Sonilo Video SFX (BudgetPixel)

Sonilo Video SFX is the sound-design node that syncs itself. Feed it a video and it adds frame-accurate foley, ambience, impacts, and action sounds matched to what's actually visible on screen - a door closes in the footage, and there's a door-close on the audio track, timed to the frame. It's the difference between a foley artist who watches your footage and one who works from a written description. And it's the one audio node in the pack that gives you a genuinely usable second output: not just the SFX track, but the source video with the effects already mixed in.

Why you'd reach for it

Because hand-placing sound effects is the most tedious hour in any edit. This handles the "what should this sound like, and when" problem in one pass for footage up to six minutes. The optional prompt input steers the sound design - "realistic Foley and ambience, no music" - while the model does the synchronization. Output format is mp3 or wav for the SFX track.

The inputs that matter

• video - required. The footage to sound-design (MP4/WebM, up to 360 seconds, 30MB remote-fetch cap). Its measured duration is the billing basis.
• prompt - optional direction (up to 1000 characters).
• format - mp3 or wav.

Here's the part that stands out in this pack: the node returns both an audio output (the SFX track) and a video output - and unlike the other Sonilo nodes where that video socket is a dead placeholder, this one is real. It's the source video with the effects mixed in. Wire the video output to a Save Video node and you've got a finished sound-designed clip. Needs 2025+ ComfyUI for core AUDIO/VIDEO types.

What it costs

15 credits per second of the input video's measured duration, with a 3-second minimum. Like Sonilo Video Music, billing is on input length, not an output length you choose. A 30-second clip is 450 credits for a full sound-design pass - budget accordingly.

How it works

Standard pack mechanics with a video upload on the input side (videos go through the API's temp storage by URL), then create → poll → download. Slower than the text-to-SFX node because the model is watching footage; expect a real wait and don't mistake the polling for a hang.

Install and key setup

ComfyUI Manager → BudgetPixel, or:

cd ComfyUI/custom_nodes
git clone https://github.com/budgetpixel-ai/comfyui-budgetpixel.git
pip install -r comfyui-budgetpixel/requirements.txt

Only requests. API key from a paid plan via BUDGETPIXEL_API_KEY or a budgetpixel.json in the pack folder; keys never in workflow JSON.

Where people get burned

The input-duration bill is the main surprise - same as the music version, this is priced on your footage, so a six-minute clip is a six-minute bill. And don't ignore that second output socket: for this node specifically it's the finished mix, which is what most people actually want. Pack rules apply: force_rerun to beat the cache, paid plan for the API, interrupts stop the wait not the job.