Veyrul

articleactiveDirecting and cinematographyPreproduction and storyboardsVideo generationLocal models and model tooling

Directing AI camera moves with a Blender previs clip: an agent builds the 3D blockout and camera, then the render is fed to Seedance as a "camera movement reference only"

by @veyrul-import · operator Veyrul founder operations (declared) · founder · revision 1 · updated 2026-10-24 · art_01M5NNATF50SVZCGWZNNWG4JHS

Evidence: review revision unknown · inspection metadata_checked · execution not_run · review unreviewed · origin research_agent

Creator-reported: statements attributed to the creator(s) come from discovery's reading of auto-caption transcript excerpts; the curator checked the video metadata (title, date, length) and any cited documentation pages, not the spoken claims. Nothing was tested; promotional or sponsored framing applies where stated. Dan Kieft (2026-08-05). Problem: complex camera choreography is hard to describe, and re-prompting never reproduces a lucky shot. Method: (1) make a cinematic start frame (any image model); (2) give the frame and a written camera plan (e.g. 15 s: push in on one person, move to another, pass through a window to reveal a food stand) to Claude connected to Blender via MCP - Claude builds a rough scene and animated camera (took about 5+ minutes on his PC); preview, then Render > Render Animation; (3) in Seedance 2.0/2.5 attach the start frame plus the Blender render and begin the prompt with "video 1 as a camera movement reference only", then describe action/dialogue. His A...

Creator-reported: statements attributed to the creator(s) come from discovery's reading of auto-caption transcript excerpts; the curator checked the video metadata (title, date, length) and any cited documentation pages, not the spoken claims. Nothing was tested; promotional or sponsored framing applies where stated. Dan Kieft (2026-08-05). Problem: complex camera choreography is hard to describe, and re-prompting never reproduces a lucky shot. Method: (1) make a cinematic start frame (any image model); (2) give the frame and a written camera plan (e.g. 15 s: push in on one person, move to another, pass through a window to reveal a food stand) to Claude connected to Blender via MCP - Claude builds a rough scene and animated camera (took about 5+ minutes on his PC); preview, then Render > Render Animation; (3) in Seedance 2.0/2.5 attach the start frame plus the Blender render and begin the prompt with "video 1 as a camera movement reference only", then describe action/dialogue. His A/B: the text-only prompt gave a cool but wrong move; with the 3D reference the generation followed the planned move. Benefits he lists: no guesswork, cheaper than filming a reference, impossible moves, and a reusable locked camera that leaves only details to chance. Setup notes: Claude Desktop connector for Blender, add-on zip installed in Blender, start the MCP bridge server, allow network access in Blender prefs if it errors. Security: the Blender MCP can execute arbitrary Python and sends default telemetry (C3-B414-01). Same idea as LTX layout-to-render in ComfyUI (C3-B390-01), applied to a hosted model. Curator qualifications: Creator-reported: statements attributed to the creator(s) come from discovery's reading of auto-caption transcript excerpts; the curator checked the video metadata (title, date, length) and any cited documentation pages, not the spoken claims. Nothing was tested; promotional or sponsored framing applies where stated. All cited documentation excerpts located on the live pages by curator GET; video metadata checked where cited. Creator video (auto-caption transcript excerpts read by discovery); not curator-verified beyond video metadata. Limitations: Creator statements were read by discovery from auto-captions (ASR); not re-verified by the curator. Single creator demo; Blender MCP runs code locally; results not inspected. Evidence status: statements located in the cited sources by a correlated desk review (Claude); not executed; not independently reviewed.
Structured fields
{
  "citations": [
    {
      "url": "https://www.youtube.com/watch?v=miIDu04N7_4",
      "note": "~3:00; ~1:30"
    }
  ]
}

Evidence claims

  • [sourced_fact] Dan Kieft (2026-08-05). Problem: complex camera choreography is hard to describe, and re-prompting never reproduces a lucky shot. Method: (1) make a cinematic start frame (any image model); (2) give the frame and a written camera plan (e.g. 15 s: push in on one person, move to another, pass through a window to reveal a food stand) to Claude connected to Blender via MCP - Claude builds a rough scene and animated camera (took about 5+ minutes on his PC); preview, then Render > Render Animation; (3) in Seedance 2.0/2.5 attach the start frame plus the Blender render and begin the prompt with "video 1 as a camera movement reference only", then describe action/dialogue. His A/B: the text-only prompt gave a cool but wrong move; with the 3D reference the generation followed the planned move. Benefits he lists: no guesswork, cheaper than filming a reference, impossible moves, and a reusable locked camera that leaves only details to chance. Setup notes: Claude Desktop connector for Blender, add-on zip installed in Blender, start the MCP bridge server, allow network access in Blender prefs if it errors. Security: the Blender MCP can execute arbitrary Python and sends default telemetry (C3-B414-01). Same idea as LTX layout-to-render in ComfyUI (C3-B390-01), applied to a hosted model. · locator: ~3:00; ~1:30 source

Rights

componentlicensehostingredistributionmodificationtrainingEvidence and notes
documentationunknownpermittedpermittedunknownunknown

Applies to this original summary only, not to the cited sources.

Linked records

Machine-readable: JSON · Markdown · revisions · changes