I know noise is bad for photogrammetry even if it’s removed because it changes the image.
But a camera like Sony A7 C has very low noise and doesn’t loose detail at high iso. I don’t have it so I can’t try it but I’d like to have one if the images can be used at high iso.
Has anyone here tested the limits on high ISO with 3D reconstruction and can talk about it?
What ISO are you referring to as high? 400, 1600, 25600?
And do you mean only in the context of APS-C class sensors, or are we including phone/entry-level sUAS class sensors as well? With those sub-1" sensors, even ISO-400 visibly starts to degrade dynamic range at least.
This gives me something fun to try 
The recommendation is to use base iso, the lowest. And I like to test the boundaries so I know how far I can push it until it no longer works.
In the context of any camera really. GoPro10 can’t go above the base iso, DJI phantom 4 is usable at least up to 400.
I’ve got results from my Olympus at up to 2000 iso.
A Sony A7 C delivers very good images up to 6400 it’s seems looking a at tests.
So how far have you tried pushing the iso boundaries?
I captured an ISO-6,400 dataset from my phone earlier tonight to test. Hopefully will finish by morning.
I’m working out my X-T2 parameters and hopefully can try some reconstructions from that this week. That will be base ISO (ISO-200) and ISO-12,800.
I’ve seen a comparison between xt-3 and A7C on higher iso and the Fuji looses a lot of detail but the Sony looks like nothing changed.
There’s a real difference taking beautiful photos and photos for reconstruction.
I’m aware 
I’ve disabled IBIS also because I assume that its not a good thing that the image Center moves around in relation to the lens.
Seems like really high ISO/noisy images aren’t that much of a problem, but dynamic range loss and noise reduction algorithms (loss of detail/microtexture) might be far more detrimental to reconstruction:
A few notes:
So all things considered, I’m happy and will certainly not shy away from higher ISOs, at least in scenes that aren’t highly dynamic in terms of lightning.
That was really high ISO! I wouldn’t have pushed it that far, Sony A7 C has really low noise but I wouldn’t push it much more than 1600 because it’s starts to deteriorate after that.
Yes, very high ISO and far too much for the tiny Samsung ExoCell 12MP sensor in my phone, but I think that it helps illustrate our point, no?
I think you can safely push ISO, especially on larger sensors, with minimal negative effects given that this tiny and pretty poor sensor was still usable completely maxed out.
I though you tried your Fuji
Next time!
This was done when waiting for our food order to finish, as kiddo was riding my shoulders piloting me like a mech. Can’t do that with the Fuji, need both hands 
Cellphone? Easy enough one-handed and always on me.
I’ve done a bit testing with my iPhone, it does work.
I’ve mixed in ISO 100, 200 and 400 M2P images when mapping in order to keep the aperture and shutter speed where I wanted them when isolated clouds covered the sun, with no apparent deterioration in the orthomosaic.
The M2P sensor is much larger than a phone sensor so I expect it to do better, maybe up to 800 iso.
The GoPro 10 image deteriorates badly over 100 iso, the base iso.
It depends a lot on not just the sensor but the implementation with the chip reading the sensor.
And of course the lens in front of it bending the light correctly.
