This is my attempt at bringing some of that ReV Animated look to SDXL. Trained on ~1900 images that were generated from ReV Animated using wildcards, thought it is quite biased towards women. Half of the images were generated to be darker with the help of the LowRA Lora.
Images were generate in the resolutions 512x768, 640x640, 768x512, and then upscaled by 1.65x to reach the target training resolution of 1024px for SDXL.
Description
FAQ
Comments (2)
Amazing model, thanks. ReV is my favourite model on 1.5, i'll use this to try and et a bit of that magic back.
Do you use this on sdxl base? Did you try with other checkpoints?
I have only tried this on sdxl-base so far.