import torch
from diffusers import DiffusionPipeline
# switch to "mps" for apple devices
pipe = DiffusionPipeline.from_pretrained("jasperai/flash-sd", dtype=torch.bfloat16, device_map="cuda")
pipe.load_lora_weights("shehabelkadi/Flash-SD-Lora")
prompt = "Japanese Ink Drawing, sexy woman in a city street, wearing (eyeglasses:1.4), white loose dress, red belt and shoes, chubby., ink drawing, inkwash, Japanese Ink Drawing"
image = pipe(prompt).images[0]Flash-SD
- Prompt
- Japanese Ink Drawing, sexy woman in a city street, wearing (eyeglasses:1.4), white loose dress, red belt and shoes, chubby., ink drawing, inkwash, Japanese Ink Drawing
- Negative Prompt
- ugly, deformed, noisy, blurry, low contrast, colorful, vibrant, text, watermark
Model description
This is the LORA file to use with ComfyUI or any other Stable Diffusion UI.
Use LCM sampler (I would advise to use 4 steps for initial generation and 2-4 steps for HiRez Fix. I found it effective to use Automatic CFG node with CFG of the KSampler at 1)
Trigger words
You should use SD1.5 Model.
I used my own trained SD 1.5 model tuned and trained on multible styles.
Download model
Weights for this model are available in Safetensors format.
Download them in the Files & versions tab.
- Downloads last month
- 6