Concurrent Submit | Zhenzhen Jimeng Video API

Jimeng's video API is billed by the second - the pack's README quotes 0.6元 for 5 seconds, 1.2元 for 10 - which makes it one of the more transparently-priced video options in this pack, and also one where the cheap path is obvious: test at 5s. This node wraps that API in the concurrent pattern, so you can run ten 5-second probes in parallel, compare, and only spend the 10-second money on keepers.

The feature list is short and that's fine: prompt, a duration toggle, an aspect ratio set, a CFG knob, and image-to-video via a reference image. There's no negative prompt, no camera controls, no multi-ref. It's a clean, cheap video model for when you want a clip without the bells.

How the wrapper works

Underneath, the plain ComflyJimengVideoApi node posts your request to Jimeng's video endpoint through Zhenzhen's proxy, uploads a reference image if connected, then polls for the finished clip. The Submit wrapper keeps every input but returns a single task of type COMFLY_VIDEO_FUTURE, deferring the call to the video worker pool (default 10, COMFLY_VIDEO_CONCURRENCY env var).

The task output must flow into ComflyConcurrent_Video_Await, which waits for all connected tasks and returns video_1 … video_10 in submission order plus a status JSON. failure_mode (fail_fast / placeholder) decides whether one failure aborts the batch or blanks the slot.

Inputs that matter

prompt is required. duration is 5 or 10 seconds - the whole pricing story. aspect_ratio defaults to 16:9 with 1:1, 21:9, 9:16, 4:3, 3:4 alongside. cfg_scale is the one creative control and it defaults to a mild 0.5 (range 0.1–1.0) - for Jimeng video, higher isn't automatically better; 0.5 is where most users leave it.

image enables image-to-video - connect a frame and the model animates it rather than generating from text. seed (0–2147483647, default 0) is here for reproducibility; it's a real per-run seed rather than just cache control, so fixing it and re-running won't silently replay a cache like some of the pack's other nodes. api_key is optional once API Settings is configured.

Install and gotchas

Pack install as always: ComfyUI Manager → search Comfyui-zhenzhen, or git clone https://github.com/T8mars/Comfyui-zhenzhen into ComfyUI/custom_nodes, restart. No model downloads.

The per-second meter is the thing to respect: ten concurrent 10s clips cost roughly ten times a single 5s test, and there's no resolution dropdown to accidentally blow the budget on - the price lever here is purely duration. A 500 is usually upstream ByteDance load; re-run is the documented answer. And if a slot reports failed at the Await node after a long render, the clip is probably finished - the site's task page has the download, so don't spend credits on a re-run until you've checked it.