A24-Springbreakers-edition

Details

Model description

This is a variation of my A24 film style trained only on frames from Springbreakers, 20 frames were hand selected that featured interesting composition and good lighting. Captioning was done with CogVLM and captions were edited to be more concise / less verbose. Training was done for 40 epochs with prodigy optimizer with a batch size of 4 and a gradient accumulation of 2. After testing the epochs i chose number 36 to share based on it's generation of colorful light compared to other epochs.

Images made by this model

No Images Found.