Hacking around with a RealSense depth camera and Python
erikorndahl.comI almost couldn't watch the videos on this website because of their size, 50MB each! Just a simple `ffmpeg -i in.mp4 -crf 23 out.mp4` reduces the size by a factor of 10 without noticeable difference. If the blog author is reading this, please consider using video compression, not everybody has super fast internet.
I thought my internet was down as video would not start on my phone. Lol, gonna be a lot of wasted bandwidth with this trending..
Yikes! Never looked at the file size.. let me fix that
Thank you! Learned a valuable lesson today.
The device seems to be similar to the original Kinect (unstructured light stereo), rather than the second generation (time of flight).
You can still buy the Kinect Xbox One sensor for $89. It's a good device, even though it never caught on for gaming.
RealSense D400 cameras support assisted stereo, (un)structured light, and ToF.
https://realsense.intel.com/introducing-intel-realsense-d400...
a bit disappointed with the image quality. The original realsense r200 was probably the worst depth camera I've tried - it was barely better than two webcams and parallax processing. The new cameras doesn't look to be much better from the video..
I wish there was better driver support for the kinect2, still the best depth camera you can buy imo.
That wasn't the best example. Lots of noise in that high depth shot of a messy cafe + quick movement from the fan. If you look at the end of the new clip, even with high accuracy filtering, you're getting a pretty clean depth image closer up. I'm pretty happy with the D415 so far.
Have you tried the much newer D400 cameras? I'd love to know how they compare to the Kinect 2.
haven't tried it, but I'd be surprised if they're comparable. Afaik all the non-kinect depth cameras are structured light vs ToF with the kinect2.
The one in the Lenovo Phab 2 Pro is a 35k pixel ToF, but artificially constrained in power and thus reach to live on a phone battery.