Skip to content
Back to skills

Refinement And Export

ASecurity

"Operate Make-It-3D refinement, test rendering, multi-view

  • 247 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added September 8, 2026
toolspythonbash

Security analysis

A100/100

Pro scans all 5 files and shows the line behind each finding

Scanned September 8, 2026

npx -y skills add VectorSpaceLab/AREX-Skill --skill refinement-and-export --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Refinement And Export?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Refinement And Export
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/vectorspacelab-refinement-and-export/badge)](https://www.skillsdirectory.com/skills/vectorspacelab-refinement-and-export)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: refinement-and-export
description: "Operate Make-It-3D refinement, test rendering, multi-view
  generation, point-cloud/mesh export, and output troubleshooting after coarse
  training."
metadata:
  disco-role: operating
disable-model-invocation: true
license: NO_LICENSE
---

# Refinement and Export

Use this sub-skill after a coarse Make-It-3D workspace exists, or when a user asks about `--refine`, `--test`, videos, checkpoints, multi-view images, point clouds, OBJ mesh export, or export-specific dependencies.

## What This Sub-Skill Covers

- Refinement command construction and the source control-flow quirk around `--final --refine`.
- Test rendering from checkpoints and output locations.
- Mesh export with `--save_mesh` and its `xatlas`/`nvdiffrast` requirements.
- Point-cloud/multi-view/refine utilities and PyTorch3D/Open3D dependencies.
- Output validation and troubleshooting after partial/interrupted runs.

Setup/input issues belong to [environment-and-inputs](../environment-and-inputs/SKILL.md). Coarse command generation belongs to [coarse-training](../coarse-training/SKILL.md).

## Required Reads and Scripts

- Read [references/workflow.md](references/workflow.md) for refine/test/export command flow.
- Read [references/outputs-and-export.md](references/outputs-and-export.md) for workspace files, videos, OBJ/texture outputs, and dependency ownership.
- Read [references/troubleshooting.md](references/troubleshooting.md) for common refine/export failures.
- Run [scripts/build_refine_export_commands.py](scripts/build_refine_export_commands.py) to generate copyable commands.

## Command Pattern

```bash
python /path/to/skill/sub-skills/refinement-and-export/scripts/build_refine_export_commands.py \
  --workspace NAME --ref-path REF_ALPHA.png --text "object prompt" --save-mesh
```

The helper emits:

- a refine command using `--final --refine` to satisfy the inspected source nesting;
- a test render command with `--test`;
- an optional mesh export command with `--test --save_mesh`.

## Safety and Verification

- Do not start refinement unless the coarse workspace has checkpoints and the user accepts additional GPU time.
- Do not promise OBJ texture export until `xatlas` and `nvdiffrast` import in the runtime environment.
- Treat PyTorch3D/Open3D/contextual-loss import errors as refinement/export blockers, not generic Python issues.
- Check generated outputs before deleting or rerunning a workspace. Interrupted training can still leave useful checkpoints.

Files in this skill

  • SKILL.md2.5 KB
  • references/outputs-and-export.md2.1 KB
  • references/troubleshooting.md1.4 KB
  • references/workflow.md2 KB
  • scripts/build_refine_export_commands.py2.6 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.