Go easy; it's my first LoRA.
I've been experimenting with vintage film stocks for a while, and wanted to take a shot at a Polaroid style. As I was playing with the LoRA, I realized that it "unlocked" a few surprises, such as some interesting text handling.
Recommended Settings
This LoRA is trained on RobMix Zenith, so using it is recommended. Sample images are generated with Euler at 40 steps and a CFG of 7.0. Recommended LoRA strength is 0.5. If using ComfyUI, I apply the LoRA to the model only.
Prompting
Responds well to simple and straightforward prompts, such as the one below.
A polaroid photograph of [subject] with the text "[text]"Text is an "icing on the cake" feature here and won't be perfect. For best results, keep it brief. Some subject/text combos will work better than others.
Enjoy!
Description
This LoRA is trained on RobMix Zenith, so using it is recommended. Sample images are generated with Euler at 40 steps and a CFG of 7.0. Recommended LoRA strength is 0.5. If using ComfyUI, I apply the LoRA to the model only.
FAQ
Comments (2)
This should deserve way more views and downloads. Super cool stuff and impressed.
Turned out well, good one, and thanks! imo, bake it against a base model instead of a fine tune for v2, it works against rob mix but it looks deep fried using other fine tunes