Skill

quality-settings

Recommends optimal Metashape processing settings based on photo count, RAM, and quality/speed priorities. Outputs specific parameters and tool calls for alignment, depth maps, meshing.

Python

performance

automation

npx claudepluginhub jenkinsm13/claude-plugins --plugin metashape-mcp

Tool Access

This skill uses the workspace's default tool permissions.

Preview

Recommend processing parameters tuned to the user's dataset and hardware.

SKILL.md

Similar Skills

photo-import-setup

Sets up Metashape photogrammetry projects from scratch: imports photos/GPS/EXR masks, configures fisheye/rolling shutter/multi-camera sensors, runs image quality analysis via MCP server. For new captures.

metashape-mcp

get-available-resources

Detects CPU cores, GPUs (NVIDIA CUDA, AMD ROCm, Apple Metal), memory, disk space and generates JSON recommendations for parallel processing (joblib, multiprocessing), out-of-core (Dask, Zarr), or GPU strategies (PyTorch, JAX) before intensive scientific tasks.

1 file

superpowers

get-available-resources

Detects CPU/GPU/memory/disk resources and generates JSON recommendations for parallel processing, GPU acceleration (PyTorch/JAX), or out-of-core strategies (Dask/Zarr) in scientific tasks.

1 file

scientific-skills

Stats

Stars18

Forks2

Last CommitMar 22, 2026

Actions

View Source View Plugin View on GitHub View README

Help us improve

Share bugs, ideas, or general feedback.

Optimize Quality Settings

Recommend processing parameters tuned to the user's dataset and hardware.

Information Needed

Ask the user (or read from the project):

Photo count: Number of images in the dataset

Available RAM: System memory in GB

Priority: Quality vs speed

Guidelines by Dataset Size

Setting

<100 photos

100–500

500–2000

2000+

Match downscale

1 (High)

2 (Medium)

2–4

Keypoint limit

60000

40000

Depth map quality

2 (High)

4 (Medium)

8 (Low)

Depth filter

mild

moderate

Face count

high

medium

medium/custom

RAM Considerations

<16 GB: Use downscale 4+ for depth maps, medium face count

16–32 GB: Use downscale 2–4, high face count up to ~500 photos

32–64 GB: Use downscale 1–2, high face count up to ~1000 photos

64+ GB: Can use ultra quality for smaller datasets

GPU/CPU Rule

set_gpu_config(cpu_enable=True) — BEFORE alignment (match_photos, align_cameras)

set_gpu_config(cpu_enable=False) — BEFORE everything else (depth maps, meshing, texturing, DEM, ortho)

CPU slows GPU operations. It is ONLY beneficial for alignment.

Output

Read the chunk info, then provide specific parameter values for each processing step as ready-to-use tool calls.

Optimize Quality Settings

Recommend processing parameters tuned to the user's dataset and hardware.

Information Needed

Ask the user (or read from the project):

Photo count: Number of images in the dataset
Available RAM: System memory in GB
Priority: Quality vs speed

Guidelines by Dataset Size

Setting	<100 photos	100–500	500–2000	2000+
Match downscale	1 (High)	1 (High)	2 (Medium)	2–4
Keypoint limit	60000	40000	40000	40000
Depth map quality	2 (High)	4 (Medium)	4 (Medium)	8 (Low)
Depth filter	mild	mild	moderate	moderate
Face count	high	high	medium	medium/custom

RAM Considerations

<16 GB: Use downscale 4+ for depth maps, medium face count
16–32 GB: Use downscale 2–4, high face count up to ~500 photos
32–64 GB: Use downscale 1–2, high face count up to ~1000 photos
64+ GB: Can use ultra quality for smaller datasets

GPU/CPU Rule

set_gpu_config(cpu_enable=True) — BEFORE alignment (match_photos, align_cameras)
set_gpu_config(cpu_enable=False) — BEFORE everything else (depth maps, meshing, texturing, DEM, ortho)

CPU slows GPU operations. It is ONLY beneficial for alignment.

Output

Read the chunk info, then provide specific parameter values for each processing step as ready-to-use tool calls.