一键在 Manus 中运行任何 Skill

optics-align

星标8

分支2

更新时间2026年6月25日 19:07

Align an optical bench in the Blender Optics Simulator — steer the beam onto its target, recover a dark port, or null an interferometric tilt using the influence-matrix solvers. Use when asked to "align the bench", "get the beam on the detector", "walk the beam", "null the tilt / fringes", "maximize coupling", or "/optics-align".

安装

用 Codex 或 Claude 帮你安装复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。

在 Manus 中运行

来源

emircbngl

emircbngl/blender-optics-simulator

打开 GitHub 仓库查看创作者相关仓库

下载

在 Manus 中运行

optics-align: close the loop onto the target

Alignment solvers MUTATE the trace (the bench moves) — unlike inspect/build's read-only steps. That is expected; always report what moved and by how much.

Workflow

See the miss first — inspect_beam(target) (is there power on target? where is it landing?) and diagnose() (beam_clipped / dark_detector / mount_limit). Quantify the error before solving.
Pick the solver:
- Gross steering onto a target — auto_align(...) (the generic influence-matrix aligner: it calibrates a Jacobian over the available knobs and drives the beam onto the target element) or align_element(name, target) for a single steering mirror.
- Interferometric tilt / fringe null — tilt_null(...) (drives the inter-arm tilt to zero so the fringes go null/flat).
- Mode-matching into a cavity/fiber — mode_match (q-solver) + check coupling_efficiency.
Re-verify — re-trace and confirm with inspect_beam(target) (power recovered, beam centred) and diagnose() (the gate that fired is now clear). For a coupling task, read coupling_efficiency.
Report — what moved (element, DOF, amount), the before/after metric (power, tilt, η), and any mount_limit you hit (a DOF ran out of travel → coarse-reposition first, then re-solve).

Disciplines

A solver makes the trace no longer byte-identical — that's the point of aligning; say so.
If diagnose() shows mount_limit, the DOF is range-exhausted: coarse-reposition the mount (place_relative) to recentre its travel, THEN re-run the fine solver.
A "dark" port may be a healthy destructive-interference null, not a miss — check inspect_beam (per-arm power high, they cancel by phase) before "fixing" it. When in doubt, /optics-correct.

同仓库更多 Skills

同仓库

optics-scope

emircbngl/blender-optics-simulator

Scope an optical-simulation request honestly — map "can this bench simulate X?" to a tier ((a) reproducible now / (b) via wave.py single-plane FFT-PSF / (c) needs an engine we lack), then cite the oracle-verified case or name the missing engine + the external tool to use. Use when asked "can it simulate …", "is X in scope", "what can/can't this engine do", "do you support FDTD / BPM / Monte-Carlo / hologram reconstruction / turbulence / g²", or "/optics-scope". Anti-overclaim only: this adds ZERO simulation capability.

2026-06-278

optics-build

emircbngl/blender-optics-simulator

Build or extend an optical bench in the Blender Optics Simulator — start from a canonical example or add and place components, set their parameters and mounts, and verify the result. Use when asked to "build a bench", "set up an interferometer / telescope / AO loop", "add a lens/mirror/beamsplitter", "assemble a cage/tube/rail system", or "/optics-build".

2026-06-258

optics-correct

emircbngl/blender-optics-simulator

Run the correction-as-FEEDBACK loop on an optical bench — surface the proposed fixes, then judge each against user intent (refuse / partial / accept) rather than blindly auto-fixing. Use when asked to "fix the bench", "what's wrong and correct it", "clean up the alignment/energy", "review for problems and fix", or "/optics-correct". Never silently "improve" a bench the user built on purpose.

2026-06-258

optics-inspect

emircbngl/blender-optics-simulator

Inspect what an optical bench is actually doing — a structured, READ-ONLY numeric sweep of the Blender Optics Simulator scene (topology, per-element beam state, per-optic behaviour, sensor reads). Use when asked "what is this bench doing", "inspect the beam", "report the optical state", "is the beam aligned", "what does this detector see", or "/optics-inspect". The golden rule: never GUESS what the beam is doing — READ it.

2026-06-258

optics-sensor-render

emircbngl/blender-optics-simulator

Read and render what an optical sensor actually measures in the Blender Optics Simulator — a finite sensor's aperture capture, the modal wavefront (Zernike RMS, incl. the beam's own curvature defocus), and the dense zonal surface-figure map. Use when asked "what does the sensor / WFS see", "render the wavefront", "sensor render", "show the surface figure", "measure the aberration", or "/optics-sensor-render".

2026-06-258

name	optics-align
description	Align an optical bench in the Blender Optics Simulator — steer the beam onto its target, recover a dark port, or null an interferometric tilt using the influence-matrix solvers. Use when asked to "align the bench", "get the beam on the detector", "walk the beam", "null the tilt / fringes", "maximize coupling", or "/optics-align".

optics-align: close the loop onto the target

Alignment solvers MUTATE the trace (the bench moves) — unlike inspect/build's read-only steps. That is expected; always report what moved and by how much.

Workflow

See the miss first — inspect_beam(target) (is there power on target? where is it landing?) and diagnose() (beam_clipped / dark_detector / mount_limit). Quantify the error before solving.
Pick the solver:
- Gross steering onto a target — auto_align(...) (the generic influence-matrix aligner: it calibrates a Jacobian over the available knobs and drives the beam onto the target element) or align_element(name, target) for a single steering mirror.
- Interferometric tilt / fringe null — tilt_null(...) (drives the inter-arm tilt to zero so the fringes go null/flat).
- Mode-matching into a cavity/fiber — mode_match (q-solver) + check coupling_efficiency.
Re-verify — re-trace and confirm with inspect_beam(target) (power recovered, beam centred) and diagnose() (the gate that fired is now clear). For a coupling task, read coupling_efficiency.
Report — what moved (element, DOF, amount), the before/after metric (power, tilt, η), and any mount_limit you hit (a DOF ran out of travel → coarse-reposition first, then re-solve).

Disciplines

A solver makes the trace no longer byte-identical — that's the point of aligning; say so.
If diagnose() shows mount_limit, the DOF is range-exhausted: coarse-reposition the mount (place_relative) to recentre its travel, THEN re-run the fine solver.
A "dark" port may be a healthy destructive-interference null, not a miss — check inspect_beam (per-arm power high, they cancel by phase) before "fixing" it. When in doubt, /optics-correct.