⌗ Frame Set Frame Dimensions Scaled

Frame Set Frame Dimensions Scaled [DVB] is a one-trick node with a very specific job: it reads the width and height of a frame set and hands them back to you multiplied by a factor. That's the entire mechanism - round(width * factor) and round(height * factor) out as two INTs. It doesn't resize anything, it doesn't crop anything, it just answers the question "what would these dimensions be at this scale?" fast and in integers.

Why that's useful enough to ship

The README calls it out for one thing: calculating intermediate step sizes. The camera nodes in this pack - the pans, zooms, and rolls - work by cropping to a smaller frame than the source, and their outputs are always smaller than their inputs. To build a stepped zoom or a sequence of crop sizes, you need to compute what "80% of the current frame" actually is in pixels, and that's exactly what this node does. It's the glue between a frame set and a resize or crop elsewhere in the graph.

It also quietly answers a question you'll ask constantly while composing: "what resolution is this set, and what would half of it be?" Instead of hunting for the dimensions elsewhere, drop this node in, set factor to 0.5, and both numbers pop out as ready-to-wire INTs. The factor input is a float with a 0.01 minimum, so scaling down to 1% is allowed - though realistically you're using it between roughly 0.25 and 4.

The honest framing

It's a utility, not a feature - but utility nodes are the connective tissue of a working graph. Pair it with the pack's camera nodes or a standard Resize node and you get a workflow that stays correct when you change the source resolution, because the scaled dimensions are derived from the actual frame set rather than typed in by hand. That's the quiet win: fewer magic numbers, more self-adjusting graph.

The input side is frames plus factor; the outputs are width and height as integers. Note the rounding - because both outputs are INTs, a 0.333 factor on a 641-wide set gives you 213, not 213.453. If you need the precise float for anything, keep that in mind; for sizing and cropping, rounded is what you want anyway.

Install

cd ComfyUI/custom_nodes
git clone https://github.com/alt-key-project/comfyui-dream-video-batches.git
cd comfyui-dream-video-batches
pip install -r requirements.txt

Restart ComfyUI, or search "Dream Video Batches" in ComfyUI Manager. No model downloads - this is a multiplication on a pair of ints. One thing to remember for the whole pack: frames in means FRAME_SET currency, so the set you feed here must have been created by Create Frame Set [DVB] or another DVB node, and anything you take out of the pack still needs Unwrap Frame Set [DVB] before standard nodes will accept it.