Vantage-Nodes
This release introduces Vantage Nodes, a powerful utility and infrastructure node pack for ComfyUI, designed for advanced workflows, batch processing, and experimental pipelines.
Nodes (31)
Do real math inside the graph, not in your head
Stitch image batches together, no resize magic
Grow a latent batch one chunk at a time — the accumulator node
One input in, three typed outputs out — stop fighting ComfyUI's type system
Conditional Pass Through
Turn a comma-separated string into a proper multiline list
The tiny node that ends off-by-one loops for good
Turn a batch of images into one contact-sheet image
Stitch two latent batches into one — safely, without resizing a single pixel
One node that swaps prompts and LoRAs per scene
The Qwen TTS model grabber that makes the rest of the pack work
Write a script, tag the speakers, get a finished voiceover with music
Reuse saved voices without rebuilding them every time
Clone a voice from three seconds of audio, no API key required
Design a voice with a sentence instead of hunting for a sample
Filter a multiline list with a regex — keep only the lines that match
Collapse a multiline list back into one string with a delimiter
Pull one line out of a multiline list by number — and get the count while you're at it
Pick a random line from your prompt list — seeded, so you can rerun it
String List Stepper for per-frame text
First non-None wins — the fallback node that works on any data type
Switch Any (By Index) explained
Pick one image out of four with an integer — the routing glue for image workflows
Choose one latent batch from four with a number — the latent router
Check two batches match before the cat() blows up
Load a quantized UNet GGUF straight from your diffusion_models folder
A boolean gate for image batches — pass the images through or hand back nothing
Load a whole folder of images, batch-resized, background removed, in one node
Encode prompts for Krea 2 — with reference-image fusion and a prompt-adherence booster
Take a batch apart for per-frame processing
Run a 40-block Wan video model on a smaller card by swapping blocks to RAM
Vantage Nodes for ComfyUI
A curated collection of high-quality utility and infrastructure nodes for ComfyUI, focused on flexibility, batch workflows, video pipelines, and advanced control logic.
This repository is designed as a practical standard library for ComfyUI power users, workflow authors, and extension developers.
✨ Highlights
- 🔁 Advanced switching & fallback control nodes
- 🧮 Powerful expression-based calculator
- 🧵 Robust batch utilities for IMAGE and LATENT
- 🖼️ Smart image batch merging with resize / pad / crop
- 🎥 WAN Video block swapping (WanVideoWrapper–compatible)
- 🧊 GGUF UNet loader for GGUF-based diffusion workflows
- 🧩 Clean categories under
Vantage / … - 🛠️ Production-safe, well-tested, and extensible
📂 Installation
Clone into your ComfyUI custom_nodes directory:
cd ComfyUI/custom_nodes
git clone https://github.com/vantagewithai/Vantage-Nodes.git
pip install -r requirements.txt
Restart ComfyUI after installation.
🗂️ Node Categories
All nodes are grouped under:
Vantage /
├─ Types
├─ Control
├─ String
├─ Math
├─ Image
├─ Latent
├─ Video
└─ Loaders
🧮 Math & Logic Nodes
Advanced Calculator
Category: Vantage / Math
- Expression-based calculator with:
- Variables
A–E(optional, auto-cast INT / FLOAT / STRING) - Multiline expressions
- Case-insensitive variables and functions
- Variables
- Supported functions:
ceil, floor, round, abs min, max pow, sqrt frac(x) # fractional part
Example:
ceil(
(A + B) * C
/ max(D, 1)
) + frac(E)
Switch Any
Category: Vantage / Control
Returns the first non-None input from up to 10 optional ANY inputs.
Perfect for:
- Fallback values
- Optional branches
- Conditional pipelines
Switch Any (By Index)
Category: Vantage / Control
Selects one of multiple ANY inputs using an integer index (0-based).
🧵 String Utilities
- Multiline string indexer
- Stepper (next / previous)
- Random picker (seeded)
- Joiner
- Regex filter
- Delimiter / CSV splitter
Designed for:
- Prompt lists
- Batch prompts
- LLM outputs
- Procedural text workflows
🖼️ Image Nodes
Join Image Batch
Category: Vantage / Image
Joins two IMAGE batches into one.
Supports advanced resolution handling:
- Resize modes:
none(strict)resizepad(aspect-preserving)crop(center crop)
- Reference resolution:
- A or B
- Interpolation:
- nearest / bilinear / bicubic
Safely handles:
- Batch images from VAE Decode
- Mixed resolutions
- Optional inputs
Append Image Batch
Accumulate IMAGE batches iteratively.
Switch Image (By Index)
Select one IMAGE input by index.
🧊 Latent Nodes
Join Latent Batch
Strict batch join for LATENT tensors.
- Preserves metadata
- Enforces shape safety
- No spatial resizing (latent-safe)
Append Latent Batch
Append LATENT batches for iterative workflows.
Switch Latent (By Index)
Index-based LATENT selector.
🎥 Video Nodes
VantageWanBlockSwap
Category: Vantage / Video
A WAN Video Block Swapper, based on:
- ComfyUI-WanVideoWrapper
Allows advanced block-level swapping / routing inside WAN video models.
Use cases:
- Video style transfer
- Hybrid motion pipelines
- Experimental WAN architectures
🧊 Loaders
VantageGGUFLoader
Category: Vantage / Loaders
A UNet GGUF loader, based on:
- ComfyUI-GGUF
Features:
- Load GGUF-based UNet models
- Integrates with standard ComfyUI pipelines
- Designed for memory-efficient and experimental setups
🧠 Design Philosophy
- ✅ Explicit > implicit
- ✅ Safe defaults
- ✅ Optional inputs everywhere possible
- ✅ No silent failures
- ✅ Batch-first mindset
- ✅ Consistent naming & categories
This repo is intentionally utility-focused, not UI-heavy.
🛠️ Development Notes
- All nodes are defined using native ComfyUI APIs
- No frontend modifications required
- Dropdowns implemented using list-based enums
- GPU-safe torch operations
- Compatible with standard and batch pipelines
📜 Credits
This project builds upon ideas and implementations from:
- ComfyUI
- ComfyUI-WanVideoWrapper
- ComfyUI-GGUF
All credit for foundational work goes to their respective authors.
📄 License
MIT License
🚀 Roadmap (Planned)
- IMAGE / LATENT batch split
- Video batch utilities
- Auto aspect-ratio tools
- Batch inspectors & debuggers
- More loader abstractions
Credits
Krea 2-related parts of this repository were developed with help and reference from the following projects:
Special thanks to their authors for their work and shared ideas that helped support the Krea 2 implementation here.
The Multi Image Loader node was also developed with help and reference from:
- WhatDreamsCost-ComfyUI for the multi-image loading and gallery-based workflow ideas [web:23][web:250]
- ComfyUI-RMBG for background removal workflows, model handling, and related RMBG integration ideas [web:246][web:251]
Special thanks to those authors as well for the work that helped shape the Multi Image Loader functionality in this project.
🤝 Contributing
Contributions, issues, and feature requests are welcome.
If you use these nodes in production or complex workflows, feedback is highly appreciated.