# IllustriousCLIPTextEncoder Turns your creative ideas into instructions the AI understands. This is where your prompt becomes reality! ## What it does - Takes your written description and translates it for the AI - Makes sure the AI understands exactly what you want to create - Optimizes your text for best results with Illustrious models - The bridge between your imagination and the final image ## Inputs - **text**: Your creative description (what you want to see) - **clip**: The translator model (auto usually works great) - **normalize**: Keeps everything balanced (leave ON) ## Outputs - **CONDITIONING**: Your prompt ready for the AI to use ## Why This Matters Think of it as your AI interpreter: - **Your words** → This encoder → **AI instructions** → Beautiful art - Without this, the AI can't understand what you want - Better encoding = more accurate results ## Writing Better Prompts **Structure that works:** 1. **Subject first**: "1girl", "landscape", "cat" 2. **Important details next**: "blue hair, smiling" 3. **Scene/setting**: "beach, sunset" 4. **Style last**: "digital art, masterpiece" **Power Words:** - **Quality boosters**: masterpiece, best quality, high resolution - **Style definers**: anime, realistic, oil painting, watercolor - **Mood setters**: dramatic lighting, soft focus, vibrant colors - **Detail enhancers**: intricate, detailed, sharp focus ## The Magic of Weights Control what's important: - **Normal**: `blue hair` - **Important**: `(blue hair)` or `(blue hair:1.2)` - **Less important**: `[blue hair]` or `(blue hair:0.8)` - **Very important**: `((blue hair))` or `(blue hair:1.5)` ## Prompt Examples That Work **Anime Portrait:** ``` 1girl, blue hair, long hair, smile, school uniform, classroom, window light, soft focus, masterpiece ``` **Fantasy Landscape:** ``` fantasy landscape, floating islands, waterfalls, crystal formations, magical atmosphere, golden hour, epic scale, highly detailed ``` **Character Action:** ``` 1boy, warrior, sword, dynamic pose, fire magic, battle scene, dramatic lighting, motion blur, intense expression ``` ## Common Prompt Patterns **For Best Quality:** Start with: `masterpiece, best quality, high resolution` **For Specific Styles:** - Anime: Add `anime style, cel shading` - Realistic: Add `photorealistic, raw photo` - Artistic: Add `oil painting, impressionist` **For Better Composition:** - Add: `rule of thirds, golden ratio` - Focus: `centered, focus on subject` - Depth: `depth of field, bokeh background` ## Troubleshooting **"AI isn't following my prompt":** - Make important parts stronger with () - Remove conflicting descriptions - Be more specific **"Results look generic":** - Add unique details - Use style descriptors - Include mood/atmosphere words **"Wrong focus/subject":** - Put main subject first - Use weights to emphasize - Add "focus on [subject]" ## Pro Tips - Order matters! Put important stuff first - Use underscores for names: `hatsune_miku` - Combine with negative prompt for best control - Test same seed with different prompts - Build a library of prompts that work ## Quick Formula ``` [Subject] + [Appearance] + [Action/Pose] + [Location/Background] + [Style] + [Quality] ``` Example: ``` 1girl + silver hair, red eyes + standing + moonlit forest + fantasy art + masterpiece ``` ## Example Workflows **Testing Prompts:** 1. Write your base prompt 2. Set seed to fixed number 3. Try variations 4. See what works best **Building Complexity:** 1. Start simple: "1girl, smile" 2. Add details: "blue hair, school uniform" 3. Add scene: "classroom, sunset" 4. Add quality: "masterpiece, detailed" **Style Exploration:** 1. Keep subject same 2. Change style tags 3. Compare results 4. Find your favorite --- For avoiding things: [IllustriousNegativeCLIPEncoder](IllustriousNegativeCLIPEncoder.md) For prompt building: [IllustriousPrompt](IllustriousPrompt.md)