Anymatix Image To Video

Plenty of ComfyUI workflows produce a video not by sampling frames inside one big model but by generating a batch of images - an AnimateDiff pass, an image-interpolation chain, or a set of frames you rendered one at a time - and then needing to treat that batch as an actual video for saving or further processing. AnymatixImageToVideo is the node that performs that hand-off: it takes an IMAGE batch and an fps value, optionally an AUDIO input, and returns a proper VIDEO object.

The important thing to understand is what it is not. It doesn't encode anything to an MP4 - there's no FFmpeg pass and no pixels written to disk here. It packages your batch of frames and your timing into ComfyUI's standard VideoComponents container, which the newer ComfyUI video tooling uses to carry "these images, at this frame rate, maybe with this soundtrack" around the graph. The real encode happens later, when a save node such as AnymatixSaveAnimatedMP4 or the core video-save nodes turn that VIDEO object into a file. That's also why it's described as "zero-cost" in the code: wrapping references costs you nothing, and the expensive step waits until you actually save.

Where it fits

This is the natural tail of an image-to-video pipeline. If your workflow generates frames and you want to (a) preview them as video, (b) mux in an audio track so a talking or singing result has its sound, or (c) hand the result to a save node, this is the bridge. The optional audio input is the part that makes it more than a repackaging: attach an AUDIO object and your video carries it to the muxer, so the soundtrack comes out of the same save step instead of a separate edit.

Inputs and output

• images (IMAGE) - a batch of frames, shape batch × height × width × channels.
• fps (FLOAT, default 6) - how fast the frames should play. ComfyUI's own animated outputs often default around 6-8; 24 is a common choice if you generated true video frames.
• audio (optional AUDIO) - attach a soundtrack.
• Output: video (VIDEO) → a save node like AnymatixSaveAnimatedMP4 or anything that consumes VIDEO.

Install

The standard pack routine:

cd ComfyUI/custom_nodes
git clone https://github.com/Anymatix/anymatix-comfy-nodes

then restart, or install via ComfyUI Manager (search "anymatix-comfy-nodes"). The pack pulls PyAV (av) and imageio-ffmpeg as dependencies, which is what the save side uses - this packaging node itself just needs numpy/PIL.

Issues to expect

The usual stumbles: feeding it a single image (a batch of one) and wondering why the "video" doesn't move - it won't, that's one frame - and confusing this packaging node with an actual encoder, then hunting for an MP4 on disk before you've added a save node. Check your batch count is real, wire a save node on the output, and set the fps to what your frames actually represent.