Update app.py

app.py

@@ -1,154 +1,178 @@
+import os
+import io
+import zipfile
+from datetime import datetime
+
 import gradio as gr
-import numpy as np
-import random
+from PIL import Image
 
-# import spaces #[uncomment to use ZeroGPU]
-from diffusers import DiffusionPipeline
 import torch
-
-device = "cuda" if torch.cuda.is_available() else "cpu"
-model_repo_id = "stabilityai/sdxl-turbo"  # Replace to the model you would like to use
-
-if torch.cuda.is_available():
-    torch_dtype = torch.float16
-else:
-    torch_dtype = torch.float32
-
-pipe = DiffusionPipeline.from_pretrained(model_repo_id, torch_dtype=torch_dtype)
-pipe = pipe.to(device)
-
-MAX_SEED = np.iinfo(np.int32).max
-MAX_IMAGE_SIZE = 1024
-
-
-# @spaces.GPU #[uncomment to use ZeroGPU]
-def infer(
-    prompt,
+from diffusers import AutoPipelineForText2Image, DPMSolverMultistepScheduler
+
+# --------- Helper: load model ----------
+@torch.inference_mode()
+def load_pipeline(model_id: str, torch_dtype=torch.float16, device=None):
+    pipe = AutoPipelineForText2Image.from_pretrained(
+        model_id,
+        torch_dtype=torch_dtype,
+        use_safetensors=True
+    )
+    pipe.scheduler = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config)
+    if device is None:
+        device = "cuda" if torch.cuda.is_available() else "cpu"
+    pipe = pipe.to(device)
+    # small memory tweaks
+    if device == "cuda":
+        pipe.enable_attention_slicing()
+        pipe.enable_vae_slicing()
+    return pipe, device
+
+# Cache of loaded models so switching is fast
+_PIPELINES = {}
+
+def get_pipe(model_id: str):
+    if model_id not in _PIPELINES:
+        _PIPELINES[model_id], _ = load_pipeline(model_id)
+    return _PIPELINES[model_id]
+
+# --------- Core generation ----------
+def parse_prompts(text: str):
+    # Split by comma, strip whitespace, drop empties
+    parts = [p.strip() for p in text.split(",")]
+    return [p for p in parts if p]
+
+def generate_images(
+    prompts_text,
     negative_prompt,
-    seed,
-    randomize_seed,
+    model_id,
     width,
     height,
     guidance_scale,
     num_inference_steps,
-    progress=gr.Progress(track_tqdm=True),
+    batch_per_prompt,
+    seed
 ):
-    if randomize_seed:
-        seed = random.randint(0, MAX_SEED)
-
-    generator = torch.Generator().manual_seed(seed)
-
-    image = pipe(
-        prompt=prompt,
-        negative_prompt=negative_prompt,
-        guidance_scale=guidance_scale,
-        num_inference_steps=num_inference_steps,
-        width=width,
-        height=height,
-        generator=generator,
-    ).images[0]
-
-    return image, 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 = """
-#col-container {
-    margin: 0 auto;
-    max-width: 640px;
-}
+    prompts = parse_prompts(prompts_text)
+    if not prompts:
+        return [], None, "Please enter at least one prompt (use commas to separate)."
+
+    pipe = get_pipe(model_id)
+
+    # Seeding
+    if seed is None or str(seed).strip() == "":
+        generator = torch.Generator(device=pipe.device).manual_seed(torch.seed())
+    else:
+        try:
+            seed_val = int(seed)
+        except:
+            seed_val = torch.seed()
+        generator = torch.Generator(device=pipe.device).manual_seed(seed_val)
+
+    all_images = []
+    names = []
+    for i, p in enumerate(prompts, start=1):
+        images = pipe(
+            prompt=p,
+            negative_prompt=negative_prompt if negative_prompt else None,
+            width=width,
+            height=height,
+            guidance_scale=guidance_scale,
+            num_inference_steps=num_inference_steps,
+            num_images_per_prompt=batch_per_prompt,
+            generator=generator
+        ).images
+
+        # Collect and name
+        for j, img in enumerate(images, start=1):
+            all_images.append(img)
+            safe_prompt = "".join(c for c in p[:40] if c.isalnum() or c in "-_ ").strip().replace(" ", "_")
+            if not safe_prompt:
+                safe_prompt = f"prompt_{i}"
+            names.append(f"{safe_prompt}_{j}.png")
+
+    # Build ZIP in-memory
+    buf = io.BytesIO()
+    with zipfile.ZipFile(buf, "w", zipfile.ZIP_DEFLATED) as zf:
+        for img, name in zip(all_images, names):
+            bio = io.BytesIO()
+            img.save(bio, format="PNG")
+            bio.seek(0)
+            zf.writestr(name, bio.read())
+    buf.seek(0)
+
+    zip_name = f"images_{datetime.now().strftime('%Y%m%d_%H%M%S')}.zip"
+    return all_images, (zip_name, buf), f"Generated {len(all_images)} image(s) from {len(prompts)} prompt(s)."
+
+# --------- UI ----------
+CSS = """
+.gradio-container {max-width: 1100px !important}
 """
 
-with gr.Blocks(css=css) as demo:
-    with gr.Column(elem_id="col-container"):
-        gr.Markdown(" # Text-to-Image Gradio Template")
-
-        with gr.Row():
-            prompt = gr.Text(
-                label="Prompt",
-                show_label=False,
-                max_lines=1,
-                placeholder="Enter your prompt",
-                container=False,
-            )
-
-            run_button = gr.Button("Run", scale=0, variant="primary")
+with gr.Blocks(css=CSS, theme=gr.themes.Soft()) as demo:
+    gr.Markdown("# 🖼️ Multi-Prompt Text-to-Image (Hugging Face Space)")
+    gr.Markdown("Enter **comma-separated prompts** to generate multiple images at once. Choose size, batch count, and download all results as a ZIP.")
 
-        result = gr.Image(label="Result", show_label=False)
-
-        with gr.Accordion("Advanced Settings", open=False):
-            negative_prompt = gr.Text(
-                label="Negative prompt",
-                max_lines=1,
-                placeholder="Enter a negative prompt",
-                visible=False,
+    with gr.Row():
+        with gr.Column():
+            prompts_text = gr.Textbox(
+                label="Prompts (comma-separated)",
+                placeholder="A futuristic city at sunset, A cozy cabin in the woods, A portrait of a cyberpunk samurai",
+                lines=4
             )
-
-            seed = gr.Slider(
-                label="Seed",
-                minimum=0,
-                maximum=MAX_SEED,
-                step=1,
-                value=0,
+            negative_prompt = gr.Textbox(
+                label="Negative prompt (optional)",
+                placeholder="blurry, low quality, distorted"
+            )
+            model_id = gr.Dropdown(
+                label="Model",
+                value="stabilityai/sdxl-turbo",
+                choices=[
+                    "stabilityai/sdxl-turbo",       # very fast SDXL
+                    "runwayml/stable-diffusion-v1-5",
+                    "stabilityai/stable-diffusion-2-1"
+                ]
+            )
+            size = gr.Dropdown(
+                label="Image Size",
+                value="1024x1024",
+                choices=["512x512", "768x768", "1024x1024", "768x1024 (portrait)", "1024x768 (landscape)"]
             )
-
-            randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
 
             with gr.Row():
-                width = gr.Slider(
-                    label="Width",
-                    minimum=256,
-                    maximum=MAX_IMAGE_SIZE,
-                    step=32,
-                    value=1024,  # Replace with defaults that work for your model
-                )
-
-                height = gr.Slider(
-                    label="Height",
-                    minimum=256,
-                    maximum=MAX_IMAGE_SIZE,
-                    step=32,
-                    value=1024,  # Replace with defaults that work for your model
-                )
+                guidance_scale = gr.Slider(0.0, 12.0, value=2.0, step=0.5, label="Guidance scale (SDXL-Turbo likes low)")
+                steps = gr.Slider(2, 50, value=8, step=1, label="Steps")
 
             with gr.Row():
-                guidance_scale = gr.Slider(
-                    label="Guidance scale",
-                    minimum=0.0,
-                    maximum=10.0,
-                    step=0.1,
-                    value=0.0,  # Replace with defaults that work for your model
-                )
-
-                num_inference_steps = gr.Slider(
-                    label="Number of inference steps",
-                    minimum=1,
-                    maximum=50,
-                    step=1,
-                    value=2,  # Replace with defaults that work for your model
-                )
-
-        gr.Examples(examples=examples, inputs=[prompt])
-    gr.on(
-        triggers=[run_button.click, prompt.submit],
-        fn=infer,
-        inputs=[
-            prompt,
-            negative_prompt,
-            seed,
-            randomize_seed,
-            width,
-            height,
-            guidance_scale,
-            num_inference_steps,
-        ],
-        outputs=[result, seed],
+                batch_per_prompt = gr.Slider(1, 6, value=2, step=1, label="Images per prompt")
+                seed = gr.Textbox(label="Seed (optional, integer)")
+
+            run_btn = gr.Button("Generate", variant="primary")
+
+        with gr.Column():
+            gallery = gr.Gallery(label="Results", show_label=True, columns=3, height=520)
+            zip_file = gr.File(label="Download all images (.zip)")
+            status = gr.Markdown("")
+
+    def on_size_change(s):
+        if "x" in s and s.count("x") == 1 and "(" not in s:
+            w, h = s.split("x")
+            return int(w), int(h)
+        if s == "768x1024 (portrait)":
+            return 768, 1024
+        if s == "1024x768 (landscape)":
+            return 1024, 768
+        return 1024, 1024
+
+    width = gr.State(1024)
+    height = gr.State(1024)
+    size.change(fn=on_size_change, inputs=size, outputs=[width, height])
+
+    run_btn.click(
+        fn=generate_images,
+        inputs=[prompts_text, negative_prompt, model_id, width, height, guidance_scale, steps, batch_per_prompt, seed],
+        outputs=[gallery, zip_file, status]
     )
 
 if __name__ == "__main__":
     demo.launch()
+