🐯 YFG Histograms Generator (compact)

The full Histograms Generator in this pack spits out seven outputs - every channel, filled and lined, plus luminosity. That's great for a deep dive and a lot of node to leave lying around. The compact version is the one you'll actually keep in your daily workflow: it computes a single histogram type, shows it inline, and hands you two clean outputs.

Same engine as the big sibling - NumPy bins the pixels, matplotlib draws the chart - just trimmed down. "YFG Histograms Generator (compact)" from gonzalu/ComfyUI_YFG_Comical.

The inputs

• histogram_type - pick one of RGB Histogram Filled, RGB Histogram Lines, Red Channel, Green Channel, Blue Channel, or Luminosity. The filled RGB one is the best general-purpose "does this look washed out" check; Luminosity is the exposure read.
• histogram_size (small / medium / large) - chart render resolution via DPI.
• preview - when on, the node displays the chart (or original, per display) directly on the node, so you don't even need a preview node. This is the compact part doing its job.
• display - choose whether the self-preview shows the histogram or the original image.

Outputs

Only two, which is the point:

• Original Image (IMAGE) - input passed through, so you can save a before/after pair from one node.
• Histogram (IMAGE) - the computed chart, ready to feed a SaveImage or another node.

Installing

Pack install: ComfyUI Manager → search "YFG Comical", or

cd ComfyUI/custom_nodes
git clone https://github.com/gonzalu/ComfyUI_YFG_Comical

Restart ComfyUI. Watch the pack-wide dependency trap here: this file imports cv2 and matplotlib at the top with neither declared in pyproject.toml. If the whole pack dies on import, pip install opencv-python matplotlib and restart.

Why you'd pick this over the full version

The full node is a diagnostic lab; this is the instrument. If you're auditing contrast before committing to an upscale, or you want a permanent "am I clipping?" check hanging off your final image, the compact version gives you the signal with half the wires. Switch histogram_type when you need a channel-specific look, and you've got 90% of the full node's usefulness in a fraction of the graph space.