This Model is a Merge of 4 LoRAs for photorealism on top of original Pony checkpoint:
https://civarchive.com/models/343602/retouchphoto-for-ponyv6
https://civarchive.com/models/264290?modelVersionId=363388
https://civarchive.com/models/392478/photorealism-lora
https://civarchive.com/models/376340/photo-pony
Clip-Skip = 1|2, Samplers = DPM++ 2M SDE Heun Karras, DPM++ 3M SDE Karras, DPM++ SDE, UniPC, (or any other you like);
Positive Prompt: very beautiful and aesthetic, raw photo, photography, realistic, natural skin texture, skin imperfections, score_9, score_8_up, score_7_up, score_6, score_5, score_4, dark theme, 1girl, _more tokens go here_, cinematic lighting, evening,
Negative Prompt: (disfigured fingers:1.5), old, wrinkles, ugly, bad, wrong, ((3D)), 2D, ((render, CG, CGI)), drawing, painting, sketch, vector, concept art, manga, comic, cartoon, anime, illustration, watercolor, cell shading,
experiment with positive and negative prompts (use my examples as a starting point) ...
P.S.1: Teeth don't always look natural. Use this LORA as a fix: https://civarchive.com/models/128371?modelVersionId=140535
P.S.2: For even more photographic look use this LORA: https://civarchive.com/models/299086?modelVersionId=335906
Description
initial release, if something goes wrong - turn on Tiled VAE or use SDXL VAE in settings of your favorite webui
FAQ
Comments (5)
Getting some nice results! Thank you!
with this model ALL depends on the prompt - positive AND! negative = so if you master them - the results can be really stunning!
and thanks a lot!
Nice Model as always! Thx :)
<3
The examples seen on the model page made me curious to try it myself. It's true that I work an A1111 and not in ComfyUI. But... in my case it didn't work that well. The results were a bit unexpected. Not only that with the same Seed, I was getting different results on every run, without even changing a single bit of the prompt, but at some point the results were so unpredictable even stylistically. There were times when images were purely 2D, and other times when slightly more realistic images appeared. But I absolutely never got a result that even came close to the idea of Photography.
Sorry. Maybe with another version.