IAMCCS V2V Bus To LTX

The V2V system in IAMCCS-nodes runs on two invisible buses: a media bus (the source video, image, and audio) and a control bus (the plan - durations, segment presets, prompts, overlap). Neither bus is something an LTX sampler can eat. V2V Bus To LTX is the adapter that unpacks both buses into concrete LTX-ready outputs: frame batches, prompts, FPS, a segment plan, and the audio VAE.

This is the first of the "BusTo" family, and it's the one that feeds the LTX backend path. It does the real planning work: given the control bus's duration, FPS, and segment settings, it computes the current segment's frame range, slices the source video down to current_source_images, extracts the segment's audio, and produces everything the LTX V2V sampler chain needs - all in one node.

Inputs

• media_bus and control_bus - both required. The source media and the plan.
• segment_index (optional, default 0) - which segment of the source you're currently rendering. Feed it your loop counter if generating segment-by-segment.

Outputs (the ones you'll actually wire)

• fps, generation_width, generation_height - the resolution/FPS the LTX chain should use, straight from the plan.
• global_prompt, negative_prompt - conditioning text.
• current_source_images - the source frames for this segment (the sliced portion of the source video).
• current_segment_audio - the audio for this segment, plus source_audio and audio_vae if your chain needs the full track.
• continuation_loops, planner_report, segment_index_out - loop/planning info for multi-segment runs.
• output_prefix, source_video_images, source_image, source_video_info (VHS video info), report - extras for output naming and diagnostics.

Install

Ships in IAMCCS-nodes:

cd ComfyUI/custom_nodes
git clone https://github.com/IAMCCS/IAMCCS-nodes.git

or via ComfyUI Manager (search "IAMCCS"), restart. No models - but it outputs VHS_VIDEOINFO and expects video loaded via the VHS-style path, so VideoHelperSuite should be present.

Practical notes

The segment_index input is the one that makes this node sing for long videos: bump it per segment and the node re-slices the source to the right window and hands you the matching audio slice, so a single BusToLTX + LTX chain can walk through an entire source clip. If the source has no audio (a silent reference), current_segment_audio will be empty - check report for the audio state rather than assuming silence is an error. And remember this node plans; the actual sampling is your LTX chain fed by these outputs.