simple-image-gen/app.py

import gradio as gr
import requests
import base64
from PIL import Image
from io import BytesIO
import numpy as np
import random
import os

API_URL = 'https://hub.societyai.com/models/flux-1-schnell/infer'
API_TOKEN = os.environ.get("SAI_API_TOKEN", "")
MAX_SEED = np.iinfo(np.int32).max
MAX_IMAGE_SIZE = 2048

with gr.Blocks() as demo:
    gr.Markdown("## FLUX.1-schnell Image Generation")

    with gr.Row():
        prompt = gr.Textbox(label="Prompt", lines=2, placeholder="Enter your prompt here")
    with gr.Accordion("Advanced Settings", open=False):
        with gr.Row():
            width = gr.Number(value=512, label="Width", maximum=1920)
            height = gr.Number(value=512, label="Height", maximum=1080)

        with gr.Row():
            num_steps = gr.Number(value=4, label="Number of Steps (1-4)", minimum=1, maximum=4)
            guidance_scale = gr.Slider(0.0, 10.0, 0.0, value=7.5, label="Guidance Scale")

        with gr.Row():
            seed = gr.Slider(
                label="Seed",
                minimum=0,
                maximum=MAX_SEED,
                step=1,
                value=0,
            )
            randomize_seed = gr.Checkbox(label="Randomize seed", value=True)

    generate_button = gr.Button("Generate Image")
    output_image = gr.Image(type="numpy", label="Generated Image")
    message = gr.Textbox(label="Status", interactive=False)

    def generate_image(prompt, width, height, num_steps, guidance_scale, seed, randomize_seed):
        try:
            # Validation: Ensure width and height are divisible by 16
            if width % 8 != 0 or height % 8 != 0:
                return None, "Error: Both width and height must be divisible by 8."
            if randomize_seed:
                seed = random.randint(0, MAX_SEED)
            # Prepare the data payload
            inputs = [
                {
                    "name": "PROMPT",
                    "shape": [1],
                    "datatype": "BYTES",
                    "data": [prompt]
                },
                {
                    "name": "INIT_IMAGE",
                    "shape": [1],
                    "datatype": "BYTES",
                    "data": [""] # not supported
                },
                {
                    "name": "WIDTH",
                    "shape": [1],
                    "datatype": "INT32",
                    "data": [int(width)]
                },
                {
                    "name": "HEIGHT",
                    "shape": [1],
                    "datatype": "INT32",
                    "data": [int(height)]
                },
                {
                    "name": "NUM_STEPS",
                    "shape": [1],
                    "datatype": "INT32",
                    "data": [int(num_steps)]
                },
                {
                    "name": "GUIDANCE_SCALE",
                    "shape": [1],
                    "datatype": "FP32",
                    "data": [float(guidance_scale)]
                },
                {
                    "name": "SEED",
                    "shape": [1],
                    "datatype": "INT32",
                    "data": [int(seed)]
                },
                {
                    "name": "IMAGE_STRENGTH",
                    "shape": [1],
                    "datatype": "FP32",
                    "data": [0.0] # not supported
                }
            ]

            payload = {
                "inputs": inputs,
                "outputs": [
                    {
                        "name": "IMAGE"
                    }
                ]
            }

            headers = {
                "Content-Type": "application/json",
                "Authorization": f"Bearer {API_TOKEN}"
            }

            # Send the POST request
            response = requests.post(API_URL, headers=headers, json=payload)

            if response.status_code == 200:
                # Parse the response
                result = response.json()
                image_base64 = result['outputs'][0]['data'][0]
                # Decode the base64 image data
                image_data = base64.b64decode(image_base64)
                # Convert to numpy array
                image = Image.open(BytesIO(image_data))
                image_np = np.array(image)
                return image_np, "Image generated successfully.", seed
            else:
                # Handle error
                return None, f"Error: {response.status_code} - {response.text}", seed
        except Exception as e:
            return None, f"Error: {str(e)}", seed

    generate_button.click(
        generate_image,
        inputs=[prompt, width, height, num_steps, guidance_scale, seed, randomize_seed],
        outputs=[output_image, message, seed]
    )

if __name__ == "__main__":
    demo.launch()