davinci-resolve-mcp 2.213.0 → 2.213.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/AGENTS.md +15 -6
- package/CHANGELOG.md +93 -0
- package/README.md +1 -1
- package/README.zh-CN.md +2 -2
- package/docs/reference/api-limitations.md +5 -12
- package/install.py +1 -1
- package/package.json +1 -1
- package/src/granular/common.py +1 -1
- package/src/server.py +30 -8
- package/src/utils/api_truth.py +62 -13
- package/src/utils/resolve_versions.py +3 -1
package/AGENTS.md
CHANGED
|
@@ -78,12 +78,21 @@ semantics.
|
|
|
78
78
|
limitation, add a `submit`-tagged entry to `src/utils/api_truth.py` and
|
|
79
79
|
regenerate with `scripts/gen_api_limitations.py` (a drift guard enforces it)
|
|
80
80
|
- Bundled Resolve API text: `docs/reference/resolve_scripting_api.txt` — refresh
|
|
81
|
-
it by copying the shipped
|
|
82
|
-
|
|
83
|
-
|
|
84
|
-
|
|
85
|
-
|
|
86
|
-
|
|
81
|
+
it by copying the shipped scripting README over it wholesale, never by
|
|
82
|
+
hand-adding the lines you already know about. The file's only value is that a
|
|
83
|
+
diff against it is trustworthy; a hand-patch carries a newer `Last Updated:`
|
|
84
|
+
header while still hiding everything you did not know to look for. That is
|
|
85
|
+
exactly how it sat eight weeks stale at 26 May 2026 while ten documented
|
|
86
|
+
21.0.4 methods went unwired (PR #139).
|
|
87
|
+
**The shipped README moved in 21.1.** `Developer/Scripting/README.txt` is
|
|
88
|
+
gone; the pack is now `README.md`, a typed `DaVinciResolveScript.pyi` (the
|
|
89
|
+
full signature surface, which the old README never carried) and a
|
|
90
|
+
`CHANGELOG.md` of per-version additions. Copying `README.md` over the bundled
|
|
91
|
+
file is still the right refresh, but it renumbers every line — and
|
|
92
|
+
`scripts/audit_api_parity.py`, `tests/test_resolve_bridge.py` and a few dozen
|
|
93
|
+
`resolve_scripting_api.txt line N` comments in `src/` read it. Re-run the
|
|
94
|
+
parity audit and re-anchor those comments in the same change, or leave the
|
|
95
|
+
file at its 21.0 revision and cite the `.pyi` for new work
|
|
87
96
|
|
|
88
97
|
## Key Paths
|
|
89
98
|
|
package/CHANGELOG.md
CHANGED
|
@@ -2,6 +2,99 @@
|
|
|
2
2
|
|
|
3
3
|
Release history for the DaVinci Resolve MCP Server. The latest release is summarized in the root README; older entries live here to keep the README focused.
|
|
4
4
|
|
|
5
|
+
## What's New in v2.213.1 — Fusion keyframes reach the render; the contact sheet waits for the viewer; the ledger learns Resolve 21.1
|
|
6
|
+
|
|
7
|
+
Reported in #196 by @JosephConroy93, with a repro precise enough to reproduce
|
|
8
|
+
on the first run.
|
|
9
|
+
|
|
10
|
+
### Fixed — Fusion keyframes (#196)
|
|
11
|
+
|
|
12
|
+
- **`fusion_comp add_keyframe` wrote the keyframe under `comp.Lock()`, so it
|
|
13
|
+
read back correctly and never rendered.** This is the comp-lock class this
|
|
14
|
+
repo documented in v2.98.5 for `set_input` — a value write under a lock
|
|
15
|
+
lands in the graph, `GetKeyFrames` lists it, Fusion's own page playback
|
|
16
|
+
interpolates it, and the delivered render ignores it — and `add_keyframe`
|
|
17
|
+
was the site that never got the fix. Measured on Studio 19.1.3.7 with a
|
|
18
|
+
Transform `Size` keyframed 2.0 → 1.0 over 47 frames on a media-backed clip:
|
|
19
|
+
through the shipped handler the render was bit-identical to the no-comp
|
|
20
|
+
baseline (PSNR inf); the identical writes with the lock removed rendered the
|
|
21
|
+
zoom (PSNR 13.3 dB against baseline, and frame 46 back within 44 dB of the
|
|
22
|
+
baseline as `Size` returned to 1.0). Four unlocked variants all rendered —
|
|
23
|
+
modifier locked with the write outside, `StartUndo`/`EndUndo` around both,
|
|
24
|
+
nothing wrapped, and a spline assigned directly — so the write pattern is
|
|
25
|
+
not the variable, the lock is. The handler now wraps the spline attach and
|
|
26
|
+
the first key in `StartUndo`/`EndUndo`, the escape `bulk_set_inputs`
|
|
27
|
+
already uses, which also gives the user one undo step per keyframe.
|
|
28
|
+
- **`delete_keyframe` moved off the lock for consistency, not because it was
|
|
29
|
+
broken.** It was mutation-checked: re-locking the delete and rendering still
|
|
30
|
+
removed the key from the output (frame 46 stayed at the 2.0 zoom, 5.7 dB
|
|
31
|
+
from baseline). The lock does not suppress a spline delete on 19.1.3.7.
|
|
32
|
+
|
|
33
|
+
### Notes on the report
|
|
34
|
+
|
|
35
|
+
- The reporter's own patch — write outside the lock, modifier still inside —
|
|
36
|
+
made their render fail outright on Windows 19.1.3.7 and 21.0.4.5. That
|
|
37
|
+
variant rendered correctly here on macOS 19.1.3.7, so the failure was not
|
|
38
|
+
reproduced and its cause is unknown; the shipped fix uses the undo-wrapped
|
|
39
|
+
shape instead, which rendered on every attempt.
|
|
40
|
+
- Priming (an unrelated unlocked write first) did not rescue it for the
|
|
41
|
+
reporter. That matches the v2.98.8 mechanism only partly and was not
|
|
42
|
+
re-tested here.
|
|
43
|
+
|
|
44
|
+
### Validation — #196
|
|
45
|
+
|
|
46
|
+
- `tests/test_fusion_value_write_lock.py` now recognises the keyframe write
|
|
47
|
+
shape (`tool[input][time] = value`) under a lock, and fails against the
|
|
48
|
+
pre-fix server naming `add_keyframe`'s line — the guard was confirmed to
|
|
49
|
+
fail before it was confirmed to pass.
|
|
50
|
+
- Live render witness on Studio 19.1.3.7 through the real `fusion_comp`
|
|
51
|
+
handlers: `add_keyframe` animates, `delete_keyframe` removes, `get_keyframes`
|
|
52
|
+
reads the written values. Full offline Python suite, drift guards and the
|
|
53
|
+
advanced Node suite green.
|
|
54
|
+
|
|
55
|
+
#197 and #198 contributed by @billcarroll, on the day Resolve 21.1 shipped. Also on landing: `timeline_markers get_thumbnail` read the thumbnail once too and now goes through the same settle helper; live-validated on the same scratch timeline (two of two reads), with the guard covering both sites.
|
|
56
|
+
|
|
57
|
+
### Fixed — contact sheet and get_thumbnail (#198)
|
|
58
|
+
|
|
59
|
+
- **`timeline thumbnail_contact_sheet` returned "No thumbnail available" for
|
|
60
|
+
every frame.** It moved the playhead and read the thumbnail once,
|
|
61
|
+
immediately; the viewer has not caught up when the scripting call that
|
|
62
|
+
follows a playhead move lands, so each sample came back empty. The sheet now
|
|
63
|
+
reads through `_playhead_thumbnail_settled`, the polling helper the
|
|
64
|
+
single-frame path already used — it was the one caller not switched over.
|
|
65
|
+
Live-validated on Studio 19.1.3.7: three samples across a scratch timeline,
|
|
66
|
+
all three returned thumbnails. A static guard now fails the suite if any
|
|
67
|
+
`GetCurrentClipThumbnailImage` read appears outside that helper. (#198)
|
|
68
|
+
|
|
69
|
+
### Documentation — Resolve 21.1 in the API ledger (#197)
|
|
70
|
+
|
|
71
|
+
- **Three `api_truth` entries corrected for Resolve 21.1** (#197), on the
|
|
72
|
+
strength of the contributor's attribute probe of Studio 21.1.0.14. Native
|
|
73
|
+
multicam clip creation is withdrawn as a gap (`MediaPool.CreateMulticamClip`,
|
|
74
|
+
`TimelineItem.FlattenMulticam`, `PerformMulticamSmartSwitch`,
|
|
75
|
+
`Timeline.AutoAlignClips` resolve on 21.1); transition **creation** is fixed
|
|
76
|
+
by `TimelineItem.AddTransition` while readback and cloning stay missing; and
|
|
77
|
+
the truncated `GetClipProperty('Transcription')` now has a real route around
|
|
78
|
+
it in `MediaPoolItem.GetTranscription()`, which returns per-word timing and
|
|
79
|
+
speakers for the source clip. `docs/reference/api-limitations.md` regenerated
|
|
80
|
+
(42 → 41 missing capabilities).
|
|
81
|
+
- **Provenance is stated in each entry.** No 21.1 build exists on the
|
|
82
|
+
maintainer's machine, so the entries record these as *reported by the
|
|
83
|
+
contributor, not reproduced here*, following the ledger's existing
|
|
84
|
+
measured / reported distinction. None of the methods was invoked; the claim
|
|
85
|
+
is existence and signature only, and each entry says what would falsify it.
|
|
86
|
+
- **The shipped scripting README moved in 21.1.** `Developer/Scripting/README.txt`
|
|
87
|
+
is gone in favour of `README.md`, a typed `DaVinciResolveScript.pyi` and a
|
|
88
|
+
`CHANGELOG.md`. `tests/live_resolve21_validation.py` reads either name, and
|
|
89
|
+
`AGENTS.md` explains why the bundled API text is not refreshed in the same
|
|
90
|
+
change (every `resolve_scripting_api.txt line N` anchor in `src/` would move).
|
|
91
|
+
|
|
92
|
+
### Validation — #197 and #198
|
|
93
|
+
|
|
94
|
+
- Full offline Python suite, drift guards (api-limitations, agent rules,
|
|
95
|
+
release surfaces), the advanced Node suite. Contact sheet live-validated as
|
|
96
|
+
above; the api_truth change is documentation and needs no Resolve run.
|
|
97
|
+
|
|
5
98
|
## What's New in v2.213.0 — a ColorTrace that matches on media, then applies
|
|
6
99
|
|
|
7
100
|
### Added
|
package/README.md
CHANGED
|
@@ -2,7 +2,7 @@
|
|
|
2
2
|
|
|
3
3
|
English | [简体中文](README.zh-CN.md)
|
|
4
4
|
|
|
5
|
-
[](https://github.com/samuelgursky/davinci-resolve-mcp/releases)
|
|
6
6
|
[](https://www.npmjs.com/package/davinci-resolve-mcp)
|
|
7
7
|
[](docs/reference/api-coverage.md)
|
|
8
8
|
[-blue.svg)](#server-modes)
|
package/README.zh-CN.md
CHANGED
|
@@ -2,7 +2,7 @@
|
|
|
2
2
|
|
|
3
3
|
[English](README.md) | 简体中文
|
|
4
4
|
|
|
5
|
-
[](https://github.com/samuelgursky/davinci-resolve-mcp/releases)
|
|
6
6
|
[](https://www.npmjs.com/package/davinci-resolve-mcp)
|
|
7
7
|
[](docs/reference/api-coverage.md)
|
|
8
8
|
[-blue.svg)](#服务器模式)
|
|
@@ -12,7 +12,7 @@
|
|
|
12
12
|
[](https://www.python.org/downloads/)
|
|
13
13
|
[](https://opensource.org/licenses/MIT)
|
|
14
14
|
|
|
15
|
-
> 本翻译对应 v2.213.
|
|
15
|
+
> 本翻译对应 v2.213.1 版 README。如与英文原版有出入，以 [英文原版](README.md) 为准。
|
|
16
16
|
|
|
17
17
|
一个 Model Context Protocol (MCP) 服务器，让 AI 助手通过官方脚本 API 控制 DaVinci Resolve Studio（达芬奇）。它提供完整的 API 覆盖，外加带护栏的工作流助手，涵盖剪辑、媒体池整理、渲染设置、审阅标记、调色、Fusion、Fairlight、项目生命周期任务、扩展开发，以及不碰源媒体的媒体分析。
|
|
18
18
|
|
|
@@ -12,7 +12,7 @@ that none exists).
|
|
|
12
12
|
|
|
13
13
|
**Verified on:** DaVinci Resolve Studio 21.0.2
|
|
14
14
|
|
|
15
|
-
**Totals:**
|
|
15
|
+
**Totals:** 41 missing capabilities, 50 bugs / unreliable behaviors.
|
|
16
16
|
|
|
17
17
|
The authoritative source is the runtime-queryable `api_truth` ledger
|
|
18
18
|
(`resolve_control api_truth "<query>"`); this document is generated from
|
|
@@ -79,18 +79,11 @@ equivalent, blocking full automation.
|
|
|
79
79
|
- **Reference:** [issue #73](https://github.com/samuelgursky/davinci-resolve-mcp/issues/73)
|
|
80
80
|
- **Tags:** missing-method, audio, channel
|
|
81
81
|
|
|
82
|
-
### Native multicam clip creation
|
|
83
|
-
|
|
84
|
-
- **Object:** `MediaPool`
|
|
85
|
-
- **Behavior:** There is no method to create a native multicam clip from a set of angles. Angles can be stacked onto tracks programmatically, but the multicam-clip conversion is a UI-only step.
|
|
86
|
-
- **Workaround / current handling:** Prepare a stacked timeline (media_pool setup_multicam_timeline) and finish the multicam-clip conversion in the Resolve UI.
|
|
87
|
-
- **Tags:** missing-method, media-pool, multicam
|
|
88
|
-
|
|
89
82
|
### Transition create / copy / clone
|
|
90
83
|
|
|
91
84
|
- **Object:** `Timeline / TimelineItem`
|
|
92
|
-
- **Behavior:** There is no method to ADD or CLONE an edit transition — no AddTransition/CreateTransition/AddVideoTransition on Timeline or TimelineItem (dir(), 21.0.4.5). CORRECTION, measured on Studio 21.0.4.5 (2026-08-12): this entry previously said transitions applied in the UI are 'invisible to and unmodifiable by scripts'. BOTH HALVES WERE WRONG and are withdrawn. A transition IS a first-class timeline item: a 12-frame Cross Dissolve applied through the Edit-page right-click menu enumerates in GetItemListInTrack('video', 1) as GetName()=='Cross Dissolve', GetStart()==86426, GetDuration()==12 — centered on a cut at 86432 — with a stable GetUniqueId() and a working GetTrackTypeAndIndex(). A transition authored offline into a .drp and imported reads IDENTICALLY, so the route that created it does not matter. It is also REMOVABLE: Timeline.DeleteClips([transition], False) returns True and deletes it, leaving both adjacent clips at their original starts and durations. THE DISCRIMINATOR between a transition item and a clip item is GetProperty(): a transition returns an EMPTY dict where a video clip returns 26 transform keys; it also has no MediaPoolItem and no Fusion comp. WHAT IS GENUINELY MISSING: creation
|
|
93
|
-
- **Workaround / current handling:** Automated QC of existing transitions IS possible and is the main practical need — enumerate GetItemListInTrack, treat any item whose GetProperty() is empty and whose GetMediaPoolItem() is None as a transition, and read its name, start and duration. Removal is scriptable via Timeline.DeleteClips. To CREATE one, either apply it in the Resolve UI, or author it offline and import: the advanced server's drp place_transition writes a cross dissolve at an abutting cut ({track, atFrame, durationFrames}) and it round-trips into Resolve 21.0.4.5 reading back at the expected centered range.
|
|
85
|
+
- **Behavior:** CREATION IS FIXED IN 21.1, READBACK IS NOT. Reported by @billcarroll (PR #197) from an attribute probe on Studio 21.1.0.14 (2026-09-08; not reproduced here, no 21.1 install): TimelineItem.AddTransition resolves to a <BlackmagicFusion.PyFunctionCall object>, not None. Its stub signature is AddTransition(transitionOptions) -> TimelineItem | None, where transitionOptions carries type (e.g. 'Cross Dissolve'), category ('simple'|'fusion'|'ofx'|'audio'), position ('start'|'end'), alignment ('left'|'center'|'right') and an optional duration in frames. It was NOT invoked (a call mutates the user's timeline), so this entry claims existence and signature only. WHAT REMAINS MISSING ON 21.1: reading a transition back. There is still no accessor for an existing transition's type, alignment or duration beyond its name string and frame range, and no clone verb — alignment and duration are write-only arguments to AddTransition. The pre-21.1 statement, kept as the historical record: there was no method to ADD or CLONE an edit transition — no AddTransition/CreateTransition/AddVideoTransition on Timeline or TimelineItem (dir(), 21.0.4.5). CORRECTION, measured on Studio 21.0.4.5 (2026-08-12): this entry previously said transitions applied in the UI are 'invisible to and unmodifiable by scripts'. BOTH HALVES WERE WRONG and are withdrawn. A transition IS a first-class timeline item: a 12-frame Cross Dissolve applied through the Edit-page right-click menu enumerates in GetItemListInTrack('video', 1) as GetName()=='Cross Dissolve', GetStart()==86426, GetDuration()==12 — centered on a cut at 86432 — with a stable GetUniqueId() and a working GetTrackTypeAndIndex(). A transition authored offline into a .drp and imported reads IDENTICALLY, so the route that created it does not matter. It is also REMOVABLE: Timeline.DeleteClips([transition], False) returns True and deletes it, leaving both adjacent clips at their original starts and durations. THE DISCRIMINATOR between a transition item and a clip item is GetProperty(): a transition returns an EMPTY dict where a video clip returns 26 transform keys; it also has no MediaPoolItem and no Fusion comp. WHAT IS GENUINELY MISSING (pre-21.1: creation too; on 21.1+ read the paragraph above): cloning, and any type/alignment/parameter detail — the transition's kind is knowable ONLY from its name string, and there is no way to read its alignment (centered/start/end) or edit its duration. AUDIO NUANCE (measured 2026-09-01 on 19.1.3.7, E113): an audio cross-fade enumerates in GetItemListInTrack('audio', n) with an EMPTY GetName() (24 frames, centered on the cut, between the two clips) — so on audio lanes even the kind is not readable from the name. The discriminator that holds for BOTH: GetMediaPoolItem() is None AND GetProperty() is empty — BUT a Solid Color generator AND a subtitle item read the same way (GetProperty() None, no MediaPoolItem; measured E115), so that pair only separates clips from non-clips. What separates a transition from a generator is GEOMETRY: a transition straddles a cut (one neighbour ends inside its span, another starts inside it) while a generator owns its span. timeline.get_items reports `kind` on that basis.
|
|
86
|
+
- **Workaround / current handling:** Automated QC of existing transitions IS possible and is the main practical need — enumerate GetItemListInTrack, treat any item whose GetProperty() is empty and whose GetMediaPoolItem() is None as a transition, and read its name, start and duration. Removal is scriptable via Timeline.DeleteClips. To CREATE one, either apply it in the Resolve UI, or author it offline and import: the advanced server's drp place_transition writes a cross dissolve at an abutting cut ({track, atFrame, durationFrames}) and it round-trips into Resolve 21.0.4.5 reading back at the expected centered range. On 21.1+ prefer TimelineItem.AddTransition, which takes the type, category, edge, alignment and duration directly.
|
|
94
87
|
- **Tags:** missing-method, timeline, transition
|
|
95
88
|
|
|
96
89
|
### Cloud project enumeration / export / user management
|
|
@@ -514,8 +507,8 @@ values, or automation-hostile modal prompts.
|
|
|
514
507
|
### MediaPoolItem.GetClipProperty('Transcription')
|
|
515
508
|
|
|
516
509
|
- **Object:** `MediaPoolItem`
|
|
517
|
-
- **Behavior:** Returns a PREVIEW of the transcription that ends in an ellipsis when the full transcript is longer than the property exposes.
|
|
518
|
-
- **Workaround / current handling:**
|
|
510
|
+
- **Behavior:** Returns a PREVIEW of the transcription that ends in an ellipsis when the full transcript is longer than the property exposes. Reported still true on Studio 21.1.0.14 by @billcarroll (PR #197; not reproduced here, no 21.1 install) — the property is not the fix. 21.1 adds a SEPARATE method that is not truncated: MediaPoolItem.GetTranscription(useNestedClipTranscription=False) -> {language, segments[{start, end, text, speaker, words[{start, end, text}]}]}, with timecode strings rather than frame numbers. The contributor measured it on a live 21.1.0.14 against an already-transcribed interview clip: 1550 segments, per-word start/end timecodes, a populated `speaker` field, and '(...)' as Resolve's own silence marker. Note the transcript is of the SOURCE clip, so timeline positions must be mapped through GetStart()/GetSourceStartFrame() on the timeline item.
|
|
511
|
+
- **Workaround / current handling:** On 21.1+, call MediaPoolItem.GetTranscription() instead of reading the property. On 21.0.x and earlier, treat a trailing ellipsis as truncation (see media_pool_item get_transcription's `truncated` flag).
|
|
519
512
|
- **Tags:** transcription, truncation
|
|
520
513
|
|
|
521
514
|
### ProjectManager.CreateProject (blocked by the current project)
|
package/install.py
CHANGED
|
@@ -37,7 +37,7 @@ from src.utils.update_check import (
|
|
|
37
37
|
|
|
38
38
|
# ─── Version ──────────────────────────────────────────────────────────────────
|
|
39
39
|
|
|
40
|
-
VERSION = "2.213.
|
|
40
|
+
VERSION = "2.213.1"
|
|
41
41
|
# Only hard floor: mcp[cli] requires Python 3.10+. There is no upper bound —
|
|
42
42
|
# Resolve's scripting bridge loads into newer interpreters on recent builds
|
|
43
43
|
# (Python 3.14 verified against Resolve Studio 20.3.2). Older Resolve builds
|
package/package.json
CHANGED
package/src/granular/common.py
CHANGED
|
@@ -87,7 +87,7 @@ if not logging.getLogger().handlers:
|
|
|
87
87
|
handlers=[logging.StreamHandler()],
|
|
88
88
|
)
|
|
89
89
|
|
|
90
|
-
VERSION = "2.213.
|
|
90
|
+
VERSION = "2.213.1"
|
|
91
91
|
logger = logging.getLogger("davinci-resolve-mcp")
|
|
92
92
|
logger.info(f"Starting DaVinci Resolve MCP Server v{VERSION}")
|
|
93
93
|
logger.info(f"Detected platform: {get_platform()}")
|
package/src/server.py
CHANGED
|
@@ -11,7 +11,7 @@ Usage:
|
|
|
11
11
|
python src/server.py --full # Start the 353-tool granular server instead
|
|
12
12
|
"""
|
|
13
13
|
|
|
14
|
-
VERSION = "2.213.
|
|
14
|
+
VERSION = "2.213.1"
|
|
15
15
|
|
|
16
16
|
import base64
|
|
17
17
|
import os
|
|
@@ -7067,8 +7067,15 @@ def _timeline_thumbnail_contact_sheet(proj, tl, p: Dict[str, Any]) -> Dict[str,
|
|
|
7067
7067
|
)
|
|
7068
7068
|
sampled.append(sample)
|
|
7069
7069
|
continue
|
|
7070
|
-
|
|
7071
|
-
|
|
7070
|
+
# Poll instead of reading once: the viewer has not caught
|
|
7071
|
+
# up when the scripting call right after a playhead move
|
|
7072
|
+
# lands, so a single read returns None on every sample and
|
|
7073
|
+
# the whole sheet comes back "No thumbnail available at
|
|
7074
|
+
# frame" while the same calls with a settle succeed.
|
|
7075
|
+
thumbnail, thumb_err = _playhead_thumbnail_settled(tl)
|
|
7076
|
+
if thumb_err:
|
|
7077
|
+
sample["error"] = thumb_err.get("error")
|
|
7078
|
+
elif not thumbnail:
|
|
7072
7079
|
sample["error"] = (
|
|
7073
7080
|
"No thumbnail available at frame"
|
|
7074
7081
|
if on_color
|
|
@@ -25804,8 +25811,13 @@ def timeline_markers(action: str, params: Optional[Dict[str, Any]] = None) -> An
|
|
|
25804
25811
|
# GetCurrentClipThumbnailImage returns None on every page but Color, and
|
|
25805
25812
|
# says nothing about why — hold the Color page for the read rather than
|
|
25806
25813
|
# reporting a page problem as a missing thumbnail.
|
|
25814
|
+
# Read through the settle helper: the first read after the page switch
|
|
25815
|
+
# comes back None before the viewer catches up (see PR #198, where the
|
|
25816
|
+
# contact sheet reported "No thumbnail available" on every frame).
|
|
25807
25817
|
with _color_page_for_thumbnails(get_resolve()) as on_color:
|
|
25808
|
-
thumbnail = tl
|
|
25818
|
+
thumbnail, thumb_err = _playhead_thumbnail_settled(tl)
|
|
25819
|
+
if thumb_err:
|
|
25820
|
+
return thumb_err
|
|
25809
25821
|
if thumbnail is None:
|
|
25810
25822
|
return {
|
|
25811
25823
|
"success": False,
|
|
@@ -30077,7 +30089,15 @@ def fusion_comp(action: str, params: Optional[Dict[str, Any]] = None) -> Dict[st
|
|
|
30077
30089
|
tool = comp.FindTool(p["tool_name"])
|
|
30078
30090
|
if not tool:
|
|
30079
30091
|
return _err(f"Tool '{p['tool_name']}' not found")
|
|
30080
|
-
comp.Lock()
|
|
30092
|
+
# No comp.Lock() here — see _FUSION_VALUE_WRITE_NOTE. The keyframe
|
|
30093
|
+
# assignment below is a value write, and under a lock it reads back
|
|
30094
|
+
# (GetKeyFrames lists both keys) while the RENDER ignores it: measured
|
|
30095
|
+
# on Studio 19.1.3.7 (issue #196), a Transform Size keyframed 2.0 -> 1.0
|
|
30096
|
+
# rendered bit-identical to the no-comp baseline (PSNR inf); the same
|
|
30097
|
+
# writes outside the lock rendered the zoom (PSNR 13.3 dB vs baseline).
|
|
30098
|
+
# StartUndo/EndUndo is the escape bulk_set_inputs already uses, and it
|
|
30099
|
+
# also makes "attach spline + first key" one undo step for the user.
|
|
30100
|
+
comp.StartUndo(f"Keyframe {p['input_name']}")
|
|
30081
30101
|
try:
|
|
30082
30102
|
inp = tool[p["input_name"]]
|
|
30083
30103
|
if not inp:
|
|
@@ -30119,7 +30139,7 @@ def fusion_comp(action: str, params: Optional[Dict[str, Any]] = None) -> Dict[st
|
|
|
30119
30139
|
tool[p["input_name"]][p["time"]] = p["value"]
|
|
30120
30140
|
return _ok()
|
|
30121
30141
|
finally:
|
|
30122
|
-
comp.
|
|
30142
|
+
comp.EndUndo(True)
|
|
30123
30143
|
|
|
30124
30144
|
elif action == "get_keyframes":
|
|
30125
30145
|
tool = comp.FindTool(p["tool_name"])
|
|
@@ -30145,11 +30165,13 @@ def fusion_comp(action: str, params: Optional[Dict[str, Any]] = None) -> Dict[st
|
|
|
30145
30165
|
tool = comp.FindTool(p["tool_name"])
|
|
30146
30166
|
if not tool:
|
|
30147
30167
|
return _err(f"Tool '{p['tool_name']}' not found")
|
|
30148
|
-
|
|
30168
|
+
# Same shape as add_keyframe: a spline edit is a value write, so it runs
|
|
30169
|
+
# under StartUndo/EndUndo rather than comp.Lock() (issue #196).
|
|
30170
|
+
comp.StartUndo(f"Delete keyframe {p.get('input_name')}")
|
|
30149
30171
|
try:
|
|
30150
30172
|
return _fusion_delete_keyframe(tool, p)
|
|
30151
30173
|
finally:
|
|
30152
|
-
comp.
|
|
30174
|
+
comp.EndUndo(True)
|
|
30153
30175
|
|
|
30154
30176
|
# --- Composition Control ---
|
|
30155
30177
|
elif action == "get_comp_info":
|
package/src/utils/api_truth.py
CHANGED
|
@@ -689,9 +689,22 @@ API_TRUTH: List[Dict[str, Any]] = [
|
|
|
689
689
|
"object": "MediaPoolItem",
|
|
690
690
|
"reality": "Returns a PREVIEW of the transcription that ends in an "
|
|
691
691
|
"ellipsis when the full transcript is longer than the property "
|
|
692
|
-
"exposes."
|
|
693
|
-
|
|
694
|
-
|
|
692
|
+
"exposes. Reported still true on Studio 21.1.0.14 by @billcarroll (PR #197; not "
|
|
693
|
+
"reproduced here, no 21.1 install) — the property is not "
|
|
694
|
+
"the fix. 21.1 adds a SEPARATE method that is not truncated: "
|
|
695
|
+
"MediaPoolItem.GetTranscription(useNestedClipTranscription=False) "
|
|
696
|
+
"-> {language, segments[{start, end, text, speaker, words[{start, "
|
|
697
|
+
"end, text}]}]}, with timecode strings rather than frame numbers. "
|
|
698
|
+
"The contributor measured it on a live 21.1.0.14 against an already-transcribed "
|
|
699
|
+
"interview clip: 1550 segments, per-word start/end timecodes, a "
|
|
700
|
+
"populated `speaker` field, and '(...)' as Resolve's own silence "
|
|
701
|
+
"marker. Note the transcript is of the SOURCE clip, so timeline "
|
|
702
|
+
"positions must be mapped through GetStart()/GetSourceStartFrame() "
|
|
703
|
+
"on the timeline item.",
|
|
704
|
+
"recommended": "On 21.1+, call MediaPoolItem.GetTranscription() instead of "
|
|
705
|
+
"reading the property. On 21.0.x and earlier, treat a "
|
|
706
|
+
"trailing ellipsis as truncation (see media_pool_item "
|
|
707
|
+
"get_transcription's `truncated` flag).",
|
|
695
708
|
"tags": ["transcription", "truncation"],
|
|
696
709
|
"submit": "bug",
|
|
697
710
|
},
|
|
@@ -855,18 +868,51 @@ API_TRUTH: List[Dict[str, Any]] = [
|
|
|
855
868
|
{
|
|
856
869
|
"symbol": "Native multicam clip creation",
|
|
857
870
|
"object": "MediaPool",
|
|
858
|
-
"reality": "
|
|
859
|
-
"
|
|
860
|
-
"
|
|
861
|
-
|
|
862
|
-
|
|
863
|
-
|
|
864
|
-
|
|
871
|
+
"reality": "WITHDRAWN on the strength of a contributor's probe of Studio 21.1.0.14 "
|
|
872
|
+
"(@billcarroll, PR #197, 2026-09-08; not reproduced here, no 21.1 "
|
|
873
|
+
"install): Resolve 21.1 adds "
|
|
874
|
+
"MediaPool.CreateMulticamClip(clips, multicamOptions) -> "
|
|
875
|
+
"list[MediaPoolItem], plus TimelineItem.FlattenMulticam, "
|
|
876
|
+
"TimelineItem.PerformMulticamSmartSwitch and "
|
|
877
|
+
"Timeline.AutoAlignClips. Measured by attribute probe on a live "
|
|
878
|
+
"21.1.0.14: each of those four resolves to a "
|
|
879
|
+
"<BlackmagicFusion.PyFunctionCall object>, not None — the same "
|
|
880
|
+
"discriminator that distinguishes a real method from an absent "
|
|
881
|
+
"one elsewhere in this registry. THE ENTRY IS NOT PROOF THE "
|
|
882
|
+
"METHODS WORK: none of them was invoked, because doing so mutates "
|
|
883
|
+
"the user's project. HISTORICAL, still true of 21.0.x and "
|
|
884
|
+
"earlier: there was no method to create a native multicam clip "
|
|
885
|
+
"from a set of angles; angles could be stacked onto tracks "
|
|
886
|
+
"programmatically but the multicam-clip conversion was a UI-only "
|
|
887
|
+
"step. Falsified further (either direction) by a session that "
|
|
888
|
+
"actually calls CreateMulticamClip on a scratch project.",
|
|
889
|
+
"recommended": "On 21.1+, call MediaPool.CreateMulticamClip. On 21.0.x and "
|
|
890
|
+
"earlier, prepare a stacked timeline (media_pool "
|
|
891
|
+
"setup_multicam_timeline) and finish the multicam-clip "
|
|
892
|
+
"conversion in the Resolve UI.",
|
|
893
|
+
"tags": ["media-pool", "multicam", "fixed-in-21.1"],
|
|
865
894
|
},
|
|
866
895
|
{
|
|
867
896
|
"symbol": "Transition create / copy / clone",
|
|
868
897
|
"object": "Timeline / TimelineItem",
|
|
869
|
-
"reality": "
|
|
898
|
+
"reality": "CREATION IS FIXED IN 21.1, READBACK IS NOT. Reported by @billcarroll "
|
|
899
|
+
"(PR #197) from an attribute probe on Studio 21.1.0.14 (2026-09-08; "
|
|
900
|
+
"not reproduced here, no 21.1 install): "
|
|
901
|
+
"TimelineItem.AddTransition resolves to a "
|
|
902
|
+
"<BlackmagicFusion.PyFunctionCall object>, not None. Its stub "
|
|
903
|
+
"signature is AddTransition(transitionOptions) -> TimelineItem | "
|
|
904
|
+
"None, where transitionOptions carries type (e.g. 'Cross "
|
|
905
|
+
"Dissolve'), category ('simple'|'fusion'|'ofx'|'audio'), position "
|
|
906
|
+
"('start'|'end'), alignment ('left'|'center'|'right') and an "
|
|
907
|
+
"optional duration in frames. It was NOT invoked (a call mutates "
|
|
908
|
+
"the user's timeline), so this entry claims existence and "
|
|
909
|
+
"signature only. WHAT REMAINS MISSING ON 21.1: reading a "
|
|
910
|
+
"transition back. There is still no accessor for an existing "
|
|
911
|
+
"transition's type, alignment or duration beyond its name string "
|
|
912
|
+
"and frame range, and no clone verb — alignment and duration are "
|
|
913
|
+
"write-only arguments to AddTransition. The pre-21.1 statement, "
|
|
914
|
+
"kept as the historical record: there was no method to ADD or "
|
|
915
|
+
"CLONE an edit transition — no "
|
|
870
916
|
"AddTransition/CreateTransition/AddVideoTransition on Timeline "
|
|
871
917
|
"or TimelineItem (dir(), 21.0.4.5). CORRECTION, measured on "
|
|
872
918
|
"Studio 21.0.4.5 (2026-08-12): this entry previously said "
|
|
@@ -886,7 +932,8 @@ API_TRUTH: List[Dict[str, Any]] = [
|
|
|
886
932
|
"a transition item and a clip item is GetProperty(): a "
|
|
887
933
|
"transition returns an EMPTY dict where a video clip returns 26 "
|
|
888
934
|
"transform keys; it also has no MediaPoolItem and no Fusion "
|
|
889
|
-
"comp. WHAT IS GENUINELY MISSING: creation
|
|
935
|
+
"comp. WHAT IS GENUINELY MISSING (pre-21.1: creation too; on "
|
|
936
|
+
"21.1+ read the paragraph above): cloning, and any "
|
|
890
937
|
"type/alignment/parameter detail — the transition's kind is "
|
|
891
938
|
"knowable ONLY from its name string, and there is no way to "
|
|
892
939
|
"read its alignment (centered/start/end) or edit its duration. "
|
|
@@ -913,7 +960,9 @@ API_TRUTH: List[Dict[str, Any]] = [
|
|
|
913
960
|
"server's drp place_transition writes a cross dissolve at an "
|
|
914
961
|
"abutting cut ({track, atFrame, durationFrames}) and it "
|
|
915
962
|
"round-trips into Resolve 21.0.4.5 reading back at the "
|
|
916
|
-
"expected centered range."
|
|
963
|
+
"expected centered range. On 21.1+ prefer "
|
|
964
|
+
"TimelineItem.AddTransition, which takes the type, category, "
|
|
965
|
+
"edge, alignment and duration directly.",
|
|
917
966
|
"tags": ["missing-method", "timeline", "transition"],
|
|
918
967
|
"submit": "missing",
|
|
919
968
|
},
|
|
@@ -104,7 +104,9 @@ def at_least(live: Any, required: Any) -> Optional[bool]:
|
|
|
104
104
|
#
|
|
105
105
|
# measured — probed live by this project
|
|
106
106
|
# reported — a user's live probe, credited, not independently reproduced
|
|
107
|
-
# vendor — stated by Blackmagic's shipped Developer/Scripting
|
|
107
|
+
# vendor — stated by Blackmagic's shipped Developer/Scripting docs
|
|
108
|
+
# (README.txt through 21.0; README.md, CHANGELOG.md and
|
|
109
|
+
# DaVinciResolveScript.pyi from 21.1 on)
|
|
108
110
|
# documented — a floor this server already enforces at the call site, taken
|
|
109
111
|
# from Blackmagic's release documentation. See CODE_FLOORS.
|
|
110
112
|
|