Files
DiffSynth-Studio/examples/flux2/model_inference_low_vram/Template-KleinBase4B-Upscaler.py
Artiprocher f58ba5a784 update docs
2026-04-16 20:24:22 +08:00

67 lines
2.4 KiB
Python

from diffsynth.diffusion.template import TemplatePipeline
from diffsynth.pipelines.flux2_image import Flux2ImagePipeline, ModelConfig
import torch
from modelscope import dataset_snapshot_download
from PIL import Image
vram_config = {
"offload_dtype": "disk",
"offload_device": "disk",
"onload_dtype": torch.float8_e4m3fn,
"onload_device": "cpu",
"preparing_dtype": torch.float8_e4m3fn,
"preparing_device": "cuda",
"computation_dtype": torch.bfloat16,
"computation_device": "cuda",
}
pipe = Flux2ImagePipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="black-forest-labs/FLUX.2-klein-base-4B", origin_file_pattern="transformer/*.safetensors", **vram_config),
ModelConfig(model_id="black-forest-labs/FLUX.2-klein-4B", origin_file_pattern="text_encoder/*.safetensors", **vram_config),
ModelConfig(model_id="black-forest-labs/FLUX.2-klein-4B", origin_file_pattern="vae/diffusion_pytorch_model.safetensors"),
],
tokenizer_config=ModelConfig(model_id="black-forest-labs/FLUX.2-klein-4B", origin_file_pattern="tokenizer/"),
vram_limit=torch.cuda.mem_get_info("cuda")[1] / (1024 ** 3) - 0.5,
)
template = TemplatePipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[ModelConfig(model_id="DiffSynth-Studio/Template-KleinBase4B-Upscaler")],
lazy_loading=True,
)
dataset_snapshot_download(
"DiffSynth-Studio/examples_in_diffsynth",
allow_file_pattern=["templates/*"],
local_dir="data/examples",
)
image = template(
pipe,
prompt="A cat is sitting on a stone.",
seed=0, cfg_scale=4, num_inference_steps=50,
template_inputs = [{
"image": Image.open("data/examples/templates/image_lowres_512.jpg"),
"prompt": "A cat is sitting on a stone.",
}],
negative_template_inputs = [{
"image": Image.open("data/examples/templates/image_lowres_512.jpg"),
"prompt": "",
}],
)
image.save("image_Upscaler_1.png")
image = template(
pipe,
prompt="A cat is sitting on a stone.",
seed=0, cfg_scale=4, num_inference_steps=50,
template_inputs = [{
"image": Image.open("data/examples/templates/image_lowres_100.jpg"),
"prompt": "A cat is sitting on a stone.",
}],
negative_template_inputs = [{
"image": Image.open("data/examples/templates/image_lowres_100.jpg"),
"prompt": "",
}],
)
image.save("image_Upscaler_2.png")