Academia SD Resolution Display 📐

Video workflows have a way of lying to you. The sampler runs at one resolution, the combine node spits out another, and the upscale step changes things again - and nowhere on the canvas does it say what's actually happening. This node is a tiny on-canvas readout: feed it the input video dimensions and the output video dimensions, and it displays them both on its face, so the discrepancy is visible at a glance instead of hiding in three different node settings.

How it works

It's a pure display node - four required inputs, all INTs: input_video_width, input_video_height, output_video_width, output_video_height. Wire them to wherever the numbers actually live in your workflow (sampler outputs, VAE decodes, resolution selectors), and the node renders a small text panel in its body:

Input Video: 768 x 512
Output Video: 1024 x 1024

There are no outputs and nothing is computed - the inputs are just shown, which makes it an output node you read rather than connect. The practical use is catching mismatches: if your input is 768x512 but your output target is 1024x1024, you'll see the aspect-ratio stretch staring back at you before you waste a queue on it.

Why it's useful for video

Aspect ratio is where video workflows quietly go wrong. LTX and other models have native resolutions, and feeding a stretched or oddly-sized latent in produces subtly off results that are hard to diagnose from the final clip alone. Having both sizes on one node face turns an invisible bug into a one-second visual check. It's also genuinely handy when you're iterating on upscaling setups and want to confirm the final dimensions changed without digging into three nodes.

Install

Pack install: ComfyUI Manager → "AcademiaSD", or git clone https://github.com/AcademiaSD/comfyui_AcademiaSD, restart. No extra dependencies.

The honest take

This is a pure convenience, and it's the thinnest node in the pack - there's no math, no history, no cleverness. If you already have a PreviewText or a log node in your toolbox, it does the same job with more flexibility. But it costs nothing, takes up almost no canvas space, and for the specific "input vs output video size" question it's the most direct answer you'll find. Wire it in while you're debugging a video pipeline and delete it when you're done - that's its entire lifecycle, and it does that well.