Please share what you generate with this, feedback is helpful so I can iterate on this LoRA!

With Lora (euler/beta/27/1.0 strength) vs Without

a close-up polaroid photo of a woman standing in a field wearing a sundress

Left With Lora (and Keyword) vs Without
a close-up polaroid photo of a woman standing in a field wearing a sundress, psxpolaroid

What Is It?

This LoRA is trained to stylize photos into a scanned Polaroid instant photo look. This brings in more film tones, stylized focus and the classic photo frame.

How To

All generations in the samples are used with this LoRA applied only, and a face detail node. Samples are generated using euler/beta @ 27 steps. You may notice at strengths near 1.0 and higher the model can cause some issues with hands, and distorted people in backgrounds. I've left these in the examples. I find the prompt "a polaroid photo" or "a scanned polaroid photo" helps to ensure the frame appears as expected.

Weights around 0.4-0.9 produces very natural results, strengths above can produce great results inconsistently. Use the trigger word to force the style.

Training Notes

Trained with 11 uncaptioned images at 768. No effort was made to train ???? concepts, although the dataset does lean heavily towards women.