Lens Sharpness & Sensor Performance

[CiiVii]

Okay- I was under the impression from the wording of the first post that both photos were taken in natural light only. So I was like, "Wait a minute..."

[natek313]

Originally Posted by CiiVii

Okay- I was under the impression from the wording of the first post that both photos were taken in natural light only. So I was like, "Wait a minute..."

No problem.

Reading my original post again, I should have been more clear. The very first photo is in-studio. The second is in daylight.

[zona5101]

Your Churches ad test photo is suprising to see that much difference between f7 and f5...

[OsmosisStudios]

Originally Posted by natek313

Thanks, Os.

I'm starting to think it's a combination of the difference between studio and natural light along with the difference in settings.

I did some testing this afternoon. The lighting was natural (skylight window). The subject distance was probably 15ft - 20ft or so. See attached photos (SOOC; 100% crop) and below for EXIF data (same settings as in the first post):

First photo: ISO 100; f/7.1; 1/20s

Second photo: ISO 200; f/5; 1/100s

There is a rather significant difference in image quality; look at the detail difference in the chicken.

I'll have to keep all this in mind. Plus, I have to remember I'm pixel peeping... Haha.

you're at 1/20s in the first image: that's motion blur. Try it on a tripod with a self-timer. You shouldnt see a difference.

[natek313]

I shot both of those photos on a tripod with mirror lock up on any shutter speed slower than 1/60s. I tripped the shutter with my wired remote shutter release.

[OsmosisStudios]

No idea then: it looks like slight motion blur though. Whether its the camera or the target, there's some movement there.

[natek313]

I understand what you're saying, Os.

I just realized, though, that we're not on the same page. The photo with the 1/20s shutter speed is the sharper of the two images.

[BCampbell]

Nate, I think you're seeing exactly what you'd expect to see, especially if you looked around at lens tests.

In most serious tests, you'll see a sharpness bell curve. Some lenses peak at some aperture a few stops narrower from wide open, and some peak at the diffraction limit (if you want to see something really interesting check out the chart on this page and see where different Canons hit their diffraction limit).

Most of the time, we don't really notice minute differences in sharpness. Sure, we can tell the difference between a 2.8 prime wide open vs. f/8, but sharpness differences between smaller steps are usually hidden by other factors in real-world shooting. You happened to be closely examing some photos that happened to be fairly well controlled for those other factors. And of course the differences are more apparent in the stricter test. That's pretty nice, actually, good job.

Specifically, there are probably a few things happening here. The most obvious is that by stopping down your photos are sharper. There's more to that, though. Most lenses are sharpest in the center of the frame, and fall off as you approach the edges and especially corners. Some lenses lose sharpness dramatically, others less so. You say these are 100% crops so I'm not sure where in the imaging circle you're taking the samples from, but that could have something to do with it. You may not see much of a sharpness difference between 5.6 & 7.1 in the center, but it might be more noticeable at the edges. Since we train ourselves to compose our subjects off-center, how much a lens' sharpness falls off can be pretty important.

The ISO difference is really important too. Remember that by going from 100 to 200 you're doubling the "sensitivity" of the sensor, and thus doubling the noise. This is absolutely going to have an impact, and actually I'm thinking it may be more of a factor than the aperture in your test case. You say these are SOOC; does this mean you're shooting JPG? If so, your camera is applying noise reduction, and that results in a definite hit in sharpness. If you are shooting JPG, trying shooting RAW and comparing again without any noise reduction. Also try both apertures at identical ISOs.

One final factor which probably isn't in play but is important to remember. At wider apertures we all know you have less depth of field. But this also means you have less leeway for correct focus. Usually this isn'ta problem, but if your lens doesn't focus precisely -- say it front focuses or back focuses just a bit -- that's going to have more of an effect at wider apertures, especially when you're shooting something flat, perpendicular to the focus plane (as in your test).

So, there can still be a few confounding factors! But it seems like you're on the right track.

[natek313]

You bring out some great points, BCampbell.

Originally Posted by BCampbell

Specifically, there are probably a few things happening here. The most obvious is that by stopping down your photos are sharper. There's more to that, though. Most lenses are sharpest in the center of the frame, and fall off as you approach the edges and especially corners. Some lenses lose sharpness dramatically, others less so. You say these are 100% crops so I'm not sure where in the imaging circle you're taking the samples from, but that could have something to do with it. You may not see much of a sharpness difference between 5.6 & 7.1 in the center, but it might be more noticeable at the edges. Since we train ourselves to compose our subjects off-center, how much a lens' sharpness falls off can be pretty important.

I totally forgot about sharpness in relation specific areas of the frame. Fortunately, though, when I set up my testing, the shots were taken with the subject pretty centered in the frame. The crops were taken close to center as well - perhaps not dead on center, but fairly close.

Maybe I'll do some testing on the edge(s) of the frame to see what insight that gives me.

Originally Posted by BCampbell

The ISO difference is really important too. Remember that by going from 100 to 200 you're doubling the "sensitivity" of the sensor, and thus doubling the noise. This is absolutely going to have an impact, and actually I'm thinking it may be more of a factor than the aperture in your test case. You say these are SOOC; does this mean you're shooting JPG? If so, your camera is applying noise reduction, and that results in a definite hit in sharpness. If you are shooting JPG, trying shooting RAW and comparing again without any noise reduction. Also try both apertures at identical ISOs.

I shot both photos in RAW. I did shoot both apertures at the same ISO setting, but I don't think I compared them side by side. That's something I'll have to try.

Thanks for your comment. It was a good one.

[BCampbell]

Originally Posted by natek313

Fortunately, though, when I set up my testing, the shots were taken with the subject pretty centered in the frame. The crops were taken close to center as well - perhaps not dead on center, but fairly close.

I figured as much, but it was worth pointing out... especially when we're actually talking earl-world applications.

I shot both photos in RAW. I did shoot both apertures at the same ISO setting, but I don't think I compared them side by side. That's something I'll have to try.

Ah, okay. Make sure you're not applying any kind of denoising by default in your conversion software... the noise itself shouldn't make things softer, but denoising sure does.

This has been a relaly interesting thread... if you do more tests, post an update!