track-gate: single-pass timeline render proven - prompt_override needs the type field, task_index=-1 + prompt_format=promptRelay renders all scenes in one pass ([s-e] range relay prompt, 17k+5 snap 720->719); corrects the earlier -1-is-not-a-render-path claim; proven against my timeline ref.json with the repo's own code

This commit is contained in:
Z User 2026-11-01 04:35:44 +00:00
parent 7b6a2da472
commit b45e6b72e7

View File

@ -13,13 +13,59 @@ the workflow, not in the JSON. Nothing needs to be added to the payload for mult
Side facts (same node, verified in source): Side facts (same node, verified in source):
- `task_index = -1` outputs the **whole timeline as one merged segment** (all task - `task_index = -1` switches the node to **full-timeline mode**: window 0..720,
contents joined, full frame length). It is for relay/audio-lock cases, NOT a way to LENGTH = all frames snapped to the H3 17k+5 grid (720 → 719), media from all
render 3 clips: the local H3 model is trained at ~124–362 frames (~5–15 s); a 30 s segments. What the prompt becomes depends on `prompt_format` — see the
merged pass is out of range. single-pass section below (`promptRelay` makes it render all scenes in one pass).
- `easy multiTrackInfoOutput` (tracks_info in) exposes `TASK_COUNT` = number of task - `easy multiTrackInfoOutput` (tracks_info in) exposes `TASK_COUNT` = number of task
segments — 3 for the Sam payload. Also WIDTH/HEIGHT/TOTAL_FRAMES/FPS. segments — 3 for the Sam payload. Also WIDTH/HEIGHT/TOTAL_FRAMES/FPS.
## The prompt_override input requires the `type` field (proven)
`MultiTrackEditor.execute` (`nodes/basic.py` L1949-1957): with `format = "MiniMax"`
it calls `is_minimax_prompt_override(prompt_override)` FIRST. Only a payload with
`"type": "minimax_prompt_override"` and a `prompts` list passes
(`utils/prompt_override.py` `parse_minimax_prompt_override`). Any other non-empty
value — including a full TRACKS_INFO JSON — falls to the `|`-text parser, which
treats the ENTIRE string as one prompt: one segment 0..479 whose `text` is the
literal JSON. Proven by running the repo's parser on the captured workflow
(`scripts/prove_relay_chain.py` STEP A). So: paste the compact payload (with the
type field) into `prompt_override`; full TRACKS_INFO belongs only in `track_data`.
## Single-pass timeline render: task_index=-1 + prompt_format=promptRelay
This is the pattern the user's own graph (`my timeline ref.json`) is wired for —
no ForLoop, one sampler pass, all scenes in one generation:
1. Node 186 (String → `prompt_override`): the compact payload WITH `type` field.
2. `easy multiTrackTaskOutput`: `task_index = -1` AND `prompt_format = "promptRelay"`.
3. Everything else stays as-is: TYPE output stays `t2v` (first task's explicit
task_type → matchLine → `multi_frames`), LENGTH output becomes 719, one
SamplerCustomAdvanced pass renders the whole timeline.
With `promptRelay` + full-timeline mode, TaskOutput builds the prompt from ALL
task segments (`_format_marker_task_prompt_relay`), each scene suffixed with its
native H3 frame range, joined by ` | `:
```
<scene 1 text> [0-240] | <scene 2 text> [240-480] | <scene 3 text> [480-720]
```
SYSTEM_PROMPT output is empty in this mode; the ranges are the H3 timeline
syntax (`_FRAME_RANGE_RE` parses the same `[s-e]` / `[s-e,s-type]` tokens in the
override parser). Proven by executing the repo's own functions on the Sam
payload (STEP C: 3 ranges, length 719, type t2v).
Both settings are required — `task_index = -1` with `prompt_format = "default"`
keeps only the FIRST task's prompt (STEP D): a 719-frame latent conditioned on
scene 1 alone.
Caveat from the node's own tooltip: the trained range is ~124–362 frames and
"longer is untested". 719 frames (~30 s) is the intended use of this graph shape
(`multi_frames` mode + BlockSparseAttention `min_tokens 12288` exist for long
sequences), but if quality degrades, fall back to the ForLoop pattern above,
which renders each 240-frame clip inside the trained range.
## The single-run loop (from the pack's own ForLoop example) ## The single-run loop (from the pack's own ForLoop example)
One queue run renders every clip and concatenates them: One queue run renders every clip and concatenates them: