Gradio UI implementation Guide

[Astroburner]

1. pip install gradio
2. start via cmd python
3. Insert

import gradio as gr
import torch
from hyimage.diffusion.pipelines.hunyuanimage_pipeline import HunyuanImagePipeline

Modell load

model_name = "hunyuanimage-v2.1"
pipe = HunyuanImagePipeline.from_pretrained(model_name=model_name, torch_dtype='bf16')
pipe = pipe.to("cuda")

Function with Gradio

def generate(prompt, width=1024, height=1024, steps=20, guidance=3.5, seed=42):
image = pipe(
prompt=prompt,
width=width,
height=height,
num_inference_steps=steps,
guidance_scale=guidance,
shift=5,
seed=seed,
)
return image

Gradio UI

demo = gr.Interface(
fn=generate,
inputs=[
gr.Textbox(label="Prompt", lines=5, placeholder="Schreibe hier deinen Prompt..."),
gr.Slider(512, 2048, value=1024, step=64, label="Width"),
gr.Slider(512, 2048, value=1024, step=64, label="Height"),
gr.Slider(1, 50, value=20, step=1, label="Steps"),
gr.Slider(1.0, 10.0, value=3.5, step=0.1, label="Guidance Scale"),
gr.Number(value=42, label="Seed"),
],
outputs=gr.Image(type="pil", label="Generated Image"),
title="HunyuanImage v2.1 Generator",
description="Creat images with your prompt in HunyuanImage."
)

demo.launch(server_name="0.0.0.0", server_port=8188)

[Whjie-LLM]

I have downloaded all the model files to my local server. How can I use my local model

[Astroburner]

If you follow the installation guide in the github repository under https://github.com/Tencent-Hunyuan/HunyuanImage-2.1/blob/main/ckpts/checkpoints-download.md you dont need to do anything. I did the following and all worked.

1. git clone https://github.com/Tencent-Hunyuan/HunyuanImage-2.1.git
2. cd HunyuanImage-2.1
3. pip install -r requirements.txt
4. pip install flash-attn==2.7.3 --no-build-isolation
5. pip install -U "huggingface_hub[cli]"
6. pip install modelscope gradio
7. pip install --upgrade pip
8. hf download tencent/HunyuanImage-2.1 --local-dir ./ckpts
9. hf download Qwen/Qwen2.5-VL-7B-Instruct --local-dir ./ckpts/text_encoder/llm
10. hf download google/byt5-small --local-dir ./ckpts/text_encoder/byt5-small
11. modelscope download --model AI-ModelScope/Glyph-SDXL-v2 --local_dir ./ckpts/text_encoder/Glyph-SDXL-v2

After that just run it with gradio descriped above.

import gradio as gr
import torch
from hyimage.diffusion.pipelines.hunyuanimage_pipeline import HunyuanImagePipeline
model_name = "hunyuanimage-v2.1"
pipe = HunyuanImagePipeline.from_pretrained(model_name=model_name, torch_dtype='bf16')
pipe = pipe.to("cuda")
def generate(prompt, width=1024, height=1024, steps=20, guidance=3.5, seed=42):
image = pipe(
prompt=prompt,
width=width,
height=height,
num_inference_steps=steps,
guidance_scale=guidance,
shift=5,
seed=seed,
)
return image
demo = gr.Interface(
fn=generate,
inputs=[
gr.Textbox(label="Prompt", lines=5, placeholder="Schreibe hier deinen Prompt..."),
gr.Slider(512, 2048, value=1024, step=64, label="Width"),
gr.Slider(512, 2048, value=1024, step=64, label="Height"),
gr.Slider(1, 50, value=20, step=1, label="Steps"),
gr.Slider(1.0, 10.0, value=3.5, step=0.1, label="Guidance Scale"),
gr.Number(value=42, label="Seed"),
],
outputs=gr.Image(type="pil", label="Generated Image"),
title="HunyuanImage v2.1 Generator",
description="Creat images with your prompt in HunyuanImage."
)

demo.launch(server_name="0.0.0.0", server_port=8188)