[Question] Can I draw a DEPTH MAP using it?

[exousiayk]

I'm studying on shoe print for shoe-tread impressions at crime scenes. So, I need to draw a depth map of shoe print.

How can I draw a depth map in python? I can't find it in Mitsuba Documentation.

*sample image

[njroussel]

Hi @exousiak

You'll want to use the aov integrator, with the "depth" attribute.

[exousiayk]

Thanks njroussel.

I have a one more question. Is it possible to use AOV input data as a .png or .jpg file instead of .xml?

[njroussel]

No ?

Maybe there's a misunderstanding here. The aov integrator and its depth map simply return the distance from your viewpoint to the shape. This therefore requires that you have a 3D model of your shoe.The XML file only serves a description of your scene and the parameters of the aov integrator.

If you're trying to recover the depth map from a single picture (not a 3D shape), you should look into other projects. Mitsuba is not well-suited for that problem.

[Monoxide-Chen]

How to write a configuration in an XML file to obtain the depth channel?
The example in the doc is as follows:

<integrator type="aov">
    <string name="aovs" value="dd.y:depth,nn:sh_normal"/>
    <integrator type="path" name="my_image"/>
</integrator>

Is dd.y:depth sufficient to extract the depth information?
Do we need the nn:sh_normal part and the <integrator type="path" name="my_image"/> part, or can they be omitted if we are only interested in the depth channel?

[njroussel]

Hi @Monoxide-Chen
The snippet below is the bare minimum.

<integrator type="aov">
    <string name="aovs" value="scene_depth:depth"/>
    <integrator type="path" name="my_image"/>
</integrator>