Video RGB Analysis (QQ)

How do you tell if your generated video is color-consistent across all its frames? Scrubbing through it, honestly. Your eyes will miss a slow drift or a per-frame flicker that a graph would show in one glance. VideoRGBAnalysis is that graph: it averages each frame's red, green, and blue channels, computes overall brightness, and plots all of it against the frame number in the node's UI. It's a scopes panel for your Wan output, and it turns "something looks off" into "the blue channel dips at frame 40."

It's from siraxe/ComfyUI-WanVideoWrapper_QQ, the "(QQ)" Wan pack - a QA utility that earns its place the first time you debug a flickering render.

How it works

For each frame it computes the mean of each color channel across the spatial dimensions, giving one R, G, B value per frame - a compact time series of the whole clip's color. Brightness is calculated with the Rec. 601 luminance weights (0.299*R + 0.587*G + 0.114*B), which approximates how bright a frame actually looks to a human rather than a naive average. The results are rendered as a multi-line graph in the node's UI: R, G, B, and brightness over time.

The display_mode dropdown changes what you're looking at:

• intensity (default) - the raw channel means and luminance.
• saturation - each channel normalized by that frame's brightest channel, showing how vivid each color is relative to the brightest one. Better for spotting hue-shift or washed-out sections than raw intensity.

It's an output node (no data outputs), so it sits at the end of your chain or on a parallel branch purely for inspection - run it on the same frames your video saver gets, and read the graph.

Inputs and outputs that matter

• image - your IMAGE tensor (frames first: [B, H, W, C]).
• display_mode - intensity or saturation.

That's the whole interface. No outputs to wire - the result is the graph in the UI.

Installation

cd ComfyUI/custom_nodes
git clone https://github.com/siraxe/ComfyUI-WanVideoWrapper_QQ.git

Restart ComfyUI (or ComfyUI Manager → "WanVideoWrapper_QQ"). No extra Python deps - pure torch; the pack's requirements.txt is empty. After the v1.3.4 rename to ComfyUI-SA-Nodes-QQ, delete any old wanwrapper_qq folder.

Common issues

There's not much to trip on here, but two things are worth knowing. First, it's a per-frame mean, so it won't show you spatial problems - a bright artifact in one corner averages away; this node is for temporal consistency, not composition review. Second, because it's an output node, forgetting where you put it in the graph means it silently shows you only whatever frames you fed it - if you wire it before a crop or resize, you're analyzing the wrong signal. Drop it on the same data path as your final saver and you're golden.