Update app.py

app.py

@@ -13,13 +13,12 @@ from huggingface_hub import InferenceClient
 
 
 help_text = """
-To optimize image editing results:
+To optimize image results:
 - Adjust the **Image CFG weight** if the image isn't changing enough or is changing too much. Lower it to allow bigger changes, or raise it to preserve original details.
 - Modify the **Text CFG weight** to influence how closely the edit follows text instructions. Increase it to adhere more to the text, or decrease it for subtler changes.
 - Experiment with different **random seeds** and **CFG values** for varied outcomes.
 - **Rephrase your instructions** for potentially better results.
 - **Increase the number of steps** for enhanced edits.
-- For better facial details, especially if they're small, **crop the image** to enlarge the face's presence.
 """
 
 def set_timesteps_patched(self, num_inference_steps: int, device = None):
@@ -36,7 +35,7 @@ def set_timesteps_patched(self, num_inference_steps: int, device = None):
     self._begin_index = None
     self.sigmas = self.sigmas.to("cpu") 
 
-
+# Image Editor
 edit_file = hf_hub_download(repo_id="stabilityai/cosxl", filename="cosxl_edit.safetensors")
 normal_file = hf_hub_download(repo_id="stabilityai/cosxl", filename="cosxl.safetensors")
 
@@ -57,32 +56,36 @@ if not torch.cuda.is_available():
 
 device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
 
+
+# Image Generator
 if torch.cuda.is_available():
     pipe = StableDiffusionXLPipeline.from_pretrained(
-        "sd-community/sdxl-flash",
+        "fluently/Fluently-XL-v4",
         torch_dtype=torch.float16,
         use_safetensors=True,
-        add_watermarker=False
     )
     pipe.scheduler = EulerAncestralDiscreteScheduler.from_config(pipe.scheduler.config)
+    pipe.load_lora_weights("ehristoforu/dalle-3-xl-v2", weight_name="dalle-3-xl-lora-v2.safetensors", adapter_name="dalle")
+    pipe.set_adapters("dalle")
 
 def randomize_seed_fn(seed: int, randomize_seed: bool) -> int:
     if randomize_seed:
         seed = random.randint(0, 999999)
     return seed
 
+# Generator
 @spaces.GPU(duration=30, queue=False)
 def king(type = "Image Generation",
         input_image = None,
         instruction: str = "Eiffel tower",
         steps: int = 8,
         randomize_seed: bool = False,
-        seed: int = 24,
+        seed: int = 25,
         text_cfg_scale: float = 7.3,
         image_cfg_scale: float = 1.7,
         width: int = 1024,
         height: int = 1024,
-        guidance_scale: float = 3,
+        guidance_scale: float = 6.2,
         use_resolution_binning: bool = True,
         progress=gr.Progress(track_tqdm=True),
     ):
@@ -92,7 +95,7 @@ def king(type = "Image Generation",
         image_cfg_scale = image_cfg_scale
         input_image = input_image
 
-        steps=steps*3
+        steps=steps
         generator = torch.manual_seed(seed)
         output_image = pipe_edit(
             instruction, image=input_image,
@@ -118,8 +121,7 @@ def king(type = "Image Generation",
         output_image = pipe(**options).images[0]
         return seed, output_image
 
-
-        
+# Prompt classifier
 def response(instruction, input_image=None):
     if input_image is None:
         output="Image Generation"
@@ -218,7 +220,7 @@ with gr.Blocks(css=css) as demo:
         inputs=[type,input_image, instruction],
         fn=king,
         outputs=[input_image],
-        cache_examples=False,
+        cache_examples=True,
     )
 
     gr.Markdown(help_text)