Concurrent Submit | Zhenzhen MJ Video

Midjourney doesn't just make images anymore - hand it a prompt and it'll animate it, and the animation comes out with Midjourney's characteristic look rather than the uncanny-video default. Comfly_mj_video is the pack's wrapper for that, generating four videos at a time per call (the author's changelog is explicit: "一次生成4个视频，按次收费" - four videos per request, billed per call). This Concurrent Submit twin drops it into the 10-worker video pool, so instead of waiting on one four-pack at a time, you can queue a whole run and collect the results together.

The inputs that matter

• prompt - required.
• motion - required, and it's the fun one: how much the image moves. Low motion reads as subtle parallax, high motion reads as "the subject is doing things." Getting this dialed per prompt is where MJ video results go from okay to good.
• image - optional input image to animate (img-to-video), otherwise it's text-to-video.
• api_key - your Zhenzhen key.
• notify_hook - an optional webhook URL to ping when the job finishes (handy if you're queuing a big batch and don't want to stare at the collector).
• seed / skip_error - the usual pair.

The concurrent flow

Output is a task (COMFLY_VIDEO_FUTURE), not a video. Wire it into ComflyConcurrent_Video_Await (10 slots) along with your other video jobs - the collector waits for the whole batch, returns the videos in slot order, and appends a status JSON. Since one call already returns four videos, a couple of concurrent jobs fills the collector fast.

Install and gotchas

Part of T8mars/Comfyui-zhenzhen: ComfyUI Manager → "Comfyui-zhenzhen", or git clone https://github.com/T8mars/Comfyui-zhenzhen into custom_nodes, restart. No local models, standard deps, paid token, overseas servers (proxy from China).

Two things worth knowing. One: because it's billed per call and each call returns four videos, the effective per-video price is a quarter of the per-call price - good deal, just don't multiply your bill by four by accident. Two: motion strength is a per-prompt tuning job; a batch where every prompt uses the same motion value will show it. That's not a bug, that's the parameter doing exactly what it says.