What Is It?
This LoRA is trained to emulate amateur photography. Part of a set of LoRAs I trained to try to minimize the issue of similar beautiful faces with Flux. This brings the photography styles into more realistic common images, as well as faces, body types and poses. This is a very lightweight LoRA that can do some heavy lifting. Please share your results.
How To
All images in the showcase are generated using just this LoRA and a face detail node.
Generated using euler/beta @ 27 steps, CFG 2.5. I do not use this with a trigger word, instead just pushing the strength until I get my desired effect. Generally I start around .85.
on the left w/lora on the right without:
a woman standing in a mall wearing a bodycon dress, volumetric lighting
Description
Same Dataset, New Model
Trained on amateur photographs for more generic 1girls that looks like snapshots. The showcase images are captured as follows:
LoRA Strength: 1.25 mostly
Euler/Beta @ 9 steps
ModelSamplingAuraFlow Shift: 5.90
with a FaceDetailer Node:
Euler/Beta @ 9 steps
0.26 Denoise