add-mouse-driven-orbit
Add restrained mouse-driven orbit and parallax depth to a Three.js hero by damping one pointer target and splitting it across camera translation, look-at, and small object rotations.
Install / Use
npx skills add MengTo/Skills --skill add-mouse-driven-orbitInstalls into whichever agent you are using.
SKILL.md
Installable skill definition
Quality Score
Category
Development & EngineeringSupported Platforms
Our assessment of add-mouse-driven-orbit
add-mouse-driven-orbit scores 92/100 on our quality scale, 868th of 4,700 Development & Engineering skills we index (top 19%).
Its SKILL.md is 5.8 KB long, well organised into 8 sections with 4 code examples: a solid amount of guidance for an agent.
With 6,252 GitHub stars, it is one of the more widely adopted skills in the catalogue.
Maintenance, license and trust
- The repository was last updated about 2 months ago, so add-mouse-driven-orbit is actively maintained.
- Our last check on 2026-10-05 found the source still online.
- It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
- Its trust signals score 100/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.
AI review by kimi-k2.7-code on 2026-10-02. Automated pattern scan on 2026-10-02. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
add-mouse-driven-orbit compared with similar skills
All 4 of these similar skills score higher than add-mouse-driven-orbit; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| add-mouse-driven-orbit (this skill)by MengTo | 92 | 6.3k | 54d ago | SKILL.md |
| taste-skillby Leonxlnx | 100 | 116.8k | 8d ago | CLAUDE.md |
| ai-job-searchby ai-job-search-dev | 100 | 72.1k | today | CLAUDE.md |
| claude-howtoby luongnv89 | 100 | 47.3k | 2d ago | CLAUDE.md |
| lobehubby lobehub | 100 | 85.6k | today | MCP Server |
Frequently asked questions
- How do I install add-mouse-driven-orbit?
- Run
npx skills add MengTo/Skills --skill add-mouse-driven-orbit. The install tabs above show the steps for each supported agent. - Which AI agents does add-mouse-driven-orbit work with?
- It is written for Universal, as a SKILL.md file. Other agents that read the same format can often use it too.
- Is add-mouse-driven-orbit safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It is MIT-licensed and scores 100/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is add-mouse-driven-orbit still maintained?
- The repository was last updated about 2 months ago, so add-mouse-driven-orbit is actively maintained.
Skill content
View source on GitHubname: add-mouse-driven-orbit description: Add restrained mouse-driven orbit and parallax depth to a Three.js hero by damping one pointer target and splitting it across camera translation, look-at, and small object rotations. Use for passive cinematic 3D heroes, pointer-responsive scenes, organic model parallax, and interactive depth where OrbitControls would feel like a product viewer.
Add Mouse-Driven Orbit
Turn one damped pointer target into a shallow camera arc and smaller object rotations. Reach for threejs with OrbitControls when the user must inspect a product directly; reach for build-game-camera-controls for drag, zoom, occlusion, or gameplay cameras. This Skill is passive cinematic depth, not direct manipulation.
Extracted from inner-green-3d.html, where a procedural moss root had to turn toward the pointer without sliding away from the headline, cards, and pinned silhouette landmarks.
Record intent, not layout
In the pointer handler, write normalized coordinates only:
function recordPointer(event) {
if (event.pointerType === "touch") return;
target.x = (event.clientX / viewport.width) * 2 - 1;
target.y = (event.clientY / viewport.height) * 2 - 1;
}
Cache the interactive rectangle from ResizeObserver. Do not call getBoundingClientRect() for every layer on every pointermove; synchronous layout turns a light effect into frame spikes.
On pointerleave, set the target back to 0, 0. On coarse pointers, keep the authored center pose. Passive orbit should never require touch dragging to reveal content.
Dampen in the frame loop
The source uses 0.055 per 60 Hz frame. Preserve that feel across refresh rates:
const alpha = 1 - Math.pow(1 - 0.055, dt * 60);
smooth.x += (target.x - smooth.x) * alpha;
smooth.y += (target.y - smooth.y) * alpha;
Clamp dt to 1/30 s and reset the time base after resume. A fixed per-frame lerp feels heavy at 30 Hz and twitchy at 120 Hz; an unclamped delta jumps after backgrounding.
Stop style or uniform writes once the rounded value settles. Three decimals are finer than one pixel of the landed travel:
const x = Math.round(smooth.x * 1000) / 1000;
const y = Math.round(smooth.y * 1000) / 1000;
if (x !== lastX || y !== lastY) publish(x, y);
Split the motion
Use opposing, unequal layers so the scene pivots rather than translates as one slab:
camera.position.x = -smooth.x * 26;
camera.position.y = smooth.y * 16;
camera.lookAt(camera.position.x * 0.42, camera.position.y * 0.42, 0);
nearGroup.rotation.y = smooth.x * 0.055;
nearGroup.rotation.x = smooth.y * 0.026;
farGroup.rotation.y = smooth.x * 0.030;
Treat these as the landed values for a scene framed in stage-pixel world units:
| layer | horizontal | vertical | reason | | --- | ---: | ---: | --- | | camera translation | -26 | +16 | establishes the shallow arc | | camera look-at carry | 42% | 42% | keeps the subject near its pinned composition | | near object yaw | 0.055 rad | — | exposes surface depth without showing its flank | | near object pitch | — | 0.026 rad | prevents the top surface from flattening | | far object yaw | 0.030 rad | — | separates planes without matching the foreground |
Do not rotate everything by the same amount. Equal movement reads as a flat poster following the cursor. Do not aim the camera at the fixed origin while translating it; the subject visibly slides away from the layout.
Apply CSS parallax from the same smooth pair, but give text and controls smaller depth coefficients than the 3D form. Keep transforms free for parallax; use clip, opacity, or child wrappers for unrelated reveals so animations do not overwrite each other.
Preserve framing across sizes
Build the center pose first at every breakpoint. Recalculate camera aspect and any stage-to-world scale from a ResizeObserver, guard zero-sized roots, then apply orbit offsets. Verify both extreme pointer corners: no copy collision, card clipping, or exposed empty edge.
Use keyboard-operable X and Y range controls in a demo or configurator. They prove the orbit is parameterised and give non-pointer users access to the same authored states. Keep focus visible and announce motion-mode changes.
Respect motion and lifecycle
- Under
prefers-reduced-motion: reduce, render a designed three-quarter still at approximatelyx=.28, y=-.12; keep the range controls live and redraw their selected still without interpolation. - Pause while hidden or offscreen and resume with a reset time base.
- Cap DPR at 2 and clamp
dtto 1/30 s. - Keep the canvas decorative unless the 3D object carries information. Preserve all labels and controls in semantic HTML.
- On teardown, cancel the frame, disconnect observers, remove pointer/media listeners, and dispose Three.js resources.
State the cost honestly
The orbit math is negligible. The expensive work is the scene already being redrawn and any DOM style invalidation layered on top. Share one frame loop, round settled values, avoid layout reads in pointer handlers, and profile the underlying draw before removing the interaction. Moving the camera does not make a dense scene cheaper.
Verify
- Compare center and four-corner poses with the source at 1440×900.
- Confirm camera and objects move by different, opposing amounts.
- Move quickly across the viewport; the scene must ease through the path, not snap.
- Leave the viewport; confirm a smooth return to center.
- Test 390×844 and coarse/touch input; center composition remains complete.
- Tab through X/Y controls, change them with arrow keys, and confirm visible focus.
- Test
?reduced=1: stable designed still, no autonomous easing, live controls. - Hide/show and scroll away/back; confirm no jump and only one frame loop.
- Confirm a clean console at both sizes.
Use demo/index.html as the working proof and demo/PROMPT.md to recreate or remix it.
Related Skills
taste-skill
116.8kAgent skills that give coding agents taste: art direction, reference images and a browser check instead of generic AI pages
ai-job-search
72.1kThe job search that runs on your machine. AI job application framework built on Claude Code: evaluate postings, tailor CVs, write cover letters, prep interviews. Fork it and own it.
claude-howto
47.3kA visual, example-driven guide to Claude Code — from basic concepts to advanced agents, with copy-paste templates that bring immediate value.
lobehub
85.6k🤯 LobeHub - Super-Agents on Your Team. Add super-agents to the chats your team already uses.They take on work by themselves and get better every week.
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
