I present a more realistic and human model, made with a mix of hybrid models and adjustments with trained Loras, which makes a modern and advanced SDXL Lightning in all aspects, it has a strong similarity with the Flux Dev1 model, this model has a lot of versatility in generating realistic and photographic images.
It is strictly prohibited to distribute, inside and outside civit ai, this model or respond anywhere or hosting services, in third-party image generation services, you can use it for personal use, if you know who someone distributed it without my consent , report it immediately.
To use the lora model you need to insert the token: ulthigrellper e class: woman
Recommended strength:
<lora:ultrahigh realism xl:0.6>
<lora:ultrahigh realism xl:0.7>
<lora:ultrahigh realism xl:0.8>
<lora:ultrahigh realism xl:0.9>
I recommend using the Adetailer extension.
Recommended settings:
Standard resolutions 512 x 640, 640x 768x960, 896 x1152
Resolution: 540x540 mark the Hires.fix, result 1080 x 1080 pixel
Resolution: 540x960 mark the Hires.fix, result 1080 x 1920 pixel
Use these recommended settings for generation:
Sampling method: DPM++3M SDE
Sampling steps: 18-35
CFG Scale: 4,5-8
Skip clip: 1
Sampling method: DPM++2M
Sampling steps: 18-35
CFG Scale: 4,5-8
Skip clip: 1
Sampling method: DPM++2M SDE
Sampling steps: 18-35
CFG Scale: 4,5-8
Skip clip: 1
Use the negative prompt:
Stable_Yogis_PDXL_Negatives-neg,monochrome, bad fingers, deformed fingers, asian, japanese, chinese, text, clothing, cloth, deformed neck, bad neck, long neck