ComfyUI Node
Video Render Batch (A-S7.5 in-process render)
A ComfyUI node in OldTimeRadio/v2/video with 10 inputs and 2 outputs.
Video Render Batch (A-S7.5 in-process render)
- render_report_json
- clip_manifest_json
◄modesoak►
◄beats40►
◄oom_index20►
◄frame_count25►
◄engineviz_camera►
◄portrait_path►
◄audio_path►
◄patched_ledger_json{}►
◄master_audio_path►
◄image_done►
CategoryOldTimeRadio/v2/video
Inputs (10)
| Name | Type | Default | Description |
|---|---|---|---|
| mode | COMBO | soak | Run mode. 'episode' is the production path (renders one REAL per-beat clip per shot from patched_ledger_json). 'soak' and 'single' are diagnostic harness modes -- some widgets below are read ONLY in a specific mode (see engine / oom_index). |
| beats | INT | 401–400 | Diagnostic harness only (mode=soak): synthetic beat count. Ignored in mode=episode (beats come from the planned ledger). |
| oom_index | INT | 200–399 | Diagnostic harness only -- read ONLY in mode=soak (the beat index at which to inject a simulated OOM for fallback-trail testing). Inert in mode=single and mode=episode. |
| frame_count | INT | 251–240 | Diagnostic harness only (mode=soak): per-clip frame count. Ignored in mode=episode (per-shot frame budget is planned upstream). |
| engineopt | STRING | viz_camera | Diagnostic harness only -- read ONLY in mode=single (forces the single engine to render). Inert in mode=soak and mode=episode (episode routes each beat by its planned per-role engine). |
| portrait_pathopt | STRING | — | |
| audio_pathopt | STRING | — | |
| patched_ledger_jsonopt | STRING | {} | mode=episode: the OTR_ShotLock-planned (and image-genned) ledger JSON. run_real_episode renders one REAL per-beat clip per shot; the emitted clip manifest feeds OTR_SilentComposite. |
| master_audio_pathopt | STRING | mode=episode: path to the FROZEN master mix (MP4 or WAV). Beats whose ledger line has no per-line *_wav_path get audio_ref filled by slicing [start_s, start_s+dur_s] from this file (read-only ffmpeg; master is NEVER mutated). Wire from OTR_EpisodeAssembler.output_path (the frozen master WAV). Leave unset to degrade LOUD on missing per-line wavs. | |
| image_doneopt | STRING | Image-done gate (mirrors audio_done; ST-0.2 still-spine). Wire from OTR_ImageGenDispatcher.image_done so the video render cannot start before every episode still exists on disk (W4). Opaque STRING; the token is never parsed. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| render_report_json | STRING | — |
| clip_manifest_json | STRING | — |