[beholder3]

Here is a little comparison of a 36 MPx FF image versus a 108 MPx smartphone image.

The 36 Mpx image is shown at 300% where the smartphone one is shown at 200% - extreme pixelpeeping.

While the smartphone camera has a lot of limitations (in darker areas the situation is a lot different), I think the tiny lens does resolve probably around the same as a 80-90 MPx FF. Pretty neat for a < 500 EUR (brandnew price) phone camera, both sensor and lens.

At the same pixelpitch a FF sensor would have 1,495 MPx (!).

Attached Images

 

[Fenwoodian]

The 36 Mpx image is shown at 300% where the smartphone one is shown at 200% - extreme pixelpeeping.

I'm not sure such a comparison is valid. Of course any image enlarged to 300% is going to be much worse than one enlarged to 200%. You might want to redo the test with a longer lens on the DSLR body.

Also, what lens did you use on the DSLR? Obviously there will be a huge difference between a cheap/slow kit lens and a $1,500 premium lens. Shooting technique and IBIS use would also have a big impact. I see that the shutter speeds and apertures are hugely different in the two images - another factor that invalidates this comparison. And how about sharpening - I suspect that the smartphone image was sharpened heavily in camera.

Until all of the above are addressed, I'm not going to be convinced of your conclusions.

[K1N8]

But only JPEG not RAW?

[zapp]

I don’t get your point and I don’t think you have a point.
Everything above 200% is an artificial, resampled image. What are we comparing in size, scale, details?

[biz-engineer]

You should have used f/5 instead of f/8 aperture setting for the DSLR. Using the widest possible aperture is important in the process of maxing out image quality. Rule of thumb f# = 2 * FL / 10 ~f/5 @ 24mm FL.

[house]

I think commenters above are missing the point. Regardless of various tweaks and considerations the phone shows incredible detail. Impressive stuff.

[beholder3]

I'm not sure such a comparison is valid. Of course any image enlarged to 300% is going to be much worse than one enlarged to 200%. You might want to redo the test with a longer lens on the DSLR body.

When doing a visual comparison between two images of varying pixelcount you simply have to use different scaling or the comparison is worthless. The subject has to appear same size.


The DSLR was shot at 24 mm and the prime on the smartphone is 25 mm, so they are reasonably "the same".

Also, what lens did you use on the DSLR? Obviously there will be a huge difference between a cheap/slow kit lens and a $1,500 premium lens.

The DFA 24-70/2.8. It is very good quality at 24 mm.

Shooting technique and IBIS use would also have a big impact. I see that the shutter speeds and apertures are hugely different in the two images - another factor that invalidates this comparison.

Both cameras use IBIS.

Differing shutter speeds are a consequence of differing apertures. The smartphone prime aperture I can not adjust it is set to F/1.7. This F1.7 is the equivalent to F6.3 on FF. I do not believe the difference between F8 and F6.3 is going to have a huge impact.

---------- Post added 16th Dec 2019 at 23:05 ----------

I don’t get your point and I don’t think you have a point.
Everything above 200% is an artificial, resampled image. What are we comparing in size, scale, details?

This is an excercise in pixelpeeping. I had to use enlargement to that degree to show differences.

We get to see more details fromt he 108MP.

[GUB]

Is it just my monitor -- but what is the blue/magenta blush thru the denser twiggy area of the phone one? It looks like massive ca.

[tvdtvdtvd]

When doing a visual comparison between two images of varying pixelcount you simply have to use different scaling or the comparison is worthless. The subject has to appear same size.

Wouldn't it make more sense to normalize the lesser pixel count at 100% and adjust the other sensor to match?

[Rondec]

Neither one is great. I do agree that depth of field should be way greater with the phone camera -- you should stop down the full frame camera to compensate. In addition, the shutter speed is way different, if accurate. That much difference would eliminate movement blur on the phone image which is probably present at 1/25 second on the SLR image.

[beholder3]

Is it just my monitor -- but what is the blue/magenta blush thru the denser twiggy area of the phone one? It looks like massive ca.

Either your eyes or your monitor. The degree of CA you see is much less than any typical lens shows. Lens correction on the DFA was on, so software removed any CA there.

---------- Post added 17th Dec 2019 at 09:31 ----------

Wouldn't it make more sense to normalize the lesser pixel count at 100% and adjust the other sensor to match?

No. The differences in detail resolving power for already high res sensors (anything > 16 MPx) can only be seen at very, very large magnifications. With your approach you'd just see two images which both are good.

---------- Post added 17th Dec 2019 at 09:34 ----------

do agree that depth of field should be way greater with the phone camera -- you should stop down the full frame camera to compensate.

It was stopped down to F/8 versus F1.7.

In addition, the shutter speed is way different, if accurate. That much difference would eliminate movement blur on the phone image which is probably present at 1/25 second on the SLR image.

The shutter speed is a result of the aperture differences. And no, there is no movement blur. The FF sensor simply doesn't resolve as much.

[thazooo]

I see what you're saying and probably agree with you, with certain caveats. The more pixels you have, the finer the detail can be captured. First you need a lens that resolves it all. Second you need daylight or a flash attached because the dimmer it gets, the more noise shows up, lack of photons in the smaller photo sites.
I've played around with my Optio A40 vs. *ist D. When you go passed 100-300% and get to pixel level , you can see a difference, more pixels equal more detail. Now printed, 5x7 or 8x10, you can see a slight detail increase at 5x7 but it's gone by 8x10, the A40 at 5x7, equal or there abouts at 8x10. Never did 13x19 cause I knew who the winner would be. It was a fun exercise and probably lead me to pick up a M4/3.
Just an observation and No I can't supply the files because they were dumped years ago