import torch
from diffusers import FluxPipeline

pipe = FluxPipeline.from_pretrained("black-forest-labs/FLUX.1-schnell", torch_dtype=torch.bfloat16)
pipe.load_lora_weights("yasaone1919/flux-schnell-stylized-animation-lora", weight_name="animation-000004.safetensors")
pipe.to("cuda")

image = pipe(
    "animated-style a fox standing on a cliff overlooking the ocean",
    num_inference_steps=4,
    guidance_scale=0.0,
).images[0]
image.save("out.png")

Training details

Base model: FLUX.1-schnell (fp8)
Trainer: kohya-ss/sd-scripts via fluxgym
Dataset: 10 stylized animation frames, auto-captioned with Florence-2
Network dim: 4, LR 8e-4, adafactor, 512px, bf16 mixed precision
Trained on a single RTX 3070 Ti (8 GB) using 35-block CPU swapping

yasaone1919/flux-schnell-avatar-animation-lora

FLUX.1-schnell Stylized Animation LoRA

Trigger word

Files

Usage

Training details

yasaone1919/
flux-schnell-avatar-animation-lora