prompts.

Turn this image into a professional quality studio shoot with better lighting and depth of field.

Create a product advertisement with the man, cap, and soda.

0. 使用banana pro 1. 请用banana中文翻译替换掉图片里的日文。是的，生成一张只有中文的图！ 3. 这句台词的大意是：[]

文本框中，输入以下提示词（可根据需要修改）： ``` 0. 使用banana pro 1. 请用banana中文翻译替换掉图片里的日文。是的，生成一张只有中文的图！ 3. 这句台词的大意是：[] ``` **步骤 4：配置 API** 展开

* **修改细节**：调整表情、改变发色、添加配饰等 * **场景修改**：将白天改为夜晚、将晴天改为雨天等 * **对象替换**：替换图片中的特定元素 ### 场景三：批量处理 处理数百张格式相同或需要统一修改的图片时： 1. 上传文件夹 2. 在一张图上设置好选区和提示词 3. 勾选

Positive prompts | (ultra realistic,32k,masterpiece:1.4),(high detailed skin:1.1),( high quality:1.1), (ultra high res:1.4), (masterpiece), (beautiful lighting:1.4), (high quality), (high...

Positive prompts | masterpiece, best quality, very aesthetic, 8K, HDR, newest, anime_screencap, dynamic pose, depth of field, rakugakingu, FlatNika, in the style of cksc, foreshortening, ...

A hyper-detailed digital painting in a surreal fantasy style, showcasing a colossal (ornate crystal wine glass) with intricate filigree etching, filled to the brim with a viscous crimson ...

Use $nano-banana-app-store-campaign to analyze this app, tell me which screenshots to capture, lock a Nano Banana Pro style anchor, and generate a consistent App Store screenshot campaign.

A high-contrast black-and-white artistic portrait of me (use my image with 100% accurate facial details). I am seated on the floor in a dramatic, cinematic atmosphere, wearing an oversized dark coat that enhances a stylish and mysterious aura. The pose is introspective and expressive: one hand rests near my mouth while my head is turned slightly to the side, capturing a thoughtful mood. Sharp, moody window shadows fall across the wall behind me, creating a striking balance of light and darkness. The overall style is minimal, emotional, and cinematic blending elegance, mystery, and sophistication in a timeless black-and-white aesthetic. 4k Resolution, 9:16 ratio.

Create a compact 4:5 vertical close-up image showing a young woman using the exact facial features from the attached reference photo, with no changes. The image appears on the screen of a compact Canon-style digital camera, with the camera body visible around the preview, including buttons, dials, textured surface, FUNC/SET wheel, DISP button, and an “IMAGE STABILIZER” label. No text or logos inside the photo preview.On the screen, the woman is indoors at night, lit by a harsh built-in flash. Long dark hair falls across part of her face, lips slightly parted, candid expression. Background is a dim, cluttered kitchen, softly blurred. Early-2000s digital camera vibe with cool muted tones, strong flash contrast, subtle screen grain, and authentic display glow.

Hyper-realistic studio portrait of my face, side-profile view, facing left, wearing a black turtleneck, same hairstyle as reference. Low-key lighting with strong rim light outlining the face and hair, deep black background, dramatic contrast, cinematic mood. Sharp facial features, natural skin texture, subtle highlights on nose and lips, soft falloff shadows, professional studio photography, shallow depth of field, ultra-high detail, photorealistic, minimalist composition, moody and elegant atmosphere. Aspect ratio 9:16, high resolution, no text, no watermark.

Create a cinematic close-up portrait of a women use image for face reference seen through a reflective glass window. They have messy dark brown hair and ultra-realistic skin texture with visible pores and natural imperfections. One green-hazel eye is in sharp focus while the rest of the face softly fades into blur. Lighting is moody and low-key, with glowing yellow and orange bokeh lights reflecting across the glass in the foreground. Shot with an extremely shallow depth of field at f/1.2 to create an emotional, intimate feel, enhanced by subtle film grain and high-resolution detail.

Assault rifle (AR-15/M4 style) with optic sight and suppressor

A quote card with "{argument name="quote" default="Stay hungry, stay foolish"}" by {argument name="author" default="Steve Jobs"}

Failed to generate character references

Open [http://localhost:8000](http://localhost:8000) in your browser. ## How It Works 1. **Story Analysis**: Gemini analyzes your narrative to extract characters, settings, themes, and emotional arcs 2. **Character Design**: Generates consistent character reference sheets with detailed visual descriptions 3. **Comic Layout Plan**: Intelligently breaks your story into sequential panels with scene descriptions and dialogue 4. **Panel Generation**: Creates manga/comic panels while maintaining character consistency using reference sheets 5. **Create Shareable Image**: Combine panels and character designs into a 1200x1200px social media-ready image ## Usage Visit **[app.storytomanga.com](https://app.storytomanga.com)** or run locally: 1. Choose your preferred style (Manga or Comic) 2. Paste your story (max 500 words) 3. Click "Generate Comic" 4. Watch as the AI progressively: - Analyzes your story structure - Creates character reference sheets - Plans the comic layout - Generates individual panels 5. Download individual panels or create a shareable composite image ## Importing from Reddit You can automatically load a story from a Reddit post by going to a URL with the path to the post. **Example:**

Story to Manga - Turn any short story into manga in seconds. | Product Hunt

4. Watch as the AI progressively: - Analyzes your story structure - Creates character reference sheets - Plans the comic layout - Generates individual panels 5. Download individual panels or create a shareable composite image ## Importing from Reddit You can automatically load a story from a Reddit post by going to a URL with the path to the post. **Example:** ``` https://app.storytomanga.com/r/<subreddit>/comments/... ``` This will load the story from the URL and automatically start the comic generation process. ## Deployment to AWS Amplify 1. Push your code to GitHub 2. Connect your GitHub repo to AWS Amplify 3. Add environment variable `GOOGLE_AI_API_KEY` in Amplify console 4. Deploy automatically on push ## Story Guidelines **Best Results:** - 200-500 words - Clear character descriptions - Simple, focused plots - Dialogue-heavy scenes work well **Avoid:** - Very complex plots with many characters - Stories requiring specific visual references - Adult or inappropriate content ## Troubleshooting **

Create a minimalist white background

## 📱 How to Use ### Step 1: Upload or Select Product - Use sample images for quick testing - Upload your own product image (PNG/JPG) - Best results with clear product shots ### Step 2: Configure Brand - Set brand name, tone, and colors - Define visual guidelines - Add product details and tagline ### Step 3: Generate Assets - Select asset types to create - Choose style preset - Enable "Ensure brand text" for clear labeling - Click Generate All ### Step 4: Save & Edit - Save generated assets to gallery - Use chat interface for edits - Download final results ## 💡 Pro Tips ### For Best Results: 1. **Product Images**: Use high-quality images with clear product visibility 2. **Brand Text**: Enable "Ensure brand name visible" for professional results 3. **Chat Edits**: Be specific - "Add warm sunset lighting" vs "change lighting" 4. **Multiple Versions**: Save different variations for A/B testing ### Chat Commands Examples: - "Change the background to a tropical beach" - "Add golden hour lighting" - "Make it more luxurious and premium" - "Place on a marble surface with rose petals" - "Add water droplets for freshness" - "Create a minimalist white background" ## 🔧 Customization ### Adding New Asset Types Edit `ASSET_CONFIGS` in `ui_components.py`:

# Generate from a prompt straighten generate "a perfectly straight banana" # Modify an existing image straighten generate "add wings to the cat" --image cat.jpg # Condition on multiple images straighten generate "blend styles" -i style1.png -i style2.jpg # With custom settings straighten generate "futuristic city" --iterations 10 --save-all

straighten generate "a perfectly straight banana" # Modify an existing image straighten generate

make this photo look like a painting", input_image=input_img