LIVENODING 1090 / SV Disparity Depth Map From Stereo Pair
In this episode, I am sharing my quick research on how to estimate and generate Depth Map using pair of stereoscopic images with Blender and Python.
This is related to disparity and parallax, see Apple presentation on this topic from last years.
If you are using Windows, there is a tool from StereoPhotoMaker developer called DMAG Depth Map Auto Generator to extract depth map from pair of stereo images, but does not work on Mac. So I am using Python OpenCV module to generate this and make it to work inside Blender via Sverchok node!
On an iPhone this kind of disparity process is actually performed in realtime when using Portrait Mode. Use app called DepthCam to see the disparity depth map. Google Pixel is also doing similar thing. With iPhone X, XR, XS, and its true depth front camera, the result of depth map is actually much more accurate and you get point cloud data. That is another story.
What you can do with this DEPTH MAP is to actually generate ”3D Photo” for Facebook.
But in Blender you can do more like separating background and foreground, and do some compositing works. Which you can all perform using some apps on iPhone or iPad. Maybe eventually we have 3D app like Blender on iPhone or iPad?