Alibaba Happyhorse 1.0 Reference To Video (Official Edition)

HappyHorse 1.0's reference-to-video node, official edition. Where the I2V variant takes exactly one first frame, R2V takes a whole set of reference images and lets your prompt refer to them by number - [ Image 1] for this character, [ Image 2] for that outfit. It's multi-image conditioning without any ControlNet or adapter stack, because the API does the conditioning for you.

This is the node for the workflows that usually require real craft locally: keep a character consistent across a scene, dress a subject from one reference while copying a pose from another, or stitch a storyboard's worth of stills into a single animated result. If you've ever assembled a reference-grid workflow of image adapters in ComfyUI, you'll appreciate how bluntly simple this is in comparison.

How it works

Cloud wrapper, standard BizyTRD mechanics, one distinctive input: media is a COMFY_AUTOGROW_V3 type, which means the node grows more input slots as you drop in more images. The prompt (required) is where the magic happens - per the tooltip, you refer to images in order using [ Image 1] , [ Image 2] , and so on. The number in your prompt maps to the order of the media inputs, so keep the mapping straight.

Controls: resolution (720P/1080P, default 1080P), ratio (default 16:9), duration (integer 3–15, default 15 in the schema), seed (-1 for auto), and watermark (default on). api_config overrides endpoint/key per request; skip_error outputs a placeholder instead of aborting the workflow.

Inputs and outputs

• media (required, autogrowing) - your reference images, in numbered order.
• prompt (required, multiline) - use [ Image 1], [ Image 2], etc. to reference them.
• Optional: resolution, ratio, duration, seed, watermark, api_config, skip_error.
• Outputs: video (VIDEO), request_id, urls.

Installing BizyAirPlus

ComfyUI Manager: "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. Dependencies: bizyair-cloudberry, bizytrd.

Common issues

The number-mapping is the one true trap: reorder an image in the autogrowing media input and every [ Image N] in your prompt quietly points at the wrong thing. Triple-check order before a long batch. Also worth remembering: more references usually means a slower, pricier generation, and the watermark default is on. Then the pack standard - missing button (pip show bizyair-cloudberry), cloud errors in the console, metered pricing.