sync

app.py

@@ -17,7 +17,7 @@ for taking it to the next level by enabling inpainting with the FLUX.
 """
 
 MAX_SEED = np.iinfo(np.int32).max
-MAX_IMAGE_SIZE = 2048
+IMAGE_SIZE = 1024
 DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
 
 pipe = FluxInpaintPipeline.from_pretrained(
@@ -26,10 +26,15 @@ pipe = FluxInpaintPipeline.from_pretrained(
 
 def resize_image_dimensions(
     original_resolution_wh: Tuple[int, int],
-    maximum_dimension: int = 2048
+    maximum_dimension: int = IMAGE_SIZE
 ) -> Tuple[int, int]:
     width, height = original_resolution_wh
 
+    # if width <= maximum_dimension and height <= maximum_dimension:
+    #     width = width - (width % 32)
+    #     height = height - (height % 32)
+    #     return width, height
+
     if width > height:
         scaling_factor = maximum_dimension / width
     else:
@@ -41,13 +46,10 @@ def resize_image_dimensions(
     new_width = new_width - (new_width % 32)
     new_height = new_height - (new_height % 32)
 
-    new_width = min(maximum_dimension, new_width)
-    new_height = min(maximum_dimension, new_height)
-
     return new_width, new_height
 
 
-@spaces.GPU()
+@spaces.GPU(duration=100)
 def process(
     input_image_editor: dict,
     input_text: str,
@@ -79,7 +81,7 @@ def process(
     if randomize_seed_checkbox:
         seed_slicer = random.randint(0, MAX_SEED)
     generator = torch.Generator().manual_seed(seed_slicer)
-    return pipe(
+    result = pipe(
         prompt=input_text,
         image=resized_image,
         mask_image=resized_mask,
@@ -88,7 +90,9 @@ def process(
         strength=strength_slider,
         generator=generator,
         num_inference_steps=num_inference_steps_slider
-    ).images[0], resized_mask
+    ).images[0]
+    print('INFERENCE DONE')
+    return result, resized_mask
 
 
 with gr.Blocks() as demo:
@@ -120,7 +124,7 @@ with gr.Blocks() as demo:
                     minimum=0,
                     maximum=MAX_SEED,
                     step=1,
-                    value=0,
+                    value=42,
                 )
 
                 randomize_seed_checkbox_component = gr.Checkbox(
@@ -129,14 +133,19 @@ with gr.Blocks() as demo:
                 with gr.Row():
                     strength_slider_component = gr.Slider(
                         label="Strength",
+                        info="Indicates extent to transform the reference `image`. "
+                             "Must be between 0 and 1. `image` is used as a starting "
+                             "point and more noise is added the higher the `strength`.",
                         minimum=0,
                         maximum=1,
                         step=0.01,
-                        value=0.75,
+                        value=0.85,
                     )
 
                     num_inference_steps_slider_component = gr.Slider(
                         label="Number of inference steps",
+                        info="The number of denoising steps. More denoising steps "
+                             "usually lead to a higher quality image at the",
                         minimum=1,
                         maximum=50,
                         step=1,