The aim of this workflow is simply to upscale a video generated with WAN2.1.
Resources you need:
📂Files :
ANY upscale model (depreciated):
Realistic : RealESRGAN_x4plus.pth
Anime : RealESRGAN_x4plus_anime_6B.pth
in models/upscale_models
📦Custom Nodes :
Description
Base version
FAQ
Comments (8)
Thank you for this, it will help me upscale some of my older videos.
Great workflow, simple and to the point.
Could you make a simple frame interpolation workflow, pretty please?
@UmeAiRTÂ The two hour wait was worth it.
I have a question regarding upscaling. I'm trying to upscale a 960x1440 image by a factor of 1.5x. Even though I have 64GB of RAM, my entire system often freezes during the process, to the point where the mouse cursor won't move.
Could there be an issue with my settings? Or is this something where installing something like xformers might help? (I'm currently using sage attention).
Alternatively, are there general guidelines or references for the maximum resolution or upscale factor one can typically handle as a reference point? Thank you.
I'm looking for a more optimized method, but I still have a lot of work to do. You should know that in reality, you do a native upscale of the model, often x4, then you reduce it to the format you chose.
@UmeAiRTÂ To clarify, are you suggesting that I should first use a workflow like the 'IMG to VIDEO simple workflow WAN2.1 | GGUF | LoRA | UPSCALE | TeaCache' to upscale the image by 4x, and then downscale the result to my desired 1.5x final size (instead of directly upscaling by 1.5x with my current method)? English isn't my first language, so I want to make sure I understand correctly. Apologies if there's any misunderstanding.
@zens No, I just described the process hidden in this workflow