X AI Grok Imagine Video Edit Video (Official Edition)

Video-to-video is where hosted models quietly show their teeth, and this is Grok Imagine's version: feed it an existing video plus an instruction, and it rewrites the footage to match. The official edition routes through xAI's API, so this is the node to grab when you already have xAI credentials - the channel editions of the same model go through SiliconFlow's BizyAir instead.

"Video edit" is doing a lot of work here. It covers restyling - "make it look like film noir," "convert to claymation" - and selective changes, the kind of thing that would take a full local pipeline of frame extraction, redrawing, and recombination. If you've ever tried to restyle video locally, you know the pain: the temporal consistency alone is a project. A hosted VEdit call turns that into one prompt and a bill.

How it works

Standard BizyTRD mechanics, one twist: the input is a VIDEO tensor rather than images. BizyAirPlus ON, drop in your clip, describe the edit, and the node sends the video up to the official API and gets an edited VIDEO back. Controls are minimal - prompt (required) and resolution (required, default 720p). The input resolution of your source clip matters here; garbage in, garbage out, and video upscaling is not this node's job.

api_config overrides base_url and api_key per request, which is the place to put your xAI vendor key if it differs from your BizyAir default. skip_error turns a failed request into a placeholder output instead of crashing the workflow - genuinely worth enabling when you're batch-editing a folder of clips.

Inputs and outputs

• videos (required, VIDEO) - the footage to edit.
• prompt (required, multiline) - the edit intent.
• resolution (required, default 720p).
• Outputs: video (VIDEO), request_id, urls.

Installing BizyAirPlus

ComfyUI Manager: search "BizyAirPlus". Or:

cd /path/to/ComfyUI/custom_nodes
git clone https://github.com/siliconflow/BizyAirPlus.git
cd BizyAirPlus
python -m pip install -r requirements.txt

Restart, account at bizyair.ai, API key, BizyAirPlus button ON. No model downloads - dependencies are bizyair-cloudberry and bizytrd.

Common issues

The pack's usual suspects: missing button → dependency environment (pip show bizyair-cloudberry), cloud failures → key, network, console. Two video-specific gotchas: the length of the input clip effectively caps what the model can do, so don't feed it a 10-minute file and expect a faithful one-pass edit; and video editing is priced per generation on top of the source upload, so test on a short, low-res slice of the footage first before committing the whole thing.