This LoRA was trained on ~100 vintage photos shot on Kodachrome film and attempts to capture some of it's characteristic color capabilities with a rich saturation, especially in reds and a very fine grain. It also imbues a vintage look to people and buildings and can be pushed or pulled from this look with prompting.
Description
Trained on ~100 image caption pairs with a batch size of 4 using the prodigy optimizer for 2000 steps with a slightly modified version of the ai-toolkit repo.
FAQ
Comments (8)
How can it be pushed and pulled? Even God doesn't get it.
never answers my prayers, asks ME for help...
@blairesilver13 You're not ready to hear the divine truth yet but I'm patiently waiting for you to blossom. Meanwhile tell me how to push and pull your lora and I might grant you eternal life.
@GodAlMighty just prompt it to be more vintage or more modern, thats all
@blairesilver13 God having been around when people used to underexpose film and overdelop, he thought you had trained the model in this manner.
@GodAlMighty i used to work in a film lab so yeah, i get it, you could add that kind of stuff to the prompt as well, you could also try adding Cross Processed for some interesting color shifts.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.