Update app.py

app.py

@@ -20,12 +20,12 @@ css = """
 def generate(prompt, samp_steps, seed, strength, progress=gr.Progress(track_tqdm=True)):
     if seed < 0:
         seed = random.randint(1,999999)
-    imagex = txt2img(
-        prompt,
-        num_inference_steps=1,
-        guidance_scale=0.0,
-        generator=torch.manual_seed(seed),
-    ).images[0]
+#    image = txt2img(
+#        prompt,
+#       num_inference_steps=1,
+#        guidance_scale=0.0,
+#        generator=torch.manual_seed(seed),
+#    ).images[0]
     imarray = numpy.random.rand(512,512,3) * 255
     image = Image.fromarray(imarray.astype('uint8')).convert('RGBA')
     upscaled_image = image.resize((1024,1024), 1)
@@ -40,12 +40,12 @@ def generate(prompt, samp_steps, seed, strength, progress=gr.Progress(track_tqdm
     return [final_image], seed
         
 def set_base_models():
-    txt2img = AutoPipelineForText2Image.from_pretrained(
-        "stabilityai/sdxl-turbo",
-        torch_dtype = torch.float16,
-        variant = "fp16"
-    )
-    txt2img.to("cuda")
+#    txt2img = AutoPipelineForText2Image.from_pretrained(
+#        "stabilityai/sdxl-turbo",
+#        torch_dtype = torch.float16,
+#        variant = "fp16"
+#    )
+#    txt2img.to("cuda")
     img2img = StableDiffusionImg2ImgPipeline.from_pretrained(
         "Lykon/dreamshaper-8",
         torch_dtype = torch.float16,
@@ -53,7 +53,7 @@ def set_base_models():
         safety_checker=None
     )
     img2img.to("cuda")
-    return txt2img, img2img
+    return None, img2img
 
 with gr.Blocks(css=css) as demo:
     with gr.Column():