browser-renderer-neural-network
browser-renderer-neural-network copied to clipboard
This is a browser based renderer for outputs of 3D neural networks
ServerMeshViewer
Description:
This is a WebGL based renderer for following file formats:
-
vox, vox2, voxnoc
(dense voxel grids) -
ply, obj
(meshes) -
wrf
(wireframe files) -
.json
(scene files, can be used in mitsuba) -
jpg, png, gif
(images)
If you want more file formats to be visualized, then open an issue and I'll write a shader for it as soon as I find some time
How-To Use
Just pase the absolute path of your file in the browser and it will render that file automatically:
localhost:8080/Viewer/[abspath-to-file]
Why?
- No copying from server to local machine needed
- No meshlab needed
- No remote desktop needed
- View outputs from anywhere
- All you need is a browser
- It's super fast to view predictions in the browser
I have seen people copy outputs to their local machine and visualize with meshlab lol. I believe this will save a lot of time.
Example:
Open firefox or chrome and goto:
localhost:8080/Viewer/home/amo/neural-network/prediction.voxnoc
.voxnoc (voxelwise rgb color) |
---|
![]() |
.json (scene fileformat) |
![]() |
.ply (this is a black bunny lol) |
![]() |
.wrf (wireframe format) |
![]() |
.vox2 (this is a heatmap) |
![]() |
Get started
- Clone this repo
-
cd repo-name
(enter downloaded repository folder) - Install
nodejs
(=npm
) from https://nodejs.org/en/ - Run
npm install
for client-side- This will install all dependencies specified in
package.json
- This will install all dependencies specified in
-
cd ./server/ && npm install
for server side. - Run
./build.sh
to compile - Run
./server/run.sh
to start the server.- Note that this assumes that you use
bash
- Note that this assumes that you use
- Go to localhost:8080/Viewer/[abspath-to-file]