chess board photos

[geminikop]

Hi Temuge, Thank you for your nice work and sharing. I read your post about camera calibration in detail and found your link to calibration frames didn't work. Could you please post them again so I can download? That will be very helpful. I will appreciate your any response.

Thanks, Rui

[TemugeB]

Hey thanks for letting me know. I was cleaning out my google drive and accidentally deleted the frames. So those frames are gone. Could you try with these: link? Let me know if this works, I will update the blog post with these frames if they do.

[geminikop]

Hey thanks for letting me know. I was cleaning out my google drive and accidentally deleted the frames. So those frames are gone. Could you try with these: link? Let me know if this works, I will update the blog post with these frames if they do.

The link works. I appreciate that. Thanks again.

[TemugeB]

No I mean did the calibration code work fine with the new frames?

[geminikop]

No I mean did the calibration code work fine with the new frames? Yes it works fine

[geminikop]

But the rmse return by stereo_calibrate func is >0.5. Is that good?

[TemugeB]

How much was it? I got 0.502, which is ok.

[geminikop]

I got the same. So what range is good usually?

[TemugeB]

Ideally you would want less than 0.5. I don't think I managed less than 0.3. Due to camera resolution and pixel to floating point conversions, there will always be some error, even if the calibration pattern is correctly obtained. In any case, 0.5 is acceptable and you can try different frames until you go below it. Technically what it is measuring is the error of re projection in pixel space. So an average per pixel error of 0.5 is not bad at all. In other words you have about 50% probability that the correct pixel is shifted by one pixel.

[geminikop]

I get it! Thanks.