ConvRefiner
A lightweight convolutional post-processing module trained after the fact with an adversarial loss while the diffusion backbone remains frozen, used to restore the high-frequency statistical characteristics of images generated by pixel diffusion.
Timeline (last 90 days)
ConvRefiner is a lightweight convolutional module that is trained post hoc with an adversarial loss while the diffusion backbone is fully frozen.
With only a small amount of additional training, ConvRefiner can restore the high-frequency statistical properties of generated images while keeping FID from degrading.
The project webpage has been made public, containing more implementation details and visualization results for researchers to further reproduce and compare.
All Facts (3)
ConvRefiner is a lightweight convolutional module that is trained post hoc with an adversarial loss while the diffusion backbone is fully frozen.
50%UnverifiedWith only a small amount of additional training, ConvRefiner can restore the high-frequency statistical properties of generated images while keeping FID from degrading.
50%UnverifiedThe project webpage has been made public, containing more implementation details and visualization results for researchers to further reproduce and compare.
50%