise to try my best to keep contributing to the community and I pretty appreciate the love of this project from the community!

[FurkanGozukara]

SeedVR is purely amazing

I will publish a full Gradio app hopefully upon it

but it will be only due to this repo because you have to consider with block swapping + tiled VAE encode and decode

without tiled VAE encode and decode it is a VRAM monster and not useable

https://github.com/numz/ComfyUI-SeedVR2_VideoUpscaler