Update app.py

app.py

@@ -1,154 +1,141 @@
 import gradio as gr
-import numpy as np
-import random
+import torch, random
+from diffusers import StableDiffusionXLPipeline, StableDiffusionXLImg2ImgPipeline, StableDiffusionXLInpaintPipeline, StableDiffusionLatentUpscalePipeline
 
-# import spaces #[uncomment to use ZeroGPU]
-from diffusers import DiffusionPipeline
-import torch
+# ---------------------------
+# HARD CHOICES (don’t change)
+# ---------------------------
+BASE_MODEL = "Lykon/dreamshaper-xl-1-0"          # High-quality SDXL finetune
+UPSCALE_MODEL = "stabilityai/sdxl-upscaler"      # x4 upscaler
 
 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,
-    negative_prompt,
-    seed,
-    randomize_seed,
-    width,
-    height,
-    guidance_scale,
-    num_inference_steps,
-    progress=gr.Progress(track_tqdm=True),
-):
-    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",
-]
+dtype = torch.float16 if device == "cuda" else torch.float32
+
+# Load base text2img pipeline
+pipe = StableDiffusionXLPipeline.from_pretrained(
+    BASE_MODEL, torch_dtype=dtype, use_safetensors=True
+)
+pipe.to(device)
+
+# Load upscaler (x4)
+upscale_pipe = StableDiffusionLatentUpscalePipeline.from_pretrained(
+    UPSCALE_MODEL, torch_dtype=dtype, use_safetensors=True
+).to(device)
+
+# ---------------------------
+# Presets & helpers
+# ---------------------------
+ASPECTS = {
+    "1:1 (Square | 4096x4096)": (4096, 4096),
+    "16:9 (UHD | 3840x2160)": (3840, 2160),
+    "9:16 (Phone Tall | 2160x3840)": (2160, 3840),  # Zedge/phone
+    "4:5 (Portrait | 4000x5000)": (4000, 5000),     # Adobe Stock favorite
+    "3:4 (Poster | 3000x4000)": (3000, 4000),
+}
 
-css = """
-#col-container {
-    margin: 0 auto;
-    max-width: 640px;
+STYLE_PRESETS = {
+    "Realistic Photo": {
+        "pos": "ultra-detailed realistic photo, natural lighting, sharp focus, DSLR depth of field, film grain subtle",
+        "neg": "blurry, lowres, artifact, extra fingers, deformed, watermark, text, logo, signature"
+    },
+    "3D Render": {
+        "pos": "high poly 3D render, global illumination, octane style, cinematic light, clean materials",
+        "neg": "lowpoly, noisy, aliased, bad shading, watermark, text, logo"
+    },
+    "Flat Illustration": {
+        "pos": "flat vector illustration, minimal shapes, smooth gradients, simple background, clean lines",
+        "neg": "noise, gradients banding, messy, sketchy, text"
+    },
+    "Minimalist": {
+        "pos": "minimalist composition, lots of negative space, clean background, simple geometric forms",
+        "neg": "cluttered, busy, noisy, text"
+    },
+    "Neon Wallpaper": {
+        "pos": "neon glow, vibrant colors, high contrast, dark background, crisp edges, perfect symmetry",
+        "neg": "dull, washed out, blurry, text, watermark"
+    }
 }
-"""
-
-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")
-
-        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,
-            )
-
-            seed = gr.Slider(
-                label="Seed",
-                minimum=0,
-                maximum=MAX_SEED,
-                step=1,
-                value=0,
-            )
-
-            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
-                )
-
-            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],
+
+def seed_or_random(seed):
+    if seed in [None, "", 0]:
+        return random.randint(1, 2**31 - 1)
+    return int(seed)
+
+def generate(prompt, style, aspect_key, steps, guidance, batch, seed, add_negatives):
+    if not prompt or len(prompt.strip()) < 3:
+        return [None]*batch, f"❗ Please enter a prompt."
+    w, h = ASPECTS[aspect_key]
+    s = seed_or_random(seed)
+    pos = prompt
+    neg = ""
+
+    # apply style preset
+    if style in STYLE_PRESETS:
+        pos = f"{STYLE_PRESETS[style]['pos']}, {prompt}"
+        if add_negatives:
+            neg = STYLE_PRESETS[style]['neg']
+
+    generator = torch.Generator(device=device).manual_seed(s)
+
+    images = pipe(
+        prompt=pos,
+        negative_prompt=neg,
+        guidance_scale=float(guidance),
+        num_inference_steps=int(steps),
+        width=int(w),
+        height=int(h),
+        num_images_per_prompt=int(batch),
+        generator=generator
+    ).images
+
+    # Simple quality guard: remove any None
+    images = [im for im in images if im is not None]
+    return images, f"✅ Done. Seed: {s} | {aspect_key} | {style}"
+
+def upscale(image, steps_up, guidance_up):
+    if image is None:
+        return None, "❗ Generate an image first."
+    # Latent upscaler expects an image and prompt; prompt can be empty for neutral upscale
+    result = upscale_pipe(
+        prompt="high quality, detailed, clean edges, no artifacts",
+        image=image,
+        guidance_scale=float(guidance_up),
+        num_inference_steps=int(steps_up)
+    ).images[0]
+    return result, "✅ Upscaled x4"
+
+with gr.Blocks(title="My Personal MidJourney-Style Tool") as demo:
+    gr.Markdown(
+        "# 🔮 Personal MidJourney-Style Tool\n"
+        "Optimized for **Freepik / Adobe Stock / Zedge**.\n"
+        "- Choose a **style** → set **aspect ratio** → **Generate** (x4 grid) → **Upscale x4** for stock.\n"
+        "- Tip: Use descriptive nouns, materials, lighting, and mood."
     )
 
-if __name__ == "__main__":
-    demo.launch()
+    with gr.Row():
+        with gr.Column(scale=1):
+            prompt = gr.Textbox(label="Prompt (describe what you want)", placeholder="e.g., futuristic city skyline at dusk, reflective water, birds, cinematic")
+            style = gr.Dropdown(choices=list(STYLE_PRESETS.keys()), value="Realistic Photo", label="Style preset")
+            aspect = gr.Dropdown(choices=list(ASPECTS.keys()), value="16:9 (UHD | 3840x2160)", label="Aspect Ratio")
+            steps = gr.Slider(5, 50, value=28, step=1, label="Steps (quality ↑ but slower)")
+            guidance = gr.Slider(1.0, 12.0, value=6.5, step=0.5, label="Prompt Strength (CFG)")
+            batch = gr.Slider(1, 4, value=4, step=1, label="Batch (number of images)")
+            seed = gr.Number(value=0, label="Seed (0 = random)")
+            add_negs = gr.Checkbox(value=True, label="Add smart negative prompts")
+            gen_btn = gr.Button("🚀 Generate")
+
+        with gr.Column(scale=1):
+            gallery = gr.Gallery(label="Results", show_label=True, columns=2, rows=2, height=600)
+            status = gr.Markdown()
+
+            gr.Markdown("### Upscale (pick the best image in gallery and paste below)")
+            to_upscale = gr.Image(type="pil", label="Image to upscale (paste or upload one of the results)")
+            steps_up = gr.Slider(5, 30, value=15, step=1, label="Upscale Steps")
+            guidance_up = gr.Slider(1.0, 10.0, value=5.0, step=0.5, label="Upscale Guidance")
+            up_btn = gr.Button("⬆️ Upscale x4")
+            up_out = gr.Image(type="pil", label="Upscaled Output")
+
+    gen_btn.click(generate, [prompt, style, aspect, steps, guidance, batch, seed, add_negs], [gallery, status])
+    up_btn.click(upscale, [to_upscale, steps_up, guidance_up], [up_out, status])
+
+demo.launch()