Files
regiellis-ComfyUI-EasyIllus…/web/docs/IllustriousNegativeCLIPEncoder.md
Regi Ellis e915afb65f chore(ci, docs): add publish workflow, mark Beta, and document Manager install
- Add .github/workflows/publish_action.yml for Comfy registry

- Update README: ComfyUI Manager install (recommended), Beta badge + warning

- Minor docs tweaks
2025-09-07 22:51:37 -04:00

3.0 KiB

IllustriousNegativeCLIPEncoder

Tells the AI what you DON'T want in your image. Like saying "no pickles" on your burger order!

What it does

  • Takes your list of things to avoid and makes the AI understand them
  • Prevents common problems like extra fingers, bad anatomy, or ugly colors
  • Works together with your positive prompt for complete control
  • Essential for getting clean, professional results

Inputs

  • text: What you want to avoid (bad hands, blurry, etc.)
  • clip: Text understanding model (usually auto is fine)
  • normalize: Balances the strength (keep ON for stability)

Outputs

  • CONDITIONING: Instructions for what to avoid, ready for sampler

Why You Need This

Think of it like quality control:

  • Without negatives: AI might add weird artifacts
  • With negatives: Clean, professional artwork
  • Too many negatives: Confused AI, weird results

The Golden Negative Prompts

For Anime/Illustration:

low quality, bad anatomy, bad hands, text, error, 
missing fingers, extra digit, fewer digits, cropped, 
worst quality, jpeg artifacts, signature, watermark

For Realistic:

deformed, ugly, mutilated, disfigured, text, extra limbs, 
face cut, head cut, extra fingers, poorly drawn hands, 
mutation, bad proportions, cropped head, malformed limbs

For Clean Art:

blurry, lowres, bad anatomy, bad hands, error, 
missing fingers, extra digit, cropped, worst quality, 
low quality, normal quality, jpeg artifacts

Common Negatives Explained

  • "bad hands": Fixes finger problems
  • "text/watermark": Removes unwanted text
  • "low quality": Improves overall quality
  • "blurry": Sharper images
  • "jpeg artifacts": Cleaner output
  • "extra fingers": Correct anatomy
  • "cropped": Full composition

When to Add More

Getting specific problems?

  • Add that specific issue to negatives
  • Example: Seeing doubles? Add "duplicate"
  • Wrong style? Add "realistic" (for anime) or vice versa

When to Use Less

Image looking bland?

  • You might be over-restricting
  • Remove some negatives
  • Keep only essential ones

Pro Tips

  • Start with basic negative template
  • Add specific issues as they appear
  • Less is often more
  • Some models have recommended negatives
  • Save your perfect negative as preset

Common Mistakes to Avoid

❌ Don't do this:

  • "not ugly, not bad, not horrible" (redundant)
  • Super long lists (confuses AI)
  • Contradicting your positive prompt

✅ Do this:

  • Short, specific negatives
  • Target actual problems you see
  • Use proven templates

Example Workflows

Basic Setup:

  1. Copy a golden negative prompt
  2. Connect to sampler's negative input
  3. Generate cleaner art instantly

Problem Solving:

  1. See a specific issue?
  2. Add it to negatives
  3. Regenerate - problem solved!

Style Control:

  1. Don't want realistic in anime?
  2. Add "realistic, photo" to negatives
  3. Get pure anime style

For positive prompts: IllustriousCLIPTextEncoder For complete prompting: IllustriousPrompt