Fix

app.py

@@ -1,21 +1,20 @@
 import os, gc, random
 import gradio as gr
 import torch, spaces
-from PIL import Image
 from diffusers import StableDiffusionPipeline, DPMSolverMultistepScheduler
 
 # ---- config ----
 MODEL_ID = "runwayml/stable-diffusion-v1-5"
-DTYPE = torch.float16
-HF_TOKEN = os.getenv("HF_TOKEN")  # optional (only needed for private models)
+DTYPE = torch.float16  # ZeroGPU slice supports fp16
+HF_TOKEN = os.getenv("HF_TOKEN")  # optional (only for private models)
 AUTH = {"token": HF_TOKEN} if HF_TOKEN else {}
 
-# cache for lazy loading
-_PIPE = {"sd": None}
+# lazy cache
+_PIPE = None
 
-def _get_pipe():
-    """Lazy-load SD1.5 and enable memory savers for ZeroGPU."""
-    if _PIPE["sd"] is None:
+def get_pipe():
+    global _PIPE
+    if _PIPE is None:
         pipe = StableDiffusionPipeline.from_pretrained(
             MODEL_ID,
             torch_dtype=DTYPE,
@@ -24,38 +23,30 @@ def _get_pipe():
             low_cpu_mem_usage=True,
             **AUTH
         )
+        # fast, stable scheduler
         pipe.scheduler = DPMSolverMultistepScheduler.from_config(
             pipe.scheduler.config, use_karras_sigmas=True, algorithm_type="dpmsolver++"
         )
+        # memory savers (good for Spaces/ZeroGPU)
         pipe.enable_attention_slicing()
         pipe.enable_vae_slicing()
         pipe.enable_model_cpu_offload()
-        _PIPE["sd"] = pipe
-    return _PIPE["sd"]
+        _PIPE = pipe
+    return _PIPE
 
-def _snap_dim(x: int) -> int:
-    # diffusers likes multiples of 8; stay within safe VRAM for ZeroGPU
+def snap8(x: int) -> int:
     x = max(256, min(1024, int(x)))
     return x - (x % 8)
 
-@spaces.GPU(duration=120)  # allocate a ZeroGPU slice only during generation
-def generate(prompt: str,
-             negative_prompt: str,
-             steps: int,
-             guidance_scale: float,
-             width: int,
-             height: int,
-             seed: int):
-    pipe = _get_pipe()
+@spaces.GPU(duration=120)
+def generate(prompt: str, negative: str, steps: int, cfg: float, width: int, height: int, seed: int):
+    pipe = get_pipe()
+    w, h = snap8(width), snap8(height)
 
-    w = _snap_dim(width)
-    h = _snap_dim(height)
-
-    # seed handling (reproducible on CUDA)
-    g = torch.Generator(device="cuda")
-    if int(seed) == -1:
+    # seed
+    if int(seed) < 0:
         seed = random.randint(0, 2**31 - 1)
-    g = g.manual_seed(int(seed))
+    gen = torch.Generator(device="cuda").manual_seed(int(seed))
 
     if torch.cuda.is_available():
         torch.cuda.empty_cache()
@@ -64,42 +55,33 @@ def generate(prompt: str,
     with torch.autocast(device_type="cuda", dtype=DTYPE):
         out = pipe(
             prompt=str(prompt),
-            negative_prompt=str(negative_prompt or ""),
+            negative_prompt=str(negative or ""),
             num_inference_steps=int(steps),
-            guidance_scale=float(guidance_scale),
+            guidance_scale=float(cfg),
             width=w, height=h,
-            generator=g,
+            generator=gen,
         )
-    img: Image.Image = out.images[0]
-    return img, seed
+    return out.images[0]
 
-# ---------- UI ----------
+# -------- UI (one output, zero fancy) --------
 with gr.Blocks() as demo:
-    gr.Markdown("# 🧩 Stable Diffusion 1.5 (ZeroGPU)\nText prompt → image, lean & fast.")
+    gr.Markdown("# 🎨 Stable Diffusion 1.5 — ZeroGPU (minimal)")
 
     with gr.Row():
         with gr.Column():
-            prompt = gr.Textbox(
-                label="Prompt",
-                value="a cozy reading nook with warm sunlight, soft textures, cinematic lighting, highly detailed"
-            )
-            negative = gr.Textbox(
-                label="Negative prompt",
-                value="lowres, blurry, watermark, text, logo, nsfw"
-            )
-            steps = gr.Slider(4, 50, value=28, step=1, label="Steps")
-            cfg = gr.Slider(1.0, 12.0, value=7.0, step=0.5, label="CFG scale")
-            width = gr.Slider(256, 1024, value=640, step=16, label="Width")
-            height = gr.Slider(256, 1024, value=640, step=16, label="Height")
-            seed = gr.Number(value=-1, precision=0, label="Seed (-1 = random)")
-
-            btn = gr.Button("Generate", variant="primary")
+            prompt  = gr.Textbox(label="Prompt", value="a cozy reading nook, warm sunlight, cinematic lighting, highly detailed")
+            negative = gr.Textbox(label="Negative (optional)", value="lowres, blurry, watermark, text")
+            steps   = gr.Slider(8, 40, value=28, step=1, label="Steps")
+            cfg     = gr.Slider(1.0, 12.0, value=7.0, step=0.5, label="CFG")
+            width   = gr.Slider(256, 1024, value=640, step=16, label="Width")
+            height  = gr.Slider(256, 1024, value=640, step=16, label="Height")
+            seed    = gr.Number(value=-1, precision=0, label="Seed (-1 random)")
+            btn     = gr.Button("Generate", variant="primary")
         with gr.Column():
-            out_img = gr.Image(label="Result", interactive=False)
-            out_seed = gr.Number(label="Used seed", interactive=False)
+            out = gr.Image(label="Result", interactive=False)
 
-    btn.click(generate, [prompt, negative, steps, cfg, width, height, seed], [out_img, out_seed])
+    btn.click(generate, [prompt, negative, steps, cfg, width, height, seed], out)
 
 if __name__ == "__main__":
-    demo.queue(max_size=8).launch()
-
+    # keep it plain so the Space "just builds"
+    demo.launch()

requirements.txt

@@ -5,7 +5,3 @@ accelerate>=0.31.0
 safetensors
 gradio==4.44.1
 huggingface-hub
-Pillow
-spaces
-numpy
-mediapipe