|
|
import gradio as gr |
|
|
import numpy as np |
|
|
import random |
|
|
from PIL import Image, ImageDraw, ImageFont |
|
|
import torch |
|
|
from diffusers import DiffusionPipeline |
|
|
import io |
|
|
|
|
|
|
|
|
device = "cuda" if torch.cuda.is_available() else "cpu" |
|
|
torch_dtype = torch.float16 if device == "cuda" else torch.float32 |
|
|
model_repo_id = "stabilityai/sdxl-turbo" |
|
|
pipe = DiffusionPipeline.from_pretrained( |
|
|
model_repo_id, |
|
|
torch_dtype=torch_dtype, |
|
|
variant="fp16" if device == "cuda" else None |
|
|
) |
|
|
pipe.to(device) |
|
|
|
|
|
MAX_SEED = np.iinfo(np.int32).max |
|
|
IMAGE_WIDTH = 768 |
|
|
IMAGE_HEIGHT = 768 |
|
|
WATERMARK_TEXT = "SelamGPT" |
|
|
|
|
|
|
|
|
def add_watermark(image): |
|
|
draw = ImageDraw.Draw(image) |
|
|
font_size = int(image.width * 0.03) |
|
|
try: |
|
|
font = ImageFont.truetype("Roboto-Bold.ttf", font_size) |
|
|
except: |
|
|
font = ImageFont.load_default() |
|
|
text_width = draw.textlength(WATERMARK_TEXT, font=font) |
|
|
x = image.width - text_width - 12 |
|
|
y = image.height - font_size - 10 |
|
|
draw.text((x+1, y+1), WATERMARK_TEXT, font=font, fill=(0, 0, 0, 128)) |
|
|
draw.text((x, y), WATERMARK_TEXT, font=font, fill=(255, 255, 255)) |
|
|
return image |
|
|
|
|
|
|
|
|
def generate( |
|
|
prompt, |
|
|
negative_prompt, |
|
|
seed, |
|
|
randomize_seed, |
|
|
guidance_scale, |
|
|
num_inference_steps, |
|
|
progress=gr.Progress(track_tqdm=True), |
|
|
): |
|
|
if not prompt.strip(): |
|
|
return None, "⚠️ Please enter a prompt" |
|
|
|
|
|
if randomize_seed: |
|
|
seed = random.randint(0, MAX_SEED) |
|
|
|
|
|
generator = torch.manual_seed(seed) |
|
|
result = pipe( |
|
|
prompt=prompt, |
|
|
negative_prompt=negative_prompt, |
|
|
width=IMAGE_WIDTH, |
|
|
height=IMAGE_HEIGHT, |
|
|
guidance_scale=guidance_scale, |
|
|
num_inference_steps=num_inference_steps, |
|
|
generator=generator, |
|
|
).images[0] |
|
|
|
|
|
watermarked = add_watermark(result) |
|
|
buffer = io.BytesIO() |
|
|
watermarked.convert("RGB").save(buffer, format="JPEG", quality=70) |
|
|
buffer.seek(0) |
|
|
return Image.open(buffer), seed |
|
|
|
|
|
|
|
|
examples = [ |
|
|
"Astronaut in a jungle, cold color palette, muted colors, detailed, 8k", |
|
|
"An astronaut riding a green horse", |
|
|
"A delicious ceviche cheesecake slice", |
|
|
] |
|
|
|
|
|
|
|
|
css = "#container { max-width: 700px; margin: auto; }" |
|
|
|
|
|
with gr.Blocks(css=css, title="SelamGPT Turbo Generator") as demo: |
|
|
with gr.Column(elem_id="container"): |
|
|
gr.Markdown("# 🖼️ SelamGPT Image Generator") |
|
|
|
|
|
with gr.Row(): |
|
|
prompt = gr.Textbox( |
|
|
label="Prompt", |
|
|
show_label=False, |
|
|
placeholder="Enter your prompt", |
|
|
lines=1, |
|
|
scale=3 |
|
|
) |
|
|
generate_btn = gr.Button("Generate", variant="primary") |
|
|
|
|
|
output_image = gr.Image(label="Generated Image", type="pil", format="jpeg") |
|
|
seed_display = gr.Textbox(label="Seed Used", interactive=False) |
|
|
|
|
|
with gr.Accordion("⚙️ Advanced Settings", open=False): |
|
|
negative_prompt = gr.Textbox(label="Negative Prompt", placeholder="What to avoid (optional)", max_lines=1) |
|
|
randomize_seed = gr.Checkbox(label="Randomize Seed", value=True) |
|
|
seed = gr.Slider(0, MAX_SEED, step=1, label="Seed", value=0) |
|
|
|
|
|
guidance_scale = gr.Slider(0.0, 10.0, step=0.1, label="Guidance Scale", value=0.0) |
|
|
num_inference_steps = gr.Slider(1, 10, step=1, label="Inference Steps", value=2) |
|
|
|
|
|
gr.Examples(examples=examples, inputs=[prompt]) |
|
|
|
|
|
generate_btn.click( |
|
|
fn=generate, |
|
|
inputs=[ |
|
|
prompt, |
|
|
negative_prompt, |
|
|
seed, |
|
|
randomize_seed, |
|
|
guidance_scale, |
|
|
num_inference_steps |
|
|
], |
|
|
outputs=[output_image, seed_display] |
|
|
) |
|
|
|
|
|
if __name__ == "__main__": |
|
|
demo.launch() |
