Visualize CFD simulations in Augmented Reality — works on phone, tablet, and desktop.
Go to: https://hiukim.github.io/mind-ar-js-doc/tools/compile
- Upload any image you want as your AR marker (a logo, a project label, etc.)
- Click Start → download the
targets.mindfile - Also save the image you uploaded as
marker.png - Put both files in the root of this repo
For each time step:
- Load your CFD data in ParaView
- Apply filters → Decimate (target ~50,000 polygons for AR)
- Color by your variable (pressure, velocity magnitude, etc.)
- File → Export Scene → GLB
- Name files:
step_001.glb,step_002.glb, ...step_050.glb
For a static simulation with one result: just export one file as step_001.glb.
- Go to your GitHub repo → Releases → Create a new release
- Tag:
v1.0(or any tag) - Drag and drop all your
.glbfiles - Publish release
- Right-click each file → Copy link address — these are your CDN URLs
Update manifest.json with your file URLs:
{
"steps": [
{ "url": "https://github.com/YOUR_USER/YOUR_REPO/releases/download/v1.0/step_001.glb", "label": "t = 0.0s" },
{ "url": "https://github.com/YOUR_USER/YOUR_REPO/releases/download/v1.0/step_002.glb", "label": "t = 0.5s" }
]
}For a static simulation with one step, just put one entry in the array — the play/pause controls will automatically hide.
Push everything to GitHub — Vercel auto-deploys.
Your repo should look like:
index.html
manifest.json
marker.png
targets.mind
vercel.json
README.md
| Action | Gesture |
|---|---|
| Scale model | Pinch (two fingers) |
| Rotate model | Drag (one finger) |
| Next step | ▶▶ button or tap track |
| Play/Pause | ▶ PLAY button |
| Desktop rotate | Mouse drag |
| Desktop scale | Mouse wheel |
- Decimate aggressively — AR doesn't need scientific accuracy in polygon count. Target 20k–100k polygons per step
- Use surface representation — volume rendering won't export to GLB. Use Contour filter for iso-surfaces
- Bake colors — make sure to include color data in the GLB export (check "Include Color" in export options)
- DRACO compression — if you install the
draco_encodertool, compress your GLBs withdraco_compressionfor 10x smaller files. The viewer supports DRACO automatically