Update app.py

app.py

@@ -14,7 +14,7 @@ logging.basicConfig(level=logging.DEBUG, format='%(asctime)s - %(levelname)s - %
 
 # List of popular styles
 STYLES = [
-    "Hyper-realistic", "Oil Painting", "Watercolor", "Anime", 
+    "Photorealistic", "Oil Painting", "Watercolor", "Anime", 
     "Studio Ghibli", "Black and White", "Polaroid", "Sketch", 
     "3D Render", "Pixel Art", "Cyberpunk", "Steampunk", 
     "Art Nouveau", "Pop Art", "Minimalist"
@@ -31,21 +31,65 @@ plastic, cartoonish, artificial, fake, unnatural
 app = dash.Dash(__name__, external_stylesheets=[dbc.themes.BOOTSTRAP])
 
 app.layout = dbc.Container([
-    html.H1("Insta-Image", className="my-4"),
+    html.H1("Stability AI SD3.5 Large Turbo Image Generator with Google Gemini Prompt Enhancement", 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.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.Accordion([
+                dbc.AccordionItem(
+                    [
+                        dbc.Row([
+                            dbc.Col([
+                                html.Label("CFG Scale:", className="mr-2"),
+                                dcc.Slider(id="cfg-scale", min=1, max=30, step=0.5, value=7, 
+                                           marks={1: '1', 15: '15', 30: '30'}, 
+                                           tooltip={"placement": "bottom", "always_visible": True}),
+                                dbc.Tooltip(
+                                    "Controls the influence of the prompt. Higher values adhere more closely to the prompt.",
+                                    target="cfg-scale",
+                                ),
+                            ], width=12, className="mb-3"),
+                        ]),
+                        dbc.Row([
+                            dbc.Col([
+                                html.Label("Steps:", className="mr-2"),
+                                dcc.Slider(id="steps", min=4, max=50, step=1, value=20, 
+                                           marks={4: '4', 25: '25', 50: '50'}, 
+                                           tooltip={"placement": "bottom", "always_visible": True}),
+                                dbc.Tooltip(
+                                    "Number of denoising steps. More steps can lead to higher quality but longer generation time.",
+                                    target="steps",
+                                ),
+                            ], width=12, className="mb-3"),
+                        ]),
+                        dbc.Row([
+                            dbc.Col([
+                                html.Label("Sampler:", className="mr-2"),
+                                dcc.Dropdown(
+                                    id="sampler",
+                                    options=[
+                                        {"label": "DDIM", "value": "DDIM"},
+                                        {"label": "PLMS", "value": "PLMS"},
+                                        {"label": "K_EULER", "value": "K_EULER"},
+                                        {"label": "K_EULER_ANCESTRAL", "value": "K_EULER_ANCESTRAL"},
+                                        {"label": "DPM_2", "value": "DPM_2"},
+                                        {"label": "DPM_2_ANCESTRAL", "value": "DPM_2_ANCESTRAL"},
+                                    ],
+                                    value="K_EULER_ANCESTRAL",
+                                ),
+                                dbc.Tooltip(
+                                    "The algorithm used for image generation. Different samplers can produce varying results.",
+                                    target="sampler",
+                                ),
+                            ], width=12, className="mb-3"),
+                        ]),
+                    ],
+                    title="Advanced Settings",
+                )
+            ], start_collapsed=True, className="mb-3"),
             dbc.Button("Generate Image", id="submit-btn", color="primary", className="mb-3"),
         ])
     ], className="mb-4"),
@@ -73,13 +117,14 @@ def enhance_prompt(google_api_key, prompt, style):
     Original prompt: '{prompt}'
     
     Instructions:
-    1. Expand the prompt to be more detailed, vivid, and realistic in quality.
+    1. Expand the prompt to be more detailed, vivid, and realism and always include the right camera used for the shot
     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.
+    4. Emphasize natural lighting and enhance the realism of textures and colors.
+    5. Avoid terms that might result in artificial or cartoonish appearances unless specifically requested
+    6. Maintain the original intent of the prompt while significantly improving its descriptive quality.
+    7. Provide ONLY the enhanced prompt, without any explanations or options.
+    8. Keep the enhanced prompt concise, ideally under 100 words.
 
     Enhanced prompt:
     """
@@ -100,7 +145,7 @@ 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, cfg_scale, steps):
+def generate_image(stability_api_key, enhanced_prompt, style, negative_prompt, cfg_scale, steps, sampler):
     url = "https://api.stability.ai/v2beta/stable-image/generate/sd3"
     
     headers = {
@@ -109,7 +154,7 @@ def generate_image(stability_api_key, enhanced_prompt, style, negative_prompt, c
     }
     
     data = {
-        "prompt": f"{enhanced_prompt}, Style: {style}, highly detailed, photorealistic, high quality",
+        "prompt": f"{enhanced_prompt}, Style: {style}, highly detailed, photorealistic, high quality, natural lighting, realistic textures and colors",
         "negative_prompt": negative_prompt,
         "model": "sd3.5-large-turbo",
         "output_format": "jpeg",
@@ -118,10 +163,11 @@ def generate_image(stability_api_key, enhanced_prompt, style, negative_prompt, c
         "num_images": 1,
         "steps": steps,
         "cfg_scale": cfg_scale,
+        "sampler": sampler,
     }
     
     try:
-        response = requests.post(url, headers=headers, files={"none": ''}, data=data)
+        response = requests.post(url, headers=headers, json=data)
         response.raise_for_status()
         
         logging.debug(f"Response headers: {response.headers}")
@@ -138,13 +184,13 @@ def generate_image(stability_api_key, enhanced_prompt, style, negative_prompt, c
         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, cfg_scale, steps, set_status):
+def process_and_generate(google_api_key, stability_api_key, prompt, style, cfg_scale, steps, sampler, 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, DEFAULT_NEGATIVE_PROMPT, cfg_scale, steps)
+        image_bytes = generate_image(stability_api_key, enhanced_prompt, style, DEFAULT_NEGATIVE_PROMPT, cfg_scale, steps, sampler)
         
         set_status("Image generated successfully!")
         return image_bytes, enhanced_prompt
@@ -163,10 +209,11 @@ def process_and_generate(google_api_key, stability_api_key, prompt, style, cfg_s
      State("prompt", "value"),
      State("style", "value"),
      State("cfg-scale", "value"),
-     State("steps", "value")],
+     State("steps", "value"),
+     State("sampler", "value")],
     prevent_initial_call=True
 )
-def update_output(n_clicks, google_api_key, stability_api_key, prompt, style, cfg_scale, steps):
+def update_output(n_clicks, google_api_key, stability_api_key, prompt, style, cfg_scale, steps, sampler):
     if n_clicks is None:
         raise PreventUpdate
     
@@ -176,7 +223,7 @@ def update_output(n_clicks, google_api_key, stability_api_key, prompt, style, cf
         status["message"] = message
     
     def run_process():
-        image_bytes, enhanced_prompt = process_and_generate(google_api_key, stability_api_key, prompt, style, cfg_scale, steps, set_status)
+        image_bytes, enhanced_prompt = process_and_generate(google_api_key, stability_api_key, prompt, style, cfg_scale, steps, sampler, 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"]