arkennemasis fal · MiniMax Video 01 Live Image to Video · $0.5/run

Your image goes in, a moving clip comes out, $0.50. That's the entire contract. This is the image-to-video half of the "live" checkpoint - same widgets as the plain Video 01 image-to-video node, same price, different model on fal's side.

Which raises the obvious question, and I'd rather answer it straight: fal publishes the same one-line description for both variants, the node exposes no settings that differ, and there's no Reddit thread anywhere discussing this pack to tell you which is better. So treat "live" as a variable to test on your subject, not as a documented upgrade. The node is a drop-in swap, and the pack's Shot Selector lets you gate one branch to run without executing the other, so an A/B on the same still costs one call, not two.

What image-to-video is actually for

The KB's framing of API nodes splits "you want a closed model" from "you want someone else's GPU". This node is the first kind, and the reason people tolerate it is that it makes an existing decision move. You already own the frame - composed locally, in a style your checkpoint handles, at the aspect ratio you need. You are buying exactly one thing: motion, at a price you can see before you commit.

Which is also its limitation. There's no camera vocabulary here (the Director variants have that), no identity reference (Subject Reference has that), no seed, and no length or resolution widget - the endpoint serves its own fixed-length clip in your frame's shape. For a single atmospheric shot, that's plenty. For a five-shot character sequence, you want the other nodes.

The inputs that matter

image - the first frame. One image per call. This matters more than it sounds: the pack validates everything before uploading, and a model that expects a single picture rejects a batch instead of quietly using the first. If your LoadImage is handing over a batch, split it first - otherwise you'll get an error that reads like a network fault and isn't.

prompt - what happens in the shot. Describe the motion, not the picture; the picture is already there.

prompt_optimizer - on by default, meaning the model rewrites your prompt before generating. Turn it off when you want to know what your own words did.

The outputs

video - a real ComfyUI VIDEO value. This is the one that wires somewhere: a save node, the pack's Video Dub or Narration Fit chain, a lip-sync node.

video_path - the same file as a string, saved under output/fal/fal-ai_minimax_video-01-live_image-to-video/. Handy when you're scripting something outside ComfyUI.

info - JSON containing the request id, the cost estimate, the saved files and fal's full answer. This is your receipt.

Install

cd ComfyUI/custom_nodes
git clone https://github.com/Hishamahmer/comfyui-arkennemasis

Restart ComfyUI, or search arkennemasis in ComfyUI-Manager. The pack's requirements.txt is a single line - httpx - and that serves its Codex nodes; the fal provider talks to fal.ai with Python's standard library, so this install is a clone and a restart. Then drop your key into the .env next to run_nvidia_gpu.bat:

FAL_KEY=your-key-from-fal.ai/dashboard/keys

Nothing in this pack will ever ask you to paste a key into a widget. That's deliberate: node values are stored in the workflow file verbatim, so a key in a widget becomes a key in every screenshot and every JSON you share.

Where people get burned

• Batched image → rejected pre-flight. Nothing billed.
• The motion is mush. Small, physical motion animates well; big narrative action in a single 50-cent clip usually doesn't. Keep the shot simple.
• You resubmitted and got the identical clip. reuse_identical_run (default on) saw the same inputs and reused your already-paid file. Turn it off for a fresh sample - and note that there's no seed box to steer it with, so a new sample is genuinely a new roll.
• An abandoned node sat at "waiting for its turn". max_concurrent defaults to 1 across the run. Raise it on the nodes you want side by side.