I’m having some problems aligning image sets taken with my Olympus E-M5 II and 12-40 2.8 lens. In most cases RC just misses most of my pictures. When it does align, it’s very inaccurate. Agisoft PS doesn’t have any problems with it, and it aligns perfectly. I’ve tried adding the camera to the sensordb, and the 35mm eqv. focal length is recognised correctly, but I only got marginal improvements.
It works fine with a Nikon d3300 that I tried, so the photos should be ok.
M4/3 cameras are great for photogrammetry: they usually come with flipout lcds, they have a stabilised sensor for challenging lighting, the lenses tend to be quite sharp and they have larger DOF than apsc/FF cameras, so I’d like to stick with it if possible.
Thanks for the advice! I was using default settings and looks like it works with these changes. I used K+Brown4 with tangential 2 though, just to make sure.
Still have a couple of questions: would using the sensor stabilisation negatively effect the alignment/reconstruction?
Also, mft cameras have automatic lens correction applied to the raw, and the lenses are designed with this in mind. Would disabling it help, or make it worse?
Thanks for the advice! I was using default settings and looks like it works with these changes. I used K+Brown4 with tangential 2 though, just to make sure.
If you use “small” sensors then its advisable to use these settings for distortion as the sensor and lenses have many optical issues so its advisable to use these settings. ( Brown4 - 4 params of distortion are calculated )
Still have a couple of questions: would using the sensor stabilization negatively effect the alignment/reconstruction?
Also, mft cameras have automatic lens correction applied to the raw, and the lenses are designed with this in mind. Would disabling it help, or make it worse?
Sensor stabilization have minimal impact on alignment / recon. It can cause issues if the “shaking” of the camera get over certain treshold, but frankly that are cases of very strong UAV movement and stabilization issues. In general its not a issue as i say.
Lens corrections, in principle I am against in-camera corrections as they are not so precise ( onboard CPU processing power + used algos ) and they are ok for photographer or videographer use cases but not so much for photogrammetry.