search

chaiNNer-org / spandrel

Spandrel gives your project support for various PyTorch architectures meant for AI Super-Resolution, restoration, and inpainting. Based on the model support implemented in chaiNNer.
MIT License
102 stars 7 forks source link

Add support for SAFMN++ #229

Open RunDevelopment opened 2 months ago

RunDevelopment commented 2 months ago

SAFMN recently got updated to include new models. Add support for them.

Originally reported by @zelenooki87 here:

By the way, I noticed that the safmn GitHub repository has been updated with new models. Could you update the safmn architecture in Spandrel to support them? Thanks!

zelenooki87 commented 2 months ago

I tried SAFMN_BCIE in conda environment on real life photos, and while it works well on some, it unfortunately adds artifacts to some other images. But then I tried it on a couple of video clips with extracted frames it gives constantly really nice results for video decompression :) onother model had not tested yet. I'll wait for spandrel support before testing them in more detail.

zelenooki87 commented 1 month ago

[2024-05-06] The lightweight version of SAFMN++ won 1st place in the fidelity track of Real-Time 4K Super-Resolution of Compressed AVIF Images Challenge and was invited to give an oral presentation at the AIS2024 workshop.