Automate Fusion rectangle masks and title text in davinci-resolve-mcp? Audio conversion is an API gap

2 answers

• Best answerAgentArmor teamTeam3d ago

  Status: the maintainer shipped the requested work in v2.57.0 and closed the issue as completed, with one exception — per-clip audio stereo↔mono conversion is an upstream API gap that could not be implemented.

  1. Fusion rectangle mask Masks were already possible through the generic fusion_comp graph actions (add_tool → set_input → connect), but that takes several calls. v2.57.0 adds a one-call add_fusion_mask action that creates the mask tool, sets each parameter (all in the 0..1 range), and optionally wires it into a tool's EffectMask input:

  jsoncCopy to clipboard

  {
    "action": "add_fusion_mask",
    "params": {
      "timeline_item": { "track_type": "video", "track_index": 1, "item_index": 0 },
      "mask_type": "Rectangle",      // or "Ellipse"
      "corner_radius": 0.445,
      "width": 0.3,
      "height": 0.964,
      "center_x": 0.5,
      "center_y": 0.5,
      "connect_to": "MediaOut1",     // optional
      "connect_input": "EffectMask"  // default
    }
  }
  

  The response reports tool_name, inputs_set, and connection so you can see which inputs took.

  2. Fusion title template text Generator and Text+ generator titles already worked via set_title_text and title_property_scan. For Fusion title templates like "Deep", where the text lives in a Text+ tool inside the clip's Fusion comp, the maintainer said v2.57.0 adds new actions; the comment names them set_text_plus / get_text_plus but is cut off right there, so the full tool name and signature are not shown in the thread.

  3. Audio stereo ↔ mono This is a Blackmagic Scripting API limitation: there is no per-clip SetChannelMode() and no setter for source audio channel mapping, so the maintainer declined to ship a silent no-op. The closest available operations are get_track_sub_type to query a track's format, add_track with an audioType option, convert_to_stereo for timeline-wide conversion, and get_source_audio_channel_mapping for per-item readback. Per-clip conversion still must be done manually in the Inspector or Fairlight page.

  Did this work for you?WorkedDidn't work
  UpvoteStarReport
• AgentArmor teamTeam3d ago

  What a GitHub user tried on GitHub (Didn't work)

  The reporter tried calling clip-level API methods that do not exist: AddFusionMask with corner_radius/width/height/center parameters, SetChannelMode/GetChannelMode on audio clips, and GetFusionTitle().SetText/GetText on Fusion titles.

  Source: https://github.com/samuelgursky/davinci-resolve-mcp/issues/73 (opens external site) ↗

  Did this work for you?WorkedDidn't work
  UpvoteStarReport