Claude/loving lamport 5 st6 m

[blofi]

Adding Vertical output + face detect (for pan and scan help).

---

Summary by cubic

Adds subject-tracked 9:16 vertical render mode and a Resolve export that outputs ProRes 422 HQ shots with frame-based handles for easy timeline handoff.

• New Features

  • --vertical: landscape scales to 1920 height with subject-tracked crop to 1080; portrait scales normally; draft uses 720×1280.
  • Vertical extraction: outputs to clips_*_vertical/; base files are base*_vertical.mp4.
  • export_resolve.py: creates resolve_shots/ with ProRes 422 HQ .mov shot clips trimmed with --handles frames (default 25), plus timeline.fcpxml (1.9), timeline.edl, and README.txt. Add --zip to also produce resolve_package.zip.

• Dependencies

  • Added uv.lock to pin dependencies.

^{Written for commit c7637a8. Summary will update on new commits. Review in cubic}

[cubic-dev-ai]

2 issues found across 3 files

Prompt for AI agents (unresolved issues)

Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="helpers/render.py">

<violation number="1" location="helpers/render.py:180">
P1: Vertical mode does not enforce fixed 9:16 dimensions for portrait inputs, which can cause downstream concat-copy failures due to mixed segment resolutions.</violation>
</file>

<file name="SKILL.md">

<violation number="1" location="SKILL.md:78">
P2: Documentation inconsistency: `clips_graded_vertical/` output directory introduced in `render.py` description but missing from canonical 'Directory layout' section.</violation>
</file>

<sub>Reply with feedback, questions, or to request a fix.

Fix all with cubic | Re-trigger cubic</sub>

[cubic-dev-ai]

P1: Vertical mode does not enforce fixed 9:16 dimensions for portrait inputs, which can cause downstream concat-copy failures due to mixed segment resolutions.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At helpers/render.py, line 180:

<comment>Vertical mode does not enforce fixed 9:16 dimensions for portrait inputs, which can cause downstream concat-copy failures due to mixed segment resolutions.</comment>

<file context>
@@ -171,15 +176,30 @@ def extract_segment(
-    if draft:
-        scale = "scale=-2:1280" if portrait else "scale=1280:-2"
+
+    if vertical and not portrait:
+        # Landscape → 9:16: scale to target height (making it wider than needed),
+        # then centre-crop to the target width. Default crop x centres automatically.
</file context>

[cubic-dev-ai]

P2: Documentation inconsistency: clips_graded_vertical/ output directory introduced in render.py description but missing from canonical 'Directory layout' section.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At SKILL.md, line 78:

<comment>Documentation inconsistency: `clips_graded_vertical/` output directory introduced in `render.py` description but missing from canonical 'Directory layout' section.</comment>

<file context>
@@ -75,7 +75,7 @@ Helpers (`helpers/transcribe.py`, `helpers/render.py`, etc.) live alongside this
 - **`pack_transcripts.py --edit-dir <dir>`** — `transcripts/*.json` → `takes_packed.md` (phrase-level, break on silence ≥ 0.5s).
 - **`timeline_view.py <video> <start> <end>`** — filmstrip + waveform PNG. On-demand visual drill-down. **Not a scan tool** — use it at decision points, not constantly.
-- **`render.py <edl.json> -o <out>`** — per-segment extract → concat → overlays (PTS-shifted) → subtitles LAST. `--preview` for 720p fast. `--build-subtitles` to generate master.srt inline.
+- **`render.py <edl.json> -o <out>`** — per-segment extract → concat → overlays (PTS-shifted) → subtitles LAST. `--preview` for 720p fast. `--build-subtitles` to generate master.srt inline. `--vertical` for 9:16 1080×1920 output (landscape sources centre-cropped; clips land in `clips_graded_vertical/` so both cuts coexist).
 - **`grade.py <in> -o <out>`** — ffmpeg filter chain grade. Presets + `--filter '<raw>'` for custom.
 
</file context>

[cubic-dev-ai]

2 issues found across 3 files (changes from recent commits).

Prompt for AI agents (unresolved issues)

Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="helpers/export_resolve.py">

<violation number="1" location="helpers/export_resolve.py:267">
P1: Missing validation of temporal inputs can silently produce wrong clips or cause opaque ffmpeg failures</violation>

<violation number="2" location="helpers/export_resolve.py:272">
P1: Unsanitized EDL source name used in output path construction, enabling path traversal/out-of-directory writes.</violation>
</file>

<sub>Reply with feedback, questions, or to request a fix.

Fix all with cubic | Re-trigger cubic</sub>

[cubic-dev-ai]

P1: Missing validation of temporal inputs can silently produce wrong clips or cause opaque ffmpeg failures

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At helpers/export_resolve.py, line 267:

<comment>Missing validation of temporal inputs can silently produce wrong clips or cause opaque ffmpeg failures</comment>

<file context>
@@ -0,0 +1,344 @@
+        src_name = r["source"]
+        src_path = resolve_path(sources[src_name], edit_dir)
+        meta = src_meta[str(src_path)]
+        seg_start = float(r["start"])
+        seg_end = float(r["end"])
+        cut_dur = seg_end - seg_start
</file context>

[cubic-dev-ai]

P1: Unsanitized EDL source name used in output path construction, enabling path traversal/out-of-directory writes.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At helpers/export_resolve.py, line 272:

<comment>Unsanitized EDL source name used in output path construction, enabling path traversal/out-of-directory writes.</comment>

<file context>
@@ -0,0 +1,344 @@
+        cut_dur = seg_end - seg_start
+        beat = r.get("beat") or r.get("note") or ""
+
+        out_name = f"shot_{i + 1:02d}_{src_name}.mp4"
+        out_path = shots_dir / out_name
+        print(f"  [{i + 1:02d}] {src_name}  {seg_start:.2f}-{seg_end:.2f}  ({cut_dur:.2f}s)  {beat}")
</file context>