TurboRender-flux-dev generates better results compared to the original flux dev baseline within the 4-6 step range. Additionally, when using more steps (we suggest 10-15 steps), it delivers even finer details, offering improved performance compared with LCM LoRA.

The LoRA was trained in just 8 hours with 1x A100 GPU.

Update: A more powerful few-step LoRA is comming soon!

Usage example

python

from diffusers import DiffusionPipeline

pipe = DiffusionPipeline.from_pretrained("black-forest-labs/FLUX.1-dev")
pipe.load_lora_weights("DarkMoonDragon/TurboRender-flux-dev")
image = pipe(
    prompt="your prompt",
    guidance_scale=3.5,
    height=1024,
    width=1024,
    num_inference_steps=8,
	generator=torch.Generator(device='cuda').manual_seed(0)
).images[0]

Trigger words

No trigger word, plug and play, accelerates generation process & adds more details!!!

Download model

Weights for this model are available in Safetensors format.

Download them in the Files & versions tab.