"Sparse" alignment image upscaling?

[floe]

Hi again, I'm sort-of misusing the issue tracker as a forum now, feel free to close if not intended :-)

That being said, I noticed that e.g. when using rs::stream::depth_aligned_to_color at a higher color resolution, you get a "sparse" image. This is, of course, expected as https://github.com/IntelRealSense/librealsense/blob/master/src/image.cpp#L356 only maps one single source pixel to a single target pixel.

In libfreenect2, we had implemented a naive-but-mostly-sufficient nearest-neighbor upscaling procedure for this case so that you can get a dense, color-aligned depth image. Would you be interested in a PR implementing a similar behaviour for librealsense? Then I'll prepare one.

[ddiakopoulos]

Hi Florian, thanks for your pull request! This is indeed functionality that is desirable and fitting for inclusion in librealsense. Give us a couple days to work with our legal team on a CLA agreement for this project.

[floe]

OK, let me know what you need. Instead of a full-scale CLA, would it be sufficient if I also declare my patch to be under Apache-2.0 license?

[sgorsten]

Satisfied by pull request #9, merged to master in commit ed4bc85.

Sorry for the long wait.