Qwen → ChronoEdit Bridge

This is a plumbing node. Its job is to take an image you made with Qwen - usually a Qwen-Image-Edit result - and get it into the exact shape a ChronoEdit video pipeline expects: the right resolution, the right frame count, resized without breaking the aspect ratio in a way that ruins the first frame. It's the adapter between "I have a nice edited still" and "now animate it," and like most adapters it's unglamorous and load-bearing at the same time.

A bit of context, because ChronoEdit is niche enough that it barely shows up in community chatter (a corpus search for it comes back empty, so temper your expectations about finding tutorials). ChronoEdit is a video-editing pipeline built in the Wan lineage - the same Alibaba video family, frozen at Wan 2.2, that most 2026 open video work sits on. The pattern this node serves is the one that gives the whole pack its name: use Qwen's strong instruction editing to nail a first frame, then hand that frame to a Wan-family video model to bring it to life. This node is the "hand it over" step, specialized for the ChronoEdit target.

How it works

You give it your image and the target video geometry - width, height, and how many frames the clip will be - and it resizes the still to those dimensions using the fit strategy you choose, then passes both the conformed image and the numeric parameters downstream so the rest of the ChronoEdit graph is fed consistent values. No model runs here; it's resize-and-forward.

The inputs and outputs that matter

• image - your Qwen-generated or edited still.
• target_width / target_height - the video resolution. Defaults are 832×480, a standard low-res Wan-family shape, which tells you the kind of pipeline this is aimed at.
• num_frames (default 17) - how long the clip is. This is passed through so the sampler and the rest of the graph agree on frame count.
• resize_mode (stretch / fit / fill) - how the still maps to the target box. fit (the default) preserves aspect ratio with padding; fill crops to cover; stretch distorts. Pick fit or fill unless you want squashed output.

Outputs: the conformed image, the width, height and num_frames as integers to wire into downstream nodes, and an info string describing what it did.

How to install it

ComfyUI Manager → ComfyUI-QwenImageWanBridge → install → restart. Or:

cd ComfyUI/custom_nodes
git clone https://github.com/fblissjr/ComfyUI-QwenImageWanBridge

then restart. The node itself is lightweight image math; the heavy lifting is the Qwen model that makes the still and the ChronoEdit/Wan model that animates it, which you supply separately.

Common issues & troubleshooting

The first frame looks stretched or cropped wrong. That's resize_mode. stretch distorts to fill the box; if your source and target aspect ratios differ a lot, use fit to letterbox or fill to crop instead.

Downstream nodes complain about dimensions or frame count. The whole reason this node exists is to keep those numbers consistent - so wire the width, height and num_frames outputs into the rest of the graph rather than typing the values again by hand, and they'll match by construction.

There's barely any documentation for ChronoEdit itself. True, and not something this node can fix. It's an obscure target with little community writeup, so you'll be leaning on the pack's example workflows and your own experimentation. If you just want image-to-video from a Qwen still, the better-trodden path is the sibling QwenToWanFirstFrameLatent into a standard Wan I2V graph, which far more people run.