VideoHelperSuite reference: frame loading, audio, composition and saved output

Purpose and scope

VHS_VideoCombine illustrates the media layer around generated content. Reading, sampling frames, combining and saving should not be confused with the generation model itself.

Source-supported context

The README documents Load Video and Video Combine options for FPS, formats, audio and storage. A specific August 2026 issue reports precision-dependent audio behavior; that remains the reporter's test, not ours. Source 1 Source 2

Before you start

1. Input FPS, forced sampling, frame skips and caps determine which frames enter processing.

2. Final frame_rate, audio, codec and pixel format determine the output file.

3. save_output=true selects ComfyUI output storage; false selects temp storage in the reviewed node source. A frontend preview, file location and embedded workflow metadata are separate checks: playback does not prove the graph can be restored. VHS source

Steps and checks

The sequence below is an editorial procedure based on the cited context, not a diagnosis already confirmed for your environment.

Step 1. Test reading and composition with a short valid input before adding a large generation model or complex audio effects.

Step 2. Record actual input/output frame counts and target FPS, then compare expected duration with the saved file. Inspect audio-track presence, duration and level separately; a video preview alone cannot establish any of them.

Step 3. Read the full traceback and FFmpeg stderr to separate a missing executable from rejected codec or format arguments. If sound is wrong, inspect the upstream AUDIO and saved track separately; stderr alone cannot establish that waveform samples were nonfinite.

Step 4. After dependency updates, regress video, audio and API paths separately and preserve the node version and successful sample conditions.

Completion check

Confirm the actual file path and whether it is in temp or output storage, then inspect container, frame count, duration and audio independently of frontend preview. If workflow recovery matters, verify the metadata or a separate canvas export rather than assuming a playable file contains the graph.

Limitations and cautions

The pinned Comfy Registry 1.7.9 archive declares these VHS_VideoCombine inputs: images (IMAGE, also accepts LATENT), frame_rate (FLOAT/INT; default 8, minimum 1, step 1), loop_count (INT; default 0, range 0–100), filename_prefix (STRING; default AnimateDiff), format, pingpong (false), and save_output (true). Optional: audio (AUDIO), meta_batch (VHS_BatchManager), and vae (VAE). Hidden execution context: PROMPT, EXTRA_PNGINFO, UNIQUE_ID. Output declaration: VHS_FILENAMES named Filenames; function combine_video; output node.

format is dynamic: GIF/WebP are combined with FFmpeg formats discovered from installed or registered format files, so the archive cannot identify your machine's choices. This is an AST-only parse of the pinned source, not live object_info or a runtime test. It covers this single node; interfaces for the other 39 classes were not harvested. The read-only 1.7.9 static interface record contains the archive and source-file hashes; the repository copy is docs/VHS-1.7.9-STATIC-INTERFACE.json.

Original sources

• VideoHelperSuite README — checked 2026-09-26.
• VHS audio dtype report under fp16 intermediates — checked 2026-09-26.
• Current VideoHelperSuite node source — checked 2026-09-26.
• Comfy Registry 1.7.9 source archive — SHA-256 rechecked 2026-09-26.

Source, static-interface and English-language review: 2026-09-26. No VHS execution, video export, GPU run or runtime guarantee is claimed.

Related troubleshooting and guides

• FFmpeg not found: test the executable in ComfyUI's process environment
• Video export fails or plays incorrectly: check frames, encoding and audio separately
• VHS output is nearly silent: inspect audio dtype before increasing gain