Martin Garbade

So looking at [genSUNCGdataScript.m](https://github.com/shurans/sscnet/blob/master/matlab_code/genSUNCGdataScript.m#L54-L55) brought some insights: ``` % resave depth map with bit shifting depthRaw = double(imread(filename))/1000; saveDepth (depthRaw,newFilename); function saveDepth (depth,filename) depth(isnan(depth)) =0; depth =single(depth)*1000; depthVis = uint16(depth);...

Ah ok. Thanks for that hint. This is the function I'm using to read the image. It's an extract from `ReadDepthImage`. Only assumption I'm making is that `Dtype` == float...

Well the problem, if I understand it correctly, is that by using the SUNCG toolbox I would have to generate my own new dataset, but I'd rather need the RGB...

For 2. I still can't really see through this: Lets take this image as example:  The current output for the 3D scene would look...

Getting closer: Changing ``` invalidPixInd = gridPtsPixX size(depthInpaint,2) | ... gridPtsPixY size(depthInpaint,1); ``` to ``` invalidPixInd = []; ``` gives the entire scene:  Rotating...

Rotating vertices of the CAD models only ([here](https://github.com/shurans/sscnet/blob/master/matlab_code/perpareNYUCADdata.m#L104) and [here](https://github.com/shurans/sscnet/blob/master/matlab_code/perpareNYUCADdata.m#L147)) gives: 

I just wanted to kindly re-ask. I'm still going through perpareNYUCADdata.m and try to figure out how to rotate back the scene such that the 3D scene is aligned with...

Part Solution: It seems like [this](https://github.com/shurans/sscnet/blob/master/matlab_code/perpareNYUCADdata.m#L119-L142) entire if-clause is not necessary / causing the trouble (not entirely sure why). If it is set to false one can rotate the scene...

I don't understand the question: You cannot find the RGB images corresponding to the 'bin' files for NYU?

I'm sorry I haven't worked with SUNCG yet only with the NYU data. It would indeed be interesting to have aligned RGB and bin files for SUNCG. [This](https://github.com/shurans/sscnet/issues/18#issuecomment-313917134) answer by...