search

invoke-ai / InvokeAI

InvokeAI is a leading creative engine for Stable Diffusion models, empowering professionals, artists, and enthusiasts to generate and create visual media using the latest AI-driven technologies. The solution offers an industry leading WebUI, supports terminal use through a CLI, and serves as the foundation for multiple commercial products.
https://invoke-ai.github.io/InvokeAI/
Apache License 2.0
22.78k stars 2.35k forks source link

Why using RealESRGAN_x4plus in v0.1.0 instead of v0.2.5 #399

Closed glonlas closed 1 year ago

glonlas commented 2 years ago

Is your feature request related to a problem? Please describe. The gfpgan_tools.py file and preload_models.py scripts are loading the RealESRGAN x4 v0.1.0 model, while a more recent model is available: x4v3 v0.2.5 model.

Is there a rationale to keep using the v0.1.0?

Describe the solution you'd like

  1. Updating the line in preload_models.py with model_path='https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.5.0/realesr-general-x4v3.pth',
  2. Updating the line in gfpgan_tools.py with 4: 'https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.5.0/realesr-general-x4v3.pth',

Describe alternatives you've considered N/A

Additional context N/A

glonlas commented 2 years ago

Update. Changing the link to the model file will not work. It will need more deep dive. But still, I am interested to know the benefits of the x4 v0.1.0 version.

blessedcoolant commented 2 years ago

At the time of implementation, that was the best model available. So I added that. He released a new model but I cannot find the model architecture code for it. So I cannot implement that just yet.