Update app.py

app.py

@@ -1,34 +1,31 @@
 import gradio as gr
+import torch
 import numpy as np
 import random
+from PIL import Image
 
-# import spaces #[uncomment to use ZeroGPU]
-from diffusers import DiffusionPipeline
-import torch
+from src.pipeline import DiT360Pipeline
 
 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)
+model_repo = "/media/nfs/tmp_data/fenghr/download/FLUX.1-dev"
+lora_weights = "/media/nfs/tmp_data/fenghr/DiT360-FLUX-Panorama-Lora"
 
-MAX_SEED = np.iinfo(np.int32).max
-MAX_IMAGE_SIZE = 1024
+pipe = DiT360Pipeline.from_pretrained(model_repo, torch_dtype=torch_dtype).to(device)
+pipe.load_lora_weights(lora_weights)
 
+MAX_SEED = np.iinfo(np.int32).max
+MAX_IMAGE_SIZE = 2048
 
-# @spaces.GPU #[uncomment to use ZeroGPU]
 def infer(
     prompt,
-    negative_prompt,
     seed,
     randomize_seed,
     width,
-    height,
     guidance_scale,
     num_inference_steps,
     progress=gr.Progress(track_tqdm=True),
@@ -36,119 +33,112 @@ def infer(
     if randomize_seed:
         seed = random.randint(0, MAX_SEED)
 
-    generator = torch.Generator().manual_seed(seed)
+    height = width // 2
+    generator = torch.Generator(device=device).manual_seed(seed)
+
+    full_prompt = f"This is a panorama. The images shows {prompt.strip()}"
 
     image = pipe(
-        prompt=prompt,
-        negative_prompt=negative_prompt,
-        guidance_scale=guidance_scale,
+        full_prompt,
+        width=int(width),
+        height=int(height),
         num_inference_steps=num_inference_steps,
-        width=width,
-        height=height,
+        guidance_scale=guidance_scale,
         generator=generator,
     ).images[0]
+    image.save("test.png")
 
     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",
+    "A medieval castle stands proudly on a hilltop surrounded by autumn forests, with golden light spilling across the landscape.",
+    "A futuristic cityscape under a starry night sky.",
+    "A tranquil beach with palm trees and turquoise water at sunset.",
+    "A snowy mountain village under northern lights, with cozy cabins and smoke rising from chimneys.",
 ]
 
+
 css = """
-#col-container {
+#main-container {
+    display: flex;
+    flex-direction: row;
+    justify-content: center;
+    align-items: flex-start;
+    gap: 2rem;
+    margin-top: 1rem;
+}
+
+#image-panel {
+    flex: 2; /* 占2/3 */
+    max-width: 900px;
     margin: 0 auto;
-    max-width: 640px;
+}
+
+#settings-panel {
+    flex: 1; /* 占1/3 */
+    max-width: 280px;
+}
+
+#prompt-box textarea {
+    resize: none !important; /* 去掉上下箭头 */
 }
 """
 
-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) as demo:
+    gr.Markdown("# 🌀 DiT360: High-Fidelity Panoramic Image Generation")
+    gr.Markdown("Official Gradio demo for **DiT360**, a panoramic image generation model based on hybrid training.")
+
+    with gr.Row(elem_id="main-container"):
+        with gr.Column(elem_id="image-panel"):
+            result = gr.Image(label="Generated Panorama", show_label=False, type="pil", height=360)
+
+            with gr.Column():
+                prompt = gr.Textbox(
+                    elem_id="prompt-box",
+                    placeholder="Describe your panoramic scene here...",
+                    show_label=False,
+                    lines=2,
+                    container=False,
+                )
+                run_button = gr.Button("Generate", variant="primary")
 
-        result = gr.Image(label="Result", show_label=False)
+            gr.Examples(examples=examples, inputs=[prompt])
 
-        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.Column(elem_id="settings-panel"):
+            gr.Markdown("### ⚙️ Settings")
+            gr.Markdown(
+                "For best results, it is **not recommended** to modify `width` or `guidance scale`. "
+                "The height is automatically set to half the width (2:1 aspect ratio)."
             )
 
-            seed = gr.Slider(
-                label="Seed",
-                minimum=0,
-                maximum=MAX_SEED,
-                step=1,
-                value=0,
-            )
+            seed = gr.Slider(0, MAX_SEED, value=0, step=1, label="Seed")
+            randomize_seed = gr.Checkbox(label="Randomize seed", value=False)
 
-            randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
+            width = gr.Slider(1024, MAX_IMAGE_SIZE, value=2048, step=64, label="Width (fixed 2:1)")
+            height_display = gr.Number(value=1024, label="Height", interactive=False)
 
-            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
-                )
+            guidance_scale = gr.Slider(0.0, 10.0, value=2.8, step=0.1, label="Guidance Scale")
+            num_inference_steps = gr.Slider(28, 100, value=50, step=1, label="Inference Steps")
 
-                height = gr.Slider(
-                    label="Height",
-                    minimum=256,
-                    maximum=MAX_IMAGE_SIZE,
-                    step=32,
-                    value=1024,  # Replace with defaults that work for your model
-                )
+            def update_height(width):
+                return width // 2
 
-            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
-                )
+            width.change(fn=update_height, inputs=width, outputs=height_display)
 
-                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.Markdown(
+        "💡 *Tip: Try descriptive prompts like “A mountain village at sunrise with mist over the valley.” "
+        "DiT360 will automatically add a trigger word.*"
+    )
 
-        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,
-        ],
+        inputs=[prompt, seed, randomize_seed, width, guidance_scale, num_inference_steps],
         outputs=[result, seed],
     )
 
+
 if __name__ == "__main__":
     demo.launch()