25 Best ComfyUI Custom Nodes You Need to Install in 2025

ComfyUI's power comes from its custom nodes. The base installation handles basics, but custom nodes unlock advanced workflows, better quality, and massive time savings. After testing hundreds of nodes, here are the 25 you actually need.

Quick Answer: Essential custom nodes include ComfyUI Manager (for installing everything else), Impact Pack (face fixing, segmentation), WAS Node Suite (utilities), Efficiency Nodes (workflow optimization), and model-specific nodes for whatever you're generating. Install ComfyUI Manager first, then add nodes as your workflows demand them.

Must-Have Foundation Nodes

1. ComfyUI Manager

What it does: Installs, updates, and manages all other custom nodes through a convenient UI.

Why essential: Without this, you're manually cloning git repositories. ComfyUI Manager makes installation one-click.

Install first: This enables everything else.

Features:

• Browse and install nodes from UI
• Update all nodes at once
• Check for missing nodes in workflows
• Disable problematic nodes without deleting

2. Impact Pack

What it does: Face detection, segmentation, detail enhancement, and batch processing.

Why essential: The FaceDetailer alone makes this mandatory. Automatically detects and enhances faces in generations.

Key nodes:

• FaceDetailer (fixes faces automatically)
• SAM Segmentation (precise masking)
• Detailer pipes (workflow organization)
• Batch processing utilities

Use case: Every portrait workflow benefits from FaceDetailer.

3. WAS Node Suite

What it does: Comprehensive utility collection with 100+ nodes for various tasks.

Why essential: Fills gaps in base ComfyUI. Text manipulation, image operations, math, conditionals.

Key nodes:

• Image operations (resize, crop, blend)
• Text processing
• Number/math operations
• File operations
• Debugging utilities

Use case: You'll use something from WAS in almost every complex workflow.

4. ComfyUI Essentials

What it does: Additional fundamental operations missing from base installation.

Why essential: Complements base nodes with better alternatives and missing functions.

Key nodes:

• Better image operations
• Mask utilities
• Batch processing
• Quality of life improvements

5. rgthree's Nodes

What it does: Workflow organization and power user features.

Why essential: Makes complex workflows manageable with better organization tools.

Key nodes:

• Seed everywhere (consistent seeds across workflow)
• Context nodes (pass multiple values together)
• Power LoRA Loader (better LoRA management)
• Bookmark nodes (navigate large workflows)

Use case: Any workflow with more than 20 nodes.

Quality Enhancement Nodes

6. Ultimate SD Upscale

What it does: Tiled upscaling with ControlNet for high-quality enlargement.

Why essential: Standard upscaling loses detail. This maintains quality at 4x+ scales.

How it works: Breaks image into tiles, upscales each with AI, blends seamlessly.

Use case: Taking 1024px generations to 4K+.

7. ControlNet Auxiliary Preprocessors

What it does: Prepares images for ControlNet (edge detection, depth maps, pose estimation).

Why essential: ControlNet requires preprocessed images. This handles all preprocessing automatically.

Key preprocessors:

• Canny edge detection
• Depth estimation (MiDaS, Zoe)
• OpenPose for body/hand pose
• Lineart extraction
• Segmentation

8. IPAdapter Plus

What it does: Image prompting, style transfer, face swapping, reference-based generation.

Why essential: The cornerstone of character consistency workflows.

Key features:

• Reference image influence
• FaceID for identity preservation
• Style reference without LoRA training
• Multiple reference combining

Use case: AI influencer creation, character consistency, style matching.

9. ReActor

What it does: Face swapping in generated images.

Why essential: When you need a specific face without LoRA training.

Features:

• Single face swap
• Multi-face handling
• Quality options
• Batch processing

10. ComfyUI FaceRestore

What it does: Dedicated face restoration using CodeFormer and GFPGAN.

Why essential: Better face restoration than generic upscalers.

Models supported:

• CodeFormer (best quality)
• GFPGAN (faster)
• RestoreFormer

Video Generation Nodes

11. ComfyUI-VideoHelperSuite

What it does: Video loading, saving, and basic manipulation.

Why essential: Core video workflow support. Load videos, save outputs, handle frames.

Key nodes:

• Load Video
• Video Combine (frames to video)
• Load Images from Directory
• Video Info

12. ComfyUI-LTXVideo

What it does: Full LTX-2 support including text-to-video, image-to-video, and 4K upscaling.

Why essential: The only way to use LTX-2 in ComfyUI.

Key features:

• LTX-2 model loading
• 4K spatial upscaler
• Multi-keyframe support
• Audio generation

See: LTX-2 complete guide

13. ComfyUI-WanVideoWrapper

What it does: Wan 2.2 video model support.

Why essential: Use Wan video models for image-to-video and text-to-video.

See: Wan 2.2 workflow guide

14. AnimateDiff Evolved

What it does: AnimateDiff motion modules for video generation from static images.

Why essential: Alternative video generation approach, works with SDXL models.

15. ComfyUI Frame Interpolation

What it does: Increases frame rate by generating intermediate frames.

Why essential: Smooth video from low-fps generations.

Models:

• RIFE (fast, good quality)
• FILM (slower, better quality)

Efficiency and Workflow Nodes

16. Efficiency Nodes

What it does: Optimized sampling, batch generation, XY plotting.

Why essential: Faster workflows, easier parameter testing.

Key nodes:

• Efficient KSampler (optimized sampling)
• XY Plot (test parameter ranges)
• Batch processing

17. ComfyUI-KJNodes

What it does: Collection of utility nodes for common operations.

Why essential: Quality of life improvements throughout workflows.

Key nodes:

• Get/Set nodes (variable management)
• Condition combining
• Image utilities
• String operations

18. ComfyUI Easy Use

What it does: Simplified versions of complex operations.

Why essential: Reduce node count for common tasks.

Simplifies:

• LoRA loading
• Model switching
• Common workflows

19. ComfyUI Noise

What it does: Advanced noise generation and manipulation.

Why essential: Better control over generation randomness.

Features:

• Custom noise patterns
• Noise scheduling
• Seed manipulation

Specialized Purpose Nodes

20. ComfyUI Inpaint Nodes

What it does: Advanced inpainting with various approaches.

Why essential: Better inpainting than base implementation.

Methods:

• MAT inpainting
• LaMa inpainting
• Differential diffusion

21. ComfyUI Segment Anything

What it does: SAM (Segment Anything Model) for precise automatic masking.

Why essential: Select any object with a click, perfect masks automatically.

Use cases:

• Background removal
• Object isolation
• Selective editing

22. ComfyUI GGUF

What it does: Load GGUF quantized models for reduced VRAM usage.

Why essential: Run larger models on limited hardware.

Supports:

• Flux GGUF versions
• Quantized checkpoints
• Memory optimization

23. ComfyUI-Crystools

What it does: Metadata handling, workflow management, debugging.

Why essential: Save generation metadata, debug workflows.

Features:

• Metadata embedding
• Workflow comparison
• Resource monitoring

24. ComfyUI LoRA Power Tool

What it does: Advanced LoRA management and combination.

Why essential: Better control when stacking multiple LoRAs.

Features:

• Per-block LoRA weights
• LoRA comparison
• Merge visualization

25. ComfyUI Prompt Control

What it does: Advanced prompt scheduling and manipulation.

Why essential: Change prompts during generation, create complex prompt sequences.

Features:

• Prompt scheduling
• Prompt interpolation
• Dynamic prompt changes

Installation Guide

Using ComfyUI Manager (Recommended)

1. Install ComfyUI Manager first (git clone into custom_nodes)
2. Restart ComfyUI
3. Click "Manager" in the menu
4. Search for desired node pack
5. Click Install
6. Restart ComfyUI

Manual Installation

cd ComfyUI/custom_nodes
git clone [repository-url]
pip install -r requirements.txt  # if present
# Restart ComfyUI

Handling Missing Nodes

When loading a workflow with missing nodes:

1. ComfyUI Manager shows "Missing Nodes" notification
2. Click to see which nodes are needed
3. Install missing nodes
4. Restart and reload workflow

Avoiding Common Issues

Too Many Nodes

Problem: Startup becomes slow, memory usage increases.

Solution: Only install what you actively use. Disable unused nodes in Manager.

Conflicting Nodes

Problem: Some nodes override others or cause errors.

Solution: Check for conflicts in Manager. Disable conflicting node sets.

Outdated Nodes

Problem: Nodes break after ComfyUI updates.

Solution: Use Manager's "Update All" regularly. Check node GitHub for compatibility.

Missing Dependencies

Problem: Node loads but fails at runtime.

Solution: Check requirements.txt in node folder. Install missing packages manually.

My Personal Node Stack

For reference, here's what I keep installed for daily use:

Always loaded:

• ComfyUI Manager
• Impact Pack
• WAS Node Suite
• rgthree's Nodes
• IPAdapter Plus
• ControlNet Preprocessors

For video work:

• VideoHelperSuite
• LTXVideo nodes
• WanVideoWrapper
• Frame Interpolation

For quality:

• Ultimate SD Upscale
• FaceRestore

Specialized (enabled when needed):

• Segment Anything
• ReActor
• GGUF support

This gives maximum capability without excessive overhead.

Frequently Asked Questions

How many nodes is too many?

When startup exceeds 30 seconds or you're using less than half your installed nodes, it's time to clean up.

Do nodes work across ComfyUI versions?

Usually yes, but major ComfyUI updates can break nodes. Update nodes after updating ComfyUI.

Can I create my own nodes?

Yes. ComfyUI nodes are Python. See the developer documentation for creating custom nodes.

Which nodes are most important for AI influencer work?

IPAdapter Plus, Impact Pack (FaceDetailer), and video nodes (LTX-2, Wan).

Do nodes slow down generation?

Unused nodes don't affect generation speed, only startup time and memory usage.

Wrapping Up

Custom nodes transform ComfyUI from capable to exceptional. Start with the foundation nodes, add quality enhancers, then expand based on your specific needs.

Essential first installs:

1. ComfyUI Manager
2. Impact Pack
3. WAS Node Suite
4. ControlNet Preprocessors
5. IPAdapter Plus

These five handle 80% of workflow needs. Add specialized nodes as your work demands them.

For workflows using these nodes, explore our ComfyUI tutorials. For generation without local setup, Apatero.com provides powerful generation without installing anything.

Quick Reference: Node Categories

Bookmark this list for reference when building new workflows.

Node Installation Priority

If you're just starting, install nodes in this order:

Day 1: ComfyUI Manager, Impact Pack Week 1: WAS Suite, ControlNet Preprocessors, IPAdapter Plus As needed: Video nodes, specialized tools

This staged approach prevents overwhelm while ensuring you have essentials immediately.

Remember: a clean installation with carefully chosen nodes outperforms a cluttered setup with everything installed. Quality over quantity applies to custom nodes too.

Troubleshooting Common Node Issues

Node not appearing after install:

• Restart ComfyUI completely (not just refresh)
• Check for error messages in terminal
• Verify the node appears in Manager's installed list

Node causing crashes:

• Disable the node in Manager
• Check GitHub issues for known problems
• Try reinstalling after ComfyUI update

Workflow says "missing nodes":

• Use Manager's "Install Missing" feature
• Some nodes have been renamed or deprecated
• Check if node pack changed names

Slow startup with many nodes:

• Disable unused node packs
• Remove experimental nodes you're not using
• Consider multiple ComfyUI installations for different purposes

These issues are common and usually have simple solutions. The ComfyUI Discord is excellent for troubleshooting persistent problems.