Zhenzhen Video G Omni Flash

Zhenzhen_Video_G_Omni_Flash is the entry-level video node in the author's own "Zhenzhen Video G" line - the plain, single-model one: zhenzhen-video-g-omni-flash, a prompt, up to two reference images, and a 4-to-15-second clip. "Flash" in the name tells you the intent: this is the fast, cheap lane of the family, the one you reach for when you're iterating on ideas and don't want the bill (or the wait) of the flagship tiers.

Like the other Zhenzhen-branded models in T8mars/ComfyUI_Seedance, this is a hosted black box served through the author's reseller API - no open weights, no inspectable lineage, per-clip billing. What the node does give you is the same reliable packaging as the rest of the pack: upload, submit to /v1/videos, poll, download with integrity checks, and decode to a ComfyUI VIDEO. The value proposition is "a video model on my existing key that's cheap enough to use for iteration," not "a mystery flagship you should bet your project on."

How it works

Fill in prompt (up to 20480 characters, with the @Image 1 material-reference syntax supported in multimodal modes), pick seconds (4 to 15), resolution (720p or 1080p), and ratio (the standard seven, forwarded as metadata.ratio). negative_prompt goes to metadata for steering. seed is native: -1 random, non-negative forwarded to metadata.seed. Connect image1 and/or image2 to do reference/first-frame work - they're submitted as images[] - or leave them empty for pure text-to-video.

That's the complete surface. No mode switcher, no 4k, no audio toggle - which is exactly the point of the "Flash" lane.

Inputs and outputs that matter

• prompt - the creative input.
• seconds - 4 to 15.
• resolution - 720p or 1080p.
• ratio - 7 options including adaptive.
• negative_prompt - optional, forwarded to metadata.
• seed - -1 random or forwarded to metadata.seed.
• image1 / image2 - optional references, submitted as images[].

Outputs: video, video_url, task_id, response.

Install

One node in T8mars/ComfyUI_Seedance:

# ComfyUI Manager: search "ComfyUI Seedance", or:
comfy node install seedance

Or clone https://github.com/T8mars/ComfyUI_Seedance.git into ComfyUI/custom_nodes, pip install -r requirements.txt, restart, and configure a key from api.seedance.nz/console via the Seedance API Config node or SEEDANCE_API_KEY.

Where people get burned

Mostly by assuming "Flash" and the Omni Flash Lowprice node are the same thing - they're not. This node is the fixed, simple one; the Lowprice variant (separate page) is a four-mode node with 4k and reference-video support. Also, don't sleep on the ratio control: adaptive follows the input, but for reference work you usually want the ratio to match your source image's orientation, so set it explicitly. And the standing house-brand advice: run your first test at 4 seconds and 720p, check the decode, then scale. The hosted model's moderation applies regardless of what you type, and every run is metered - Flash means cheap, not free.