generate.py

# === Standard Library ===
import requests
import os
from datetime import datetime
import random
import time
import base64

# === Third-Party Libraries ===
import torch
from PIL import Image, PngImagePlugin
from diffusers import StableDiffusionPipeline

# === Configuration ===
model_id = "runwayml/stable-diffusion-v1-5"
output_dir = "generated_images"
os.makedirs(output_dir, exist_ok=True)
ROTATIONS = 32

base_prompt = "antiwar"
negative_prompt = (
    "(nsfw:1.5), (easynegative:1.3) (bad_prompt:1.3) badhandv4 bad-hands-5 (negative_hand-neg) "
    "(bad-picture-chill-75v), (worst quality:1.3), (low quality:1.3), (bad quality:1.3), "
    "(a shadow on skin:1.3), (a shaded skin:1.3), (a dark skin:1.3), (blush:1.3), "
    "(signature, watermark, username, letter, copyright name, copyright, chinese text, artist name, name tag, "
    "company name, name tag, text, error:1.5), (bad anatomy:1.5), (low quality hand:1.5), (worst quality hand:1.5)"
)

generation_config = {
    "vae": "vae-ft-mse-840000",
    "sampler": "Euler a",
    "steps": 25,
    "guidance_scale": 7.0
}

GIST_LOG_FILE = "gist_log.md"

# === Initialize Model ===
device = "cuda" if torch.cuda.is_available() else "cpu"
print(f"Using device: {device}")
pipe = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16).to(device)

# === Functions ===

def add_metadata_and_save(image: Image.Image, filepath: str, prompt: str, negative_prompt: str, seed: int):
    """Embed generation metadata into a PNG and save it."""
    meta = PngImagePlugin.PngInfo()
    meta.add_text("Prompt", prompt)
    meta.add_text("NegativePrompt", negative_prompt)
    meta.add_text("Model", model_id)
    meta.add_text("VAE", generation_config["vae"])
    meta.add_text("Sampler", generation_config["sampler"])
    meta.add_text("Steps", str(generation_config["steps"]))
    meta.add_text("Seed", str(seed))
    meta.add_text("Date", datetime.now().strftime("%Y-%m-%d %H:%M:%S"))

    image.save(filepath, "PNG", pnginfo=meta)


def upload_to_gist(image_path, prompt, negative_prompt, seed, model_id):
    """
    Uploads an image and metadata to GitHub Gist using Base64 encoding.
    Returns Gist URL if successful.
    """
    # HF_SECRET INSERT HERE 
    USERNAME = "ajsbsd"

    headers = {
        "Authorization": f"token {GITHUB_TOKEN}",
        "Accept": "application/vnd.github+json"
    }

    try:
        with open(image_path, "rb") as img_file:
            image_bytes = img_file.read()
            image_data = base64.b64encode(image_bytes).decode("utf-8")
            print(f"✅ Image encoded. Length: {len(image_data)} characters")
    except Exception as e:
        print(f"❌ Failed to read image: {e}")
        return None

    # Build metadata
    metadata = (
        f"Prompt: {prompt}\n"
        f"Negative Prompt: {negative_prompt}\n"
        f"Seed: {seed}\n"
        f"Model: {model_id}\n"
        f"VAE: {generation_config['vae']}\n"
        f"Sampler: {generation_config['sampler']}\n"
        f"Steps: {generation_config['steps']}\n"
        f"Guidance Scale: {generation_config['guidance_scale']}\n"
        f"Date: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}"
    )

    print(f"README.md content preview: {f'![Generated Image](data:image/png;base64,{image_data})'[:200]}...")
    readme_content = f"![Generated Image](data:image/png;base64,{image_data})"

    print("README.md content length:", len(readme_content))  # Optional debug
    print("README.md sample:", readme_content[:200])  # Optional debug


    payload = {
        "description": "Stable Diffusion Generated Image",
        "public": True,
        "files": {
            os.path.basename(image_path): {
                "content": image_data,
                "encoding": "base64"
            },
            "metadata.txt": {
                "content": metadata
            },
            "README.md": {
              "content": readme_content
            }
        }
    }

    response = requests.post("https://api.github.com/gists",  headers=headers, json=payload)

    if response.status_code == 201:
        gist_url = response.json()["html_url"]
        print(f"✅ Uploaded to GitHub Gist: {gist_url}")
        return gist_url
    else:
        print(f"❌ Failed to create Gist: {response.status_code} - {response.text[:200]}")
        return None


def generate_and_process_images(num_images: int = 1):
    """Generate images with metadata and upload to GitHub Gist."""
    for i in range(num_images):
        variation = ", vibrant colors, neon lights" if i % 2 == 0 else ", soft pastel tones, morning light"
        prompt = base_prompt + variation
        seed = random.randint(10000000, 99999999)
        generator = torch.Generator(device=device).manual_seed(seed)

        print(f"Generating image {i + 1} with seed {seed}...")

        result = pipe(
            prompt=prompt,
            negative_prompt=negative_prompt,
            num_inference_steps=generation_config["steps"],
            guidance_scale=generation_config["guidance_scale"],
            generator=generator,
        )

        image = result.images[0]
        timestamp = datetime.now().strftime("%Y%m%d_%H%M%S_%f")
        filename = f"{output_dir}/image_{timestamp}_{i}.png"

        add_metadata_and_save(image, filename, prompt, negative_prompt, seed)
        print(f"Saved: {filename}")

        # Upload to GitHub Gist
        #gist_url = upload_to_gist(filename, prompt, negative_prompt, seed, model_id)
        #if gist_url:
        #    with open(GIST_LOG_FILE, "a") as f:
        #        f.write(f"- [{prompt}]({gist_url})\n")
        #    print(f"📌 Gist created: {gist_url}")


# === Execution ===
if __name__ == "__main__":
    generate_and_process_images(num_images=ROTATIONS)
    del pipe
    torch.cuda.empty_cache()