Control Points

[caglarenes]

Hello. I'm trying to calculate a face model but results of my model isn't very good. I did several times before at another people succesfully but my last 3Dmodel has no nose for example. I take shots with 24mm lens and 5D mark III. My images are good but there is another problem I think. (I think 24mm lens is mistake. I'm alsa try to calibrate my lens with 3DLapyx)

And here is my question: If I add control points to my photos at person's beauty spots, acne etc. does it change output quality?

I trying different keypoint presets etc. but each test take a lot of times of course. When I add control points, program can't show equiline lines correct positions. If I show correct points on every photo, does it help the 3D process?

[Andrea Alessi]

Hello caglarenes,

If you're seeing that a control point doesn't fall on the epipolar line, this could have happened:

- the subject moved or
- one (or more) cameras have not been oriented correctly or
- you are manually feeding a wrong camera calibration

You can use control points as constraints and optimize the camera parameters, but if the subject moved (which in human scanning is honestly the most common problem) you might still have problems. A good idea is usually to mask out the background, so that you can "even out" the discrepancy between the background and your subject.

Make also sure that all cameras have been oriented correctly - if you see a specific camera in the workspace that is wrong, it might help to remove it rather than keeping it.

If you can share your dataset with us, i can have a look at it

cheers,

Andrea