Video User Inputs Basic

Some workflows need five inputs wired to five places. Some need two. TKVideoUserInputsBasic is the two-input version of the pack's resolution GUI - width and height in, width and height out, nothing else. It's the "Common Video User Inputs - Basic" variant, and if you only need to feed an empty latent and a sampler the same dimensions without dragging the full five-output node onto the canvas, this is the one you want.

It exists because its big sibling, TKVideoUserInputs, is opinionated. That node forces you to pick seconds-or-frames and gives you fps and duration math whether you need it or not. Here, none of that - the author cut it down to the minimum that a video workflow actually needs when all you're doing is declaring a canvas size.

How it works

Honestly, it's a pass-through. Two integer widgets, two integer outputs, and the Python function returns exactly what you typed. The value, such as it is, comes from two places. First, the widgets snap to 32px steps (1280×1280 by default, range 100–1288), which keeps you on the resolution grid LTX wants - dimensions divisible by 32 - without you having to remember that rule. Second, like the full node, it ships with the same JS canvas UI, so you get the draggable resolution picker with the dot grid instead of typing numbers into a box.

The pair of outputs - video_width and video_height - wire into whatever consumes your canvas size. Because both are INT sockets, they feed an Empty Latent Image's width and height, a sampler, a resolution-conditional group, anything.

When to use it

Reach for this when resolution is the only thing you want to centralize, or when you're building a modular workflow and don't want the full five-value node cluttering the graph. If you also need frame count or duration, step up to the full TKVideoUserInputs - this one deliberately doesn't compute anything for you.

Install

Same pack, same drill. ComfyUI Manager, search "Handy Nodes", install, restart. Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/trashkollector/TKNodes

Restart, and remember the same frontend caveat as its sibling: the canvas GUI needs a full restart to register, and if the js/ directory didn't survive the install you'll just see plain number boxes - which honestly still work fine, since the node's logic is a pass-through either way. No models, no dependencies beyond the pack's standard requirements. It's the node you install once and forget exists, until the day a workflow is missing it and you remember how nice not-typing-32-multiples was.