comfyui-alabo-resolution

comfyui-alabo-resolution

One node to drive resolution, clip length, and frame rate across a ComfyUI video graph from a single control. Built for SD1.5 and AnimateDiff pipelines where the empty latent, the IPAdapter frame count, and the video frame rate all have to agree, and usually do not.

Why it exists

SD1.5 and AnimateDiff render best near their training resolution (around a 512 short edge). Push native resolution to 1024 and the model repeats features and runs out of memory on a 16GB card. The reliable pattern is render low, then upscale as a separate final pass. This node encodes that discipline so the numbers are set once and stay consistent everywhere downstream.

Nodes

• Alabo Resolution Preset. Aspect and scale and seconds go in, width and height and latent frame count and fps come out. Emits a notes string so the chosen values are visible on the canvas.
• Alabo Quality Preset (draft / high). A single switch for a fast draft pass or a full-quality pass.

What it emits

width, height, frames, fps, notes.

The short edge follows orientation: portrait sizes the width, landscape sizes the height, square sizes both. Every dimension snaps to a multiple of 16. The latent frame count is computed as round(seconds * fps / rife_multiplier), so the final clip is exactly seconds long once RIFE multiplies the frames back up and the video node plays them at fps.

Aspects

16:9, 9:16, 3:4, 4:3, 1.37:1 Academy, 2.35:1 scope, 1:1, 4:5, 21:9 ultrawide.

Install

Clone into ComfyUI/custom_nodes and restart ComfyUI.

cd ComfyUI/custom_nodes
git clone https://github.com/AlaboMPJ/comfyui-alabo-resolution

Usage

Right-click each downstream widget and Convert Widget to Input first, then wire:

• width into EmptyLatentImage.width
• height into EmptyLatentImage.height
• frames into IPAdapterWeights.frames
• fps into VHS_VideoCombine.frame_rate

Set rife_multiplier to match your RIFE node so the clip length stays honest through interpolation.

License

MIT. </content>