Some confused about RT when I run the code

[maoLB]

Hi, anuranbaka
I have successfully run the stuff.I'm try to change the code into real-time camera input.I have seen the #12 "about real-time camera input" Issues and almost know something about DTAM. The stuff run by dataset already give the RT values.So I'm confused about that how do you get the RT values?Is it mean that you find RT values by the usual OpenCV convention: point_in_pixels=cameraMatrix[R|T]point_in_world_coordinates?Do you mean the function solvePNP?Thank you very much if you can help me.

[maoLB]

Hi, anuranbaka
First,i use the RT values by ORB_SLAM.But i can not get the correct depth image.
So i change the dateset by another slow motion video. Reference to http://www.doc.ic.ac.uk/~ahanda/HighFrameRateTracking/downloads.html.They have both dataset type.But when i run the stuff again,i get the wrong depth image. Do this code have a great robust for video dataset or if i have make mistakes in your code?

[lianuo]

I have same problem, I donot know how to fix it.