diff --git a/.agents/skills/splat-studio-control/SKILL.md b/.agents/skills/splat-studio-control/SKILL.md index 4b84e2c..28936c4 100644 --- a/.agents/skills/splat-studio-control/SKILL.md +++ b/.agents/skills/splat-studio-control/SKILL.md @@ -62,10 +62,11 @@ CLI stdout (the summary table for `-m`). ### `convert` formats `ply`, `compressed-ply`, `sog`, `sog-unbundled`, `lod`, `spz`, `glb`, `csv`, -`html`. `options`: `iterations`, `spzVersion`, `decimate`, -`decimateAlgorithm` (`adaptive`|`uniform`), `filterNaN`, -`device` (`auto`|`cpu`), `lodLevels`/`lodKeepPercent`/`lodChunkCount`/ -`lodChunkExtent`/`lodFiles`, and `params` (for `.mjs` generator inputs). +`html`. `options`: `iterations`, `webpEffort` (0-9, SOG/HTML/LOD), `spzVersion`, `decimate`, +`decimateAlgorithm` (`adaptive`|`uniform`), `scratchDir` (blank = the project folder), +`filterNaN`, `device` (`auto`|`cpu`), `lodLevels`/`lodKeepPercent`/`lodChunkCount`/ +`lodChunkExtent`/`lodChunkMin`/`lodFileCount`/`lodFiles`, `image` (webp renders; incl. +`dofSamples` with `fStop`, `shutter` default 1), and `params` (for `.mjs` generator inputs). `uniform` is the only decimation algorithm that runs on CPU once a scene splits into multiple blocks (roughly 2M+ gaussians); adaptive with `device: "cpu"` fails there. diff --git a/.agents/skills/splat-studio-workflows/SKILL.md b/.agents/skills/splat-studio-workflows/SKILL.md index b4a81ad..e0fa8cb 100644 --- a/.agents/skills/splat-studio-workflows/SKILL.md +++ b/.agents/skills/splat-studio-workflows/SKILL.md @@ -24,7 +24,7 @@ After every job-starting call: `jobs(action:"wait", id, timeout_ms)` → on `don On `{refused:true}`: raise voxelSize or crop via filterBox/filterSphere; `overridePreflight:true` only with user buy-in. - **"A beauty shot"** → `render_image` with CLI-space `"x,y,z"` camera strings; DoF via - `fStop`+`focusDistance`. Frame it first *(editor)*: `render_pose(action:"set")` + + `fStop`+`focusDistance` (+ `dofSamples`, default 32; each sample is a full render). Frame it first *(editor)*: `render_pose(action:"set")` + `viewport_screenshot(max_width:800)`. - **"Wrong size / wrong origin"** *(editor)* → measure/origin recipes below. - **"Share with a non-technical person"** → `convert(format:"html")` (single self-contained file). @@ -42,7 +42,8 @@ the job LOG, no file). (+ a `build-meta.json` recipe — read via `inspect(target:"lod_recipe")`; pre-feature bundles return `not-found`); deleting that entry deletes the folder. Add `decimateAlgorithm:"uniform"` to decimate every derived level at a flat rate instead of -the adaptive default. +the adaptive default. If a bake dies with "non-finite position", re-run with +`filterNaN:true` — the LOD writer aborts on the first bad gaussian. **Environment-shell LOD (combine):** lighter levels first (`convert(decimate:"50%")` per level), then `build_lod(mode:"combine", lodFiles:[...], lodEnvFlags:[...])` — env flag = always-visible @@ -53,7 +54,8 @@ resolution:"4096x2048"})` — 2:1 aspect REQUIRED (else the CLI errors); OMIT fo options (rejected, not ignored). Result can feed `set_view_option(option:"skybox", value:)`. **Motion blur:** add `cameraEnd`/`lookAtEnd` + `shutter:0.5, motionSamples:64` to a normal -render. Keep the move 10–20 cm. +render (`shutter` defaults to 1). With DoF on too, cost = `motionSamples` × `dofSamples` +renders. Keep the move 10–20 cm. **Scale to real (editor):** `load_into_viewport` → `panel(open, "panel-edit")` → `measure(action:"measure", points:[[A],[B]])` on a known span (passing points turns measure diff --git a/.claude/skills/splat-studio-control/SKILL.md b/.claude/skills/splat-studio-control/SKILL.md index 4b84e2c..28936c4 100644 --- a/.claude/skills/splat-studio-control/SKILL.md +++ b/.claude/skills/splat-studio-control/SKILL.md @@ -62,10 +62,11 @@ CLI stdout (the summary table for `-m`). ### `convert` formats `ply`, `compressed-ply`, `sog`, `sog-unbundled`, `lod`, `spz`, `glb`, `csv`, -`html`. `options`: `iterations`, `spzVersion`, `decimate`, -`decimateAlgorithm` (`adaptive`|`uniform`), `filterNaN`, -`device` (`auto`|`cpu`), `lodLevels`/`lodKeepPercent`/`lodChunkCount`/ -`lodChunkExtent`/`lodFiles`, and `params` (for `.mjs` generator inputs). +`html`. `options`: `iterations`, `webpEffort` (0-9, SOG/HTML/LOD), `spzVersion`, `decimate`, +`decimateAlgorithm` (`adaptive`|`uniform`), `scratchDir` (blank = the project folder), +`filterNaN`, `device` (`auto`|`cpu`), `lodLevels`/`lodKeepPercent`/`lodChunkCount`/ +`lodChunkExtent`/`lodChunkMin`/`lodFileCount`/`lodFiles`, `image` (webp renders; incl. +`dofSamples` with `fStop`, `shutter` default 1), and `params` (for `.mjs` generator inputs). `uniform` is the only decimation algorithm that runs on CPU once a scene splits into multiple blocks (roughly 2M+ gaussians); adaptive with `device: "cpu"` fails there. diff --git a/.claude/skills/splat-studio-workflows/SKILL.md b/.claude/skills/splat-studio-workflows/SKILL.md index b4a81ad..e0fa8cb 100644 --- a/.claude/skills/splat-studio-workflows/SKILL.md +++ b/.claude/skills/splat-studio-workflows/SKILL.md @@ -24,7 +24,7 @@ After every job-starting call: `jobs(action:"wait", id, timeout_ms)` → on `don On `{refused:true}`: raise voxelSize or crop via filterBox/filterSphere; `overridePreflight:true` only with user buy-in. - **"A beauty shot"** → `render_image` with CLI-space `"x,y,z"` camera strings; DoF via - `fStop`+`focusDistance`. Frame it first *(editor)*: `render_pose(action:"set")` + + `fStop`+`focusDistance` (+ `dofSamples`, default 32; each sample is a full render). Frame it first *(editor)*: `render_pose(action:"set")` + `viewport_screenshot(max_width:800)`. - **"Wrong size / wrong origin"** *(editor)* → measure/origin recipes below. - **"Share with a non-technical person"** → `convert(format:"html")` (single self-contained file). @@ -42,7 +42,8 @@ the job LOG, no file). (+ a `build-meta.json` recipe — read via `inspect(target:"lod_recipe")`; pre-feature bundles return `not-found`); deleting that entry deletes the folder. Add `decimateAlgorithm:"uniform"` to decimate every derived level at a flat rate instead of -the adaptive default. +the adaptive default. If a bake dies with "non-finite position", re-run with +`filterNaN:true` — the LOD writer aborts on the first bad gaussian. **Environment-shell LOD (combine):** lighter levels first (`convert(decimate:"50%")` per level), then `build_lod(mode:"combine", lodFiles:[...], lodEnvFlags:[...])` — env flag = always-visible @@ -53,7 +54,8 @@ resolution:"4096x2048"})` — 2:1 aspect REQUIRED (else the CLI errors); OMIT fo options (rejected, not ignored). Result can feed `set_view_option(option:"skybox", value:)`. **Motion blur:** add `cameraEnd`/`lookAtEnd` + `shutter:0.5, motionSamples:64` to a normal -render. Keep the move 10–20 cm. +render (`shutter` defaults to 1). With DoF on too, cost = `motionSamples` × `dofSamples` +renders. Keep the move 10–20 cm. **Scale to real (editor):** `load_into_viewport` → `panel(open, "panel-edit")` → `measure(action:"measure", points:[[A],[B]])` on a known span (passing points turns measure diff --git a/README.md b/README.md index 22781c9..470909c 100644 --- a/README.md +++ b/README.md @@ -7,7 +7,7 @@ all in a dockable, Unity/Unreal-style tab editor you can rearrange and save per workspace. -**Built with [PlayCanvas](https://github.com/playcanvas/engine) `2.21.3` · [@playcanvas/splat-transform](https://github.com/playcanvas/splat-transform) `3.2.0`** +**Built with [PlayCanvas](https://github.com/playcanvas/engine) `2.24.0` · [@playcanvas/splat-transform](https://github.com/playcanvas/splat-transform) `3.10.3`** [![License: Apache 2.0](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](LICENSE) ![Platform: Windows](https://img.shields.io/badge/platform-Windows-blue.svg) @@ -236,11 +236,11 @@ Then in the browser: `--resolution`/`--background`). **📷 from viewer** seeds the camera from the 3D view. **Projection** switches pinhole ↔ equirectangular 360° panorama. **Depth of field** (`--f-stop`/`--focus-distance`, pinhole only) and **motion blur** - (`--camera-end`/`--shutter`/`--motion-samples`) are exposed too. -- **Device** dropdown — choose the GPU adapter (listed via `-L/--list-gpus`) or + (`--camera-pos-end`/`--shutter`/`--motion-samples`) are exposed too. +- **Device** dropdown — choose the GPU adapter (listed via `--list-gpus`) or CPU (`-g`). **Verbose** adds `--verbose --memory` diagnostics to the Job log. -- **HTML viewer** output gains **Unbundled** (`-U`, separate files) and a - **Viewer settings** JSON (`-E`). An **.lcc** input gains **LOD levels** (`-O`). +- **HTML viewer** output gains **Unbundled** (`--unbundled`, separate files) and a + **Viewer settings** JSON (`--viewer-settings`). An **.lcc** input gains **LOD levels** (`--select-lod`). ## Edit: measure-to-scale & set origin diff --git a/client/index.html b/client/index.html index 72cfa88..28c2fae 100644 --- a/client/index.html +++ b/client/index.html @@ -121,6 +121,7 @@

Export

+
-
+ -