Aperture sharpness results

[gnirtS]

I ran some tests to see where the lens sweet spot and drop off range is for the camera to get an idea of what settings produce acceptable images or not.

Surprisingly, i found the lens at optimum sharpness from f/2.8 up to about f/4 with very little difference between them.
f/5.6 is a little less sharp when pixel peeping but probably still usable.
f/8 is on the borderline and f/11 is terrible.

So from my tests it seems i can use the camera between f/2.8 and f/4 if possible, happily let it drift up to f/5.6. But diffraction starts to kill things at f/8 and f/11 is definitely unusable.

Test conditions - ISO 200 (hence some noise) to guarantee a fast shutter speed to eliminate motion blur error. ALL these images are far cleaner at ISO100 noise wise but shutter speed is too slow at narrower apertures for a spare test.

First images are near the centre


f/2.8


f/4


f/5.6


f/8


f/11


I also tested the edges, corner sharpness does not seem to vary with aperture.

 

[gnirtS]

OK now the same test for ISO. f/5 fixed aperture, shutter speed variable to cope with ISO.

*ALL* noise reduction disabled along with capture sharpening so these images will look a lot worse than normal where you apply those by default on input.

For me 100 is lovely and clean - orders of magnitude better than the M1. ISO200 with noise reduction is fine. Even 400 at a push can be cleaned up. 800 is yuck, 1600 in hell no!


ISO100


ISO200



ISO400


ISO800


ISO1600

 

[miemo]

I did find quite visible difference in CORNER sharpness between f2.8 and f4 and even still at f5.6, as I wrote in the other thread. I was doing a variation of the classic "brick wall test", so it's really easy to spot there. Will not necessarily be as clear in real-life situations with landscapes etc, like in your test shot.

Exactly similar observations on center sharpness from me as well, btw!

 

[gnirtS]

The lens designer did mention the optics were optimised for "distance" as well due to the fact its in the air so i skipped the wall test (that and its hammering down with rain).
Im not seeing much of a difference in sharpness as i think diffraction is taking over a lot of the corner changes.
That said, the corner and edge sharpness is far far better than i was expecting anyway.

Its a good little camera - im more than happy with it.

 

[miemo]

The lens designer did mention the optics were optimised for "distance" as well due to the fact its in the air so i skipped the wall test (that and its hammering down with rain).

True, distance is probably also a factor here. For my aperture test I was shooting topdown maybe round 30-40 meters high, so probably not exactly infinity-focus yet. In some actual landscape-style shots I haven't noticed the f2.8 corner softness that strongly. And in any case, corners are perfectly usable even in the close-focus f2.8 shots!

 

[gnirtS]

Infinity even at f/2.8 works out at 3.3m when you do the maths so 30-40m should be more than adequate for a distance.

 

[kilomikebravo]

Gnirt: Considering you were surprised that the wider apertures produced sharper images, it sounds like we all might need one or two more ND filters. Whaddaya think?

KB

 

[gnirtS]

I'd guess maybe 2 NDs.

An 8 and a 16. Or maybe just a 16. Certainly no more than 2.

It depends what you'd class as acceptable. f/5.6 and f/8 wasn't horrible and in the drastically reduced resolution of 4k video vs still may well be a non issue.

 

[kilomikebravo]

An excellent point, G. I am anxious to see the effects when I get the PP combo. It’s funny in a away...we’ve been through this drill many times in the past...like with point and shoots, phones, and of course, drones. Then there are secrutiy cameras, vehicle cameras, you name it and all we do is salivate for bigger and higher rez displays, more CPU horsepower and obviously, better cameras and lenses. The human is a strange bird, no matter the gender.

KB

 

[miemo]

Infinity even at f/2.8 works out at 3.3m when you do the maths so 30-40m should be more than adequate for a distance.

Hmm, well then there's also variation between different units that might be a factor here. Seeing that in much higher-quality lenses too, so wouldn't surprise to see it with the M2P camera as well. Well anyhoo, this is kind of hairsplitting in any case since we're talking about marginal differences.

 

[gnirtS]

I dont go with that, you might get tiny focal length variations but the focus distance is a factor of focal length, sensor size and distance.
It certainly cant be far away from that figure as the sensor is a constant and lens is a constant (within reason).

I dont mean "infinity focus" here which changes on every lens on every body but purely once you HAVE an accurate focus on a flight for infinity there should be no need to change focus again unless you're within about 3.3m (wide open).

Even a DSLR APS-C sensor with a f/2.8 28mm lens has infinity at 15m!

A Flexible Depth of Field Calculator to do the maths.

 

[miemo]

I dont go with that, you might get tiny focal length variations but the focus distance is a factor of focal length, sensor size and distance.

Didn't mean that there would be variance in what the infinity focus is between units, but in corner sharpness. That really is a thing with lenses in general.

 

[gnirtS]

Corner sharpness will vary for sure. Even properly made lenses have variation and im not going to accuse this mass produced camera as being properly made to those standards.

That said, compared to the mavic 1 its incredible. Mavic 1 had uneven corner sharpness and random edge sharpness eating into the scene as they never got the lens and sensor properly parallel on large numbers of batches.

 

[macfawlty]

Exactly the info I was looking for. Thanks for the test pics.

 

[CanadaDrone]

Exactly the info I was looking for. Thanks for the test pics.

It's actually physics - since we know it's a 1" sensor with 20MP, we know that after F4 diffraction will start to degrade the image based on the pixel size. You can calculate this for any sensor without even having it

 

[AcesHigh]

This is why you still should use a ND Filter for Mavic 2 Pro even with manual aperture. They came to the same conclusion that the sweet spots is something between f/2.8 and f/5.6

 

[gnirtS]

This is why you still should use a ND Filter for Mavic 2 Pro even with manual aperture. They came to the same conclusion that the sweet spots is something between f/2.8 and f/5.6

For video yes. For stills, you definitely don't need one and shouldn't use one.

 

[CanadaDrone]

For video yes. For stills, you definitely don't need one and shouldn't use one.

I wouldn't go as far as saying that. There are several good uses for ND filters for stills on the M2P, it just depends on your personal usage. I've already seen some great waterfall shots from the M2P just as one example, and they would not have been possible without a ND.