Update app.py

app.py

@@ -14,30 +14,38 @@ logging.basicConfig(level=logging.DEBUG, format='%(asctime)s - %(levelname)s - %
 
 # List of popular styles
 STYLES = [
-    "Photorealistic", "Oil Painting", "Watercolor", "Anime", 
+    "Hyper-realistic", "Oil Painting", "Watercolor", "Anime", 
     "Studio Ghibli", "Black and White", "Polaroid", "Sketch", 
     "3D Render", "Pixel Art", "Cyberpunk", "Steampunk", 
     "Art Nouveau", "Pop Art", "Minimalist"
 ]
 
-# Default negative prompt
+# Default negative prompt (hidden from UI)
 DEFAULT_NEGATIVE_PROMPT = """
 ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, 
 extra limbs, disfigured, deformed, body out of frame, bad anatomy, watermark, signature, 
-cut off, low contrast, underexposed, overexposed, bad art, beginner, amateur, distorted face
+cut off, low contrast, underexposed, overexposed, bad art, beginner, amateur, distorted face,
+plastic, cartoonish, artificial, fake, unnatural
 """
 
 app = dash.Dash(__name__, external_stylesheets=[dbc.themes.BOOTSTRAP])
 
 app.layout = dbc.Container([
-    html.H1("Stability AI SD3.5 Large Turbo Image Generator with Google Gemini Prompt Enhancement", className="my-4"),
+    html.H1("Insta-Image", className="my-4"),
     dbc.Card([
         dbc.CardBody([
             dbc.Input(id="google-api-key", type="password", placeholder="Enter Google AI API Key", className="mb-3"),
             dbc.Input(id="stability-api-key", type="password", placeholder="Enter Stability AI API Key", className="mb-3"),
             dbc.Textarea(id="prompt", placeholder="Enter your prompt", className="mb-3"),
             dcc.Dropdown(id="style", options=[{"label": s, "value": s} for s in STYLES], placeholder="Select style", className="mb-3"),
-            dbc.Textarea(id="negative-prompt", value=DEFAULT_NEGATIVE_PROMPT, className="mb-3"),
+            dbc.Row([
+                dbc.Col(dcc.Slider(id="cfg-scale", min=1, max=30, step=1, value=7, marks={1: '1', 15: '15', 30: '30'}, tooltip={"placement": "bottom", "always_visible": True}), width=6),
+                dbc.Col(html.Div("CFG Scale"), width=6),
+            ], className="mb-3"),
+            dbc.Row([
+                dbc.Col(dcc.Slider(id="steps", min=4, max=20, step=1, value=4, marks={4: '4', 12: '12', 20: '20'}, tooltip={"placement": "bottom", "always_visible": True}), width=6),
+                dbc.Col(html.Div("Steps"), width=6),
+            ], className="mb-3"),
             dbc.Button("Generate Image", id="submit-btn", color="primary", className="mb-3"),
         ])
     ], className="mb-4"),
@@ -60,16 +68,18 @@ def enhance_prompt(google_api_key, prompt, style):
     genai.configure(api_key=google_api_key)
     model = genai.GenerativeModel("gemini-2.0-flash-lite")
     enhanced_prompt_request = f"""
-    Task: Enhance the following prompt for image generation.
+    Task: Enhance the following prompt for high-quality, realistic image generation.
     Style: {style}
     Original prompt: '{prompt}'
     
     Instructions:
-    1. Expand the prompt to be more detailed and vivid.
-    2. Incorporate elements of the specified style.
-    3. Maintain the original intent of the prompt.
-    4. Provide ONLY the enhanced prompt, without any explanations or options.
-    5. Keep the enhanced prompt concise, ideally under 100 words.
+    1. Expand the prompt to be more detailed, vivid, and realistic in quality.
+    2. Incorporate elements of the specified style, focusing on realism and natural appearances.
+    3. Add details that enhance the realism of the scene, especially for elements like trees, textures, and lighting.
+    4. Avoid terms that might result in artificial or cartoonish appearances unless specified.
+    5. Maintain the original intent of the prompt while significantly improving its descriptive quality.
+    6. Provide ONLY the enhanced prompt, without any explanations or options.
+    7. Keep the enhanced prompt concise, ideally under 100 words.
 
     Enhanced prompt:
     """
@@ -90,7 +100,7 @@ def enhance_prompt(google_api_key, prompt, style):
         logging.error(f"Error in enhance_prompt: {str(e)}")
         raise
 
-def generate_image(stability_api_key, enhanced_prompt, style, negative_prompt):
+def generate_image(stability_api_key, enhanced_prompt, style, negative_prompt, cfg_scale, steps):
     url = "https://api.stability.ai/v2beta/stable-image/generate/sd3"
     
     headers = {
@@ -99,14 +109,15 @@ def generate_image(stability_api_key, enhanced_prompt, style, negative_prompt):
     }
     
     data = {
-        "prompt": f"{enhanced_prompt}, Style: {style}",
+        "prompt": f"{enhanced_prompt}, Style: {style}, highly detailed, photorealistic, high quality",
         "negative_prompt": negative_prompt,
         "model": "sd3.5-large-turbo",
         "output_format": "jpeg",
         "width": 1024,
         "height": 1024,
         "num_images": 1,
-        "steps": 4,  # SD3.5 Large Turbo generates high-quality images in just 4 steps
+        "steps": steps,
+        "cfg_scale": cfg_scale,
     }
     
     try:
@@ -127,13 +138,13 @@ def generate_image(stability_api_key, enhanced_prompt, style, negative_prompt):
         logging.error(f"Request failed: {str(e)}")
         raise Exception(f"Request failed: {str(e)}")
 
-def process_and_generate(google_api_key, stability_api_key, prompt, style, negative_prompt, set_status):
+def process_and_generate(google_api_key, stability_api_key, prompt, style, cfg_scale, steps, set_status):
     try:
         set_status("Enhancing prompt...")
         enhanced_prompt = enhance_prompt(google_api_key, prompt, style)
         
         set_status("Generating image...")
-        image_bytes = generate_image(stability_api_key, enhanced_prompt, style, negative_prompt)
+        image_bytes = generate_image(stability_api_key, enhanced_prompt, style, DEFAULT_NEGATIVE_PROMPT, cfg_scale, steps)
         
         set_status("Image generated successfully!")
         return image_bytes, enhanced_prompt
@@ -151,10 +162,11 @@ def process_and_generate(google_api_key, stability_api_key, prompt, style, negat
      State("stability-api-key", "value"),
      State("prompt", "value"),
      State("style", "value"),
-     State("negative-prompt", "value")],
+     State("cfg-scale", "value"),
+     State("steps", "value")],
     prevent_initial_call=True
 )
-def update_output(n_clicks, google_api_key, stability_api_key, prompt, style, negative_prompt):
+def update_output(n_clicks, google_api_key, stability_api_key, prompt, style, cfg_scale, steps):
     if n_clicks is None:
         raise PreventUpdate
     
@@ -164,7 +176,7 @@ def update_output(n_clicks, google_api_key, stability_api_key, prompt, style, ne
         status["message"] = message
     
     def run_process():
-        image_bytes, enhanced_prompt = process_and_generate(google_api_key, stability_api_key, prompt, style, negative_prompt, set_status)
+        image_bytes, enhanced_prompt = process_and_generate(google_api_key, stability_api_key, prompt, style, cfg_scale, steps, set_status)
         if image_bytes:
             encoded_image = base64.b64encode(image_bytes).decode('ascii')
             return f"data:image/jpeg;base64,{encoded_image}", f"Enhanced Prompt: {enhanced_prompt}", status["message"]