Download
1 variant available
fp16 SafeTensor
Train4-step00122500.safetensors
Half precision, best balance • 6.46 GB
Verified: 3 years ago
4640 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
(77)
Sep 21, 2023
SDXL 1.0
Better generalization
-There may be some minor issues, such as elf ears

460 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
4250 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
This could be the most advanced WLOP model at present
(Trigger word)Suggested prompt:wgz style,portrait of a beautiful girl highly detailed
Do not use SD 1.5 prompts , Just enter a few words you want
You can let GPT organize natural language prompt , the effect is good
Suggest to use the LMS Karras sampler ,DPM++ 3M SDE Exponential is good too.
Training is focus on face.
performance may better under certain prompt words, such as Jewelry,necklace,freckles
(Sometimes the output are too bright-colored, try to send the picture to img2img and adjust the CFG to 4-5, which can achieve lower saturation and softer strokes.
and try to use more complete sentences to improve performance ,which is important for SDXL)
