On Sharpness, ISO and Shutter Speed

Started Nov 29, 2013 | Discussions thread
Anders W Forum Pro • Posts: 21,466
Re: On Sharpness, ISO and Shutter Speed

Jack Hogan wrote:

Some of this information has appeared in recent posts, but I thought it deserved its own thread.

DPR's improved technique in capturing images for its new Studio Scene includes focus bracketing and controlled lighting, resulting in one of the better databases of systematically collected raw files available for downlonad online. Nevertheless there have been reports of inconsistencies and unexpectedly blurry images which most people (including myself) have almost subconsciously ascribed to either faulty equipment or operator error. Given the vast number of variables, pieces of equipment and information collected there are bound to be some of those. On the other hand, if we take the information for good, we may be able to gain some interesting insights on our gadgets, their performance and testing at large. As some of you know I've been playing with MTF Mapper recently, so in this post I am going to concentrate on MTF50 as an imperfect but convenient indicator of Spatial Resolution quality in the captured information (aka sharpness).

I used the slanted edges in DPR's new Studio Scene's raw files to derive vertical and horizontal MTF50 readings at base ISO in order to get the cleanest reading possible. AFAIK the captures are taken on a tripod, mirror up (where there is one), delayed release, typically at a lens' sharpest aperture, at whatever shutter speed is appropriate for the desired Exposure. The few times that I checked ISOs just above base I saw a tendency for the MTF50 readings to creep up, which I assumed was increased noise messing with MTF Mapper's algorithms. Upon further inspection, it turns out I was wrong.

Here for instance is how the MTF50 readings vary with ISO on a Nikon DF whose sensor has a 'traditional' antialiasing filter:

It doesn't vary much. Here are two more with sensors by a different manufacturer (the D800 applies digital scaling after ISO1600)

Shutter Speeds Correspond to ISO 200-6400

And since we are at it here is a more recent camera with a sensor by the same manufacturer as the last two, Sony's A7

The vertical and horizontal readings in previous graphs were within what I think would be the expected accuracy of measurements from different setups (say +/-3% from the average), but obviously something is very different in this case compared to the previous ones - most likely the AA as explained by The_Suede here . It is also clear that MTF Mapper deals with the increasing noise with aplomb, so taking the data for good it is worth investigating some of the imperfections in it.

For instance the dip in performance of the EM5 at base ISO. There have been suggestions that some cameras suffer from obvious shutter shock. Could this be it? Let's take a look at its younger brother, which reportedly shows the issue more clearly

Shutter Speeds correspond to ISO 100-6400

Whoa. Looking at the relative images in the comparometer confirms these results: below 1/320th the EM1's images are definitely blurrier. Shutter Shock, changed lighting, changed positions, changed hardware, misfocus, operator error? And what do we make of that vertical/horizontal crossover?

Steen Bay wondered how the AAless Sony A7r would look next to the AAful A7 above, because the A7 has an electronic shutter while the A7r has a mechanical one.

Interesting. This confirms reports that the A7's images look sharper than the A7r's at some ISOs in the Studio Scene - I (probably incorrectly) thought the A7r's focus had not been peaked properly. And there's that dip again.

I also wondered how the D610 (mechanical shutter) would do against the A7, since its sensor is the same size, it is made by the same manufacturer and it has the same resolution.

The first set of ISOs come from the 'Dim' Studio Scene Setting. The seond from the 'Daylight' setting

Still interesting. The large difference in horizontal and vertical readings is similar to what it was in the A7 - but here we see the inconsistent 'dip' near base ISO and few subsequent stops, which is not there in the A7 with its electronic shutter. The slower shutter speeds are from 'dim light' Studio Scene raw files, to see whether a change in lighting/WB would make a difference. It apparently doesn't.

So I have two initial questions, assuming that the data is valid:

1) What are the practical implications for a photographer of the large spatial resolution differences in the V+H directions seen in the newer sensors measured here ? Will it make a difference whether we look at the final image in portrait vs landscape orientation?

Why would it make a difference if we look at it this way or that way? Blur is blur and to my knowledge, the human visual system isn't differently sensitive to blur in one dimension than in the other. Besides, we wouldn't normally want to look at something shot in landscape orientation in portrait mode or vice versa, for pretty obvious reasons.

2) Are the dips we see in MTF50 around 1/10th of a second to about 1/300th possibly due to shutter shock - or something else? What implications does this have for people choosing a high resolution camera today?

Hard to say without knowing more about what's going on in DPR's studio. But the shutter speed range affected and the fact that the blur is largely vertical makes shutter shock the likely culprit. The problem is known to occur in that range, with a peak at about 1/100 s or slightly higher. The point where the problem disappears at the upper end of the range may be related to the max sync speed of the camera (disappearing later the higher that speed is). At the lower end, it disappears more gradually, without much of a clear cut point.

That said, it surprises me to see the shutter-shock problem (if that's what it is) appear so clearly in the DPR studio scene samples. Presumably, they use a heavy studio tripod and the focal length is not all that long (about 100 mm EFL). Furthermore, one would at least have expected DPR to take the precaution of using anti-shock delay with Oly bodies and the electronic shutter option with Pany bodies. But I certainly wouldn't have expected any clear shutter-shock problems even if they didn't, provided that my assumption about good tripod support is correct.

Everybody's thoughts are welcome.


 Anders W's gear list:Anders W's gear list
Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Olympus E-M1 Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS Panasonic Lumix G Vario 7-14mm F4 ASPH +28 more
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow