Update app.py

app.py

@@ -512,7 +512,6 @@ image_pipe = DiffusionPipeline.from_pretrained("ostris/Flex.1-alpha", torch_dtyp
 image_pipe.enable_model_cpu_offload()
 image_pipe.enable_vae_slicing()
 image_pipe.enable_vae_tiling()
-image_pipe.enable_sequential_cpu_offload()
 
 # functionality
 
@@ -673,14 +672,10 @@ def upscale_image(img,factor):
 
     return ret
 
-def all_pipes(pos,neg,artist,song,h,w):
+def all_pipes(pos,neg,h,w):
 
-    imgs = pipe_generate_image(pos,neg,h,w)
-
-    for i in range(len(imgs)):
-        imgs[i] = upscaler(imgs[i],upscale_factor=factor)
-
-    return imgs
+    img = pipe_generate_image(pos,neg,h,w)[0]
+    return img
 
 google_translate_endpoint = "https://translate.google.com/m"
 language_codes = {
@@ -1333,9 +1328,9 @@ def handle_generation(artist,song,lyrics,h,w):
         Negative: {neg}
     """)
 
-    img = all_pipes(pos,neg,pos_artist,pos_song,h,w)[0]
+    img = all_pipes(pos,neg,h,w)[0]
 
-    labeled_img = add_song_cover_text(img,artist,song,h,w)
+    labeled_img = add_song_cover_text(img,pos_artist,pos_song,h,w)
     name = f'{generate_random_string(8)}.png'
     labeled_img.save(name)
 
@@ -1375,18 +1370,18 @@ if __name__ == "__main__":
                 width = gr.Slider(
                     label="Width (px)",
                     minimum=512,
-                    maximum=3072,
-                    step=256,
-                    value=1536,
+                    maximum=1536,
+                    step=64,
+                    value=1024,
                 )
                 
     
                 height = gr.Slider(
                     label="Height (px)",
                     minimum=512,
-                    maximum=3072,
-                    step=256,
-                    value=1536,
+                    maximum=1536,
+                    step=64,
+                    value=1024,
                 )
     
                 lyrics = gr.Textbox(