neural-reality-capture
Use this skill for provider-independent reality capture that turns photographed or video-derived real places, objects, aerial sites, interiors, or people into photogrammetric meshes, NeRFs, or 3D Gaussian splats. It guides representation choice, capture planning, calibration and scale, reconstruction or training, cleanup, compression and LOD, coordinate and interchange handoff, visual/geometric QA, archival records, and property, location, people, and cultural rights checks. Do not use it for text-to-3D generation, synthetic asset invention, guaranteed metrology, or broad downstream finishing
Profile →npx skills add calesthio/generative-media-skills --skill neural-reality-capture --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
Converts photographed or video-derived inputs of real places, objects, interiors, aerial sites, or people into one of three deliverable representations: photogrammetric meshes with textures, NeRFs, or 3D Gaussian splat scenes. It guides representation choice, capture planning, calibration and scale, reconstruction or training, cleanup, compression and LOD, coordinate and interchange handoff, visual/geometric QA, archival records, and rights checks.
How it works
The skill specifies representation choices based on final use (mesh, NeRF, or 3D Gaussian splats) and emphasizes staying provider-independent. It explains when not to use certain outputs (e.g., not for text-to-3D, synthetic asset invention, or broad downstream finishing). It provides capture planning guidance, including overlap, lighting, and camera behavior rules, and formalizes a calibration and coordinate workflow: preserve originals with EXIF, group cameras, add scale markers or GCPs, decide coordinate frame, and document axes and units. It includes reconstruction workflows for Mesh, NeRF, and 3D Gaussian splat paths with steps like ingest, cull, feature detection, pose estimation, calibration, dense reconstruction, texturing, and validation. It references official and research foundations (e.g., COLMAP for SfM/MVS, NeRF papers, 3D Gaussian Splatting) as factual anchors and separates asset-specific guidance from generic workflow advice.
When to use it
Use when you need provider-independent reality capture that yields a photogrammetric mesh, a NeRF, or a 3D Gaussian splat scene, with planning, calibration, QA, and handoff for archival and rights checks. It explicitly cautions against using this for text-to-3D generation, synthetic asset invention, or broad downstream finishing after handoff, and it directs the agent to select representations by the final use case and constraints.
What it can touch
Tools listed as supported: claude-code, codex, copilot, cursor. The skill includes steps for capture planning, calibration, and reconstruction workflows that imply using COLMAP for SfM/MVS and references to NeRF and 3D Gaussian splatting literature, but it does not enumerate executable tooling beyond those references. It requires documenting scale, markers, GCPs, and coordinate conventions during setup and handoff.
Caveats
License: MIT. It emphasizes provider-independence and cautions against metrology claims unless supported by calibrated workflows and independent checks. It limits scope to photogrammetric meshes, NeRFs, or 3D Gaussian splats and forbids text-to-3D, synthetic asset invention, or broad finishing work beyond handoff. It notes potential risks in texture, geometry, and calibration reliance on controlled capture and QA.
# Neural Reality Capture Neural reality capture is the production craft of reconstructing or rendering a real subject from many overlapping observations. In this skill, the deliverable is one of three families: - a photogrammetric mesh with textures, usually for game, VFX, GIS, cultural heritage, ecommerce, or DCC handoff; - a NeRF or related neural radiance field, usually for high-quality novel-v
What does the neural-reality-capture skill do?
Use this skill for provider-independent reality capture that turns photographed or video-derived real places, objects, aerial sites, interiors, or people into photogrammetric meshes, NeRFs, or 3D Gaussian splats. It guides representation choice, capture planning, calibration and scale, reconstruction or training, cleanup, compression and LOD, coordinate and interchange handoff, visual/geometric QA, archival records, and property, location, people, and cultural rights checks. Do not use it for text-to-3D generation, synthetic asset invention, guaranteed metrology, or broad downstream finishing
How do I install it?
Run `npx skills add calesthio/generative-media-skills --skill neural-reality-capture --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From calesthio/generative-media-skills, a repository with 112 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.