- Add .github/workflows/publish_action.yml for Comfy registry - Update README: ComfyUI Manager install (recommended), Beta badge + warning - Minor docs tweaks
2.6 KiB
2.6 KiB
IllustriousSmartCacheCorrector
Fixes cache issues and speeds up your workflow. When things get stuck or slow, this is your reset button!
What does it do?
- Detects and fixes stale cached data that's slowing you down
- Refreshes specific parts without clearing everything
- Optimizes memory usage during long sessions
- Prevents the "why isn't my prompt changing?" problem
- Auto-clears cache when switching between projects
Inputs
-
cache_mode: How aggressive to be
- Smart: Only clear what's changed
- Selective: Clear specific items
- Full: Complete refresh (nuclear option)
- Auto: Let it decide based on usage
-
target: What to clear
- Prompts: Text/conditioning cache
- Images: Generated image cache
- Models: Model state cache
- All: Everything
-
threshold_mb: Clear when cache exceeds this size
-
age_minutes: Clear items older than this
-
force_refresh: Bypass smart detection
Outputs
- status: What was cleared and why
- cache_size: Current cache usage
- performance_gain: Estimated speedup
Artist Tips
- Run this when generations feel sluggish
- Use Smart mode for everyday clearing
- Full clear when switching projects
- Set threshold to 50% of your VRAM
- Age-based clearing good for long sessions
When to Use
Prompt not updating?
- Target: Prompts
- Mode: Selective
- Your changes will finally show!
Out of memory warnings?
- Target: All
- Mode: Full
- Frees up VRAM instantly
Workflow getting slow?
- Mode: Smart
- Threshold: 2048 MB
- Automatic optimization
Switching projects?
- Mode: Full
- Clean slate for new work
Common Issues & Fixes
"Cache keeps refilling":
- Lower threshold_mb
- Enable age-based clearing
- Check for memory leaks
"Lost my cached prompts":
- Use Selective mode
- Only target specific types
- Smart mode preserves recent items
"No performance improvement":
- Try Full clear
- Restart ComfyUI if needed
- Check system memory too
Pro Tips
- Schedule clears between batches
- Monitor cache_size output
- Use before long generations
- Combine with prompt cache nodes
- Clear models when switching styles
Example Workflows
Regular Maintenance:
- Every 10 generations
- Run Smart clear
- Continue working smoothly
Project Switch:
- Finish current project
- Full cache clear
- Load new project
- Fresh, fast performance
Memory Recovery:
- Getting VRAM warnings
- Target: Images + Models
- Mode: Full
- Instantly recover memory
For caching prompts: IllustriousSmartPromptCache For performance: IllustriousKSamplerPro