Update app.py

app.py

@@ -68,7 +68,7 @@ base_model = "sayakpaul/FLUX.1-merged"
 taef1 = AutoencoderTiny.from_pretrained("madebyollin/taef1", torch_dtype=dtype).to(device)
 good_vae = AutoencoderKL.from_pretrained(base_model, subfolder="vae", torch_dtype=dtype).to(device)
 # Adjust the scaling factor for the base model's output
-scaling_factor = 0.05  # You can adjust this value as needed
+scaling_factor = 1.15  # You can adjust this value as needed
 pipe = DiffusionPipeline.from_pretrained(base_model, torch_dtype=dtype, vae=taef1).to(device)
 
 MAX_SEED = 2**32 - 1
@@ -433,7 +433,7 @@ def generate_image(prompt, steps, seed, cfg_scale, width, height, progress):
             width=width,
             height=height,
             generator=generator,
-            joint_attention_kwargs={"scale": 1.0},
+            joint_attention_kwargs={"scale": 1.15},
             output_type="pil",
             good_vae=good_vae,
         ):
@@ -479,15 +479,12 @@ def run_lora(prompt, cfg_scale, steps, selected_indices, lora_scale_1, lora_scal
     prompt_mash = " ".join(prepends + [prompt] + appends)
 
     # Print formatted log
-    print("\n" + "=" * 50)
-    print("=" * 50)
     print(f"📌 Prompt: {prompt}")
     print(f"🎭 Selected LoRAs:\n{lora_details}")
     print(f"🎛️  CFG Scale: {cfg_scale} | Steps: {steps}")
     print(f"🎲 Seed: {seed}")
     print(f"🖼️  Image Size: {width} x {height}")
     print("\n" + "=" * 50 + "\n")
-    print(f"\n🔀 Final Prompt: {prompt_mash}")
 
     # Unload previous LoRA weights
     with calculateDuration("Unloading LoRA"):
@@ -523,7 +520,6 @@ def run_lora(prompt, cfg_scale, steps, selected_indices, lora_scale_1, lora_scal
         yield image, seed, gr.update(value=progress_bar, visible=True)
 
     print("✅ Image Generation Complete!")
-    print("=" * 50 + "\n")
 
 run_lora.zerogpu = False