Matt Romein

Curious if @zeal (Who mentioned playing with them in the oculus thread) or anyone else got the 

 cameras working with Max? Working on a project for grad school using them and the Oculus external for Max and having the worst luck with them. The 2 cameras both come up with jit.grab but only one can be grabbed at a time. Explored using the xcode files they provide to get both the camera feeds on syphon to pull them into Max but that proved fruitless as well. Any help or knowledge from other people that explored it would be awesome!

Curious if @zeal (Who mentioned playing with them in the oculus thread) or anyone else got the ovrvision cameras working with Max? Working on a project for grad school using them and the Oculus external for Max and having the worst luck with them. The 2 cameras both come up with jit.grab but only one can be grabbed at a time. Explored using the xcode files they provide to get both the camera feeds on syphon to pull them into Max but that proved fruitless as well. Any help or knowledge from other people that explored it would be awesome!


ovrvision-cameras-for-oculus-rift

I did manage to get it to work, but this was last year and now I don't recall exactly how. I hadn't calibrated them very accurately so the image was somewhat warped, but in any case the resolution was low for being so close to the eyes, and moreover the cameras are placed too close to each other for effective use as a see-through-AR type thing AFAICT.

Having said that, although the raw RGB images were not pleasant to look at, putting them through some filters was quite a lot of fun. Seeing the world around you through new lenses -- in 3D -- is pretty compelling. I remember particularly enjoying an edge-detection algorithm that was set to be way too sensitive.

I did manage to get it to work, but this was last year and now I don't recall exactly how. I hadn't calibrated them very accurately so the image was somewhat warped, but in any case the resolution was low for being so close to the eyes, and moreover the cameras are placed too close to each other for effective use as a see-through-AR type thing AFAICT. 

Having said that, although the raw RGB images were not pleasant to look at, putting them through some filters was quite a lot of fun. Seeing the world around you through new lenses -- in 3D -- is pretty compelling. I remember particularly enjoying an edge-detection algorithm that was set to be way too sensitive.


Awesome photo! Actually I did get it working. It turned out that I just needed to use the external power supply rather than just relying on the Rift's usb connection to draw power. Agreed that they are weird cameras in terms of spacing/resolution but I'm embracing the weird aesthetic they have.

I'm developing a hallucination simulator with them so they work pretty well and I've been playing with the concept of time and perception as well as trying out a lot of effects like edge-detection and optical flow shaders. It's a lot of fun and very disorienting. Not trying to do any typical AR stuff. Right now I'm doing research on camera calibration in Max so if you have any tips, links or patches you can think of that would be very helpful.

Sounds super interesting -- please share pictures / vids if you can!

Yes, the DK2's usb port is mostly useless -- only very very low power devices can use it, which are not usually what you'd want to stick on the headset.

Yes I've worked on some camera calibration externals -- OpenCV chessboard and calibrate camera wrappers. Not necessarily fully functional, but I can share if you want to try and give me feedback? OSX only currently.

Yes, the DK2's usb port is mostly useless -- only very very low power devices can use it, which are not usually what you'd want to stick on the headset. 

Yes I've worked on some camera calibration externals -- OpenCV chessboard and calibrate camera wrappers. Not necessarily fully functional, but I can share if you want to try and give me feedback? OSX only currently.


Yes I'll totally post some of the stuff I've been working on. I need to document my midterm examples this week so I'll post it once I finish the video.

Yes, if you wouldn't mind sharing it I would be very grateful! Happy to give feedback on how it's working and also curious to see how you approached the code. Thanks for your help!

The two objects I've been working on & using are here (maxhelps included): 

https://www.dropbox.com/sh/thp9cg1mtjwab7g/AAAlGXj1byRnDnT6H6ErQ4s5a?dl=0

Calibration means grabbing a few shots of a printed chessboard in front of the camera, and feeding them into the 'find chessboard' object to derive the screen space coordinates. These coordinates, and a matrix corresponding to the coordinates in real space (simple in that the chessboard is a plane, so z == 0), are fed to cv.jit.calibratecamera to derive the intrinsics.

All based on the standard OpenCV camera calibration routine.

The two objects I've been working on & using are here (maxhelps included): https://www.dropbox.com/sh/thp9cg1mtjwab7g/AAAlGXj1byRnDnT6H6ErQ4s5a?dl=0

Calibration means grabbing a few shots of a printed chessboard in front of the camera, and feeding them into the 'find chessboard' object to derive the screen space coordinates. These coordinates, and a matrix corresponding to the coordinates in real space (simple in that the chessboard is a plane, so z == 0), are fed to cv.jit.calibratecamera to derive the intrinsics. 

All based on the standard OpenCV camera calibration routine.


Thanks for sharing these Graham. Right now I'm having issues with the findchessboard object causing Max to crash whenever it receives a matrix. I can send you the error report if that helps or try to isolate the issue further.

Hmm, strange. Yes, please do send me the error report. Maybe off list?

Sure, what's the easiest way to do that?


I thought there was a private message feature on the forum, but I can't find it... just email me at graham at cycling74.com

For anyone reading, I updated the external in the dropbox link above, and Matt confirmed it no longer crashes. Will make this part of a more public release of stuff soon.

Ovrvision cameras for Oculus Rift