Sony a7R testing, part 1

[I’ve noticed that people are linking to this page a lot, since it is the first page of my a7R work. However, that work involves a lot of posts, so I’m adding a little roadmap.

• If you are more interested in the Sony alpha 7 (without the R), go here.
• If you want an overview of Sony’s lossy raw compression algorithm (used on both the a7 and a7R) take a look at this. If you want to see what the artifacts of that algorithm look like when presented with a test image designed to trip it up, here’s the place. If you want to know if problems in compressed images are obscured by photon noise, click here. If you’d like to see what images would look like if Sony removed the delta modulation part of their compression algorithm, they’re here. If you are concerned about artifacts and want a sense of how visible they are, go here.
• If you’re interested in the dark noise of the sensor, go here, here, and here.
• If you want to see how the sensor’s bit depth changes between continuous and single-shot mode go here,
• If you care about how the performance of the camera varies with ISO setting, look here.
• If you want to see examples of in-camera ISO settings vs post processing pushing, look here and here.
• If you’re concerned about corner luminance and chroma errors with short third-party lenses, look here, and here.
• If you want to look at corner smearing with a short symmetric lens on the a7R and the Leica M240, here’s the place. Also here. Not to mention here.
• If you want to see how a third-party short retrofocus lens does on the a7R, look here, and here. Foreshadowing: it does great.
• If you’re concerned about the tolerance of lens adapters — and you should be — look here, and here,
• If you want to see how longish third-party lenses do on the a7R, go here, here,
• If you’re interested in corner cast removal using Adobe Flat Field, look at this. If you want to see how that interacts with corner smearing, go here.
• If you’d like to see the fantastic performance of the Zony 55mm f/1.8 Sonnar FE on the a7R, look here.
• If you want to see the rumors that the a7R is soft in the corners put to rest, go here.
• If you don’t bother to keep the dust out of your adapters, read this.
• If you want to see a simulation study on the effects of anti-aliasing and fill factor on sharpness, here it is. A similar study on camera motion is here. A study showing how vibration blur affects resolution is here, and the results run counter to the conventional wisdom.
• For a visual study of horizontal and vertical camera vibration with the a7R and a 100mm lens, look at this.
• A set of photographs showing shutter shock resolution degradation with a 135mm lens is here. Here’s another set showing degradation at 1/500 second. Here’s proof that electronic flash isn’t always fact enough to stop the effects of vibration on a7R images. But trailing curtain synch can help matters.
• If you’d like to see two really expensive third party “normal” lenses compared on the a7R, look here. Also here. It’ll make you really happy you bought the Zony 55mm f/1.8 FE.
• If you want to see how what some people have called the world’s best wide-angle zoom works on the a7R, here’s a link. Here’s another.
• A study of the a7R shutter motion using a home-built accelerometer is here. Here’s a technique for ameliorating the effects of shutter shock.
• A three-part series on the physics, engineering, and practical implications of vibration control starts here. There’s a simulation study of the a7R’s shutter shock here.
• There’s a comparison of the shutter shock vibration of the Nikon D800E (in mirror-up mode) and the Sony a7R here.
• Here’s a test of a couple of ways to ameliorate the a7R shutter shock with a 135mm lens.
• How bad shutter shock affects your images depends on how sharp your lens is. Here’s why.
• Here’s a shutter shocker. The 55mm Zony is affected.

And now, on with the original post:]

An a7R arrived yesterday. Typical Sony packaging, which I like; the Leica and Apple packaging just seems excessive to me. It takes the same battery as the NEX-7, which is A Good Thing, but it doesn’t come with a charger, which is Not. Sony learned this cost savings trick with the RX-1. It didn’t become that expensive camera, and it doesn’t this one.

If you liked the RX-1 menus and controls, you’ll be right at home with those on the A7R. The long menus on that camera, however, are even longer on this one. I reiterate my call for a firmware image for serious photographers that strips out sweep panos, smile recognition, and the like.

There has been discussion of vibration induced by the closing, then opening of the shutter on the a7R that is a result of Sony’s not including an electronic first curtain on the camera as they did with the plain a7. This is the same issue as with the Leica M240, tested here, except you can’t turn off live view on the Sony camera, so there’s no workaround like that for the M240.

Since I’m going to be doing some sharpness testing, I thought I’d deal with the shutter vibration issue right away. I connected the a7R to a Nikon 400mm f/2.8 lens that has in the past shown a remarkably high sensitivity to vibration. I clipped the lens into a RRS TVC-34L Versa Series 3 tripod with an Arca Swiss Cube head. I pointed the camera at an oscilloscope with no vertical input, and the time base set to 1/10 second per division. I set the camera shutter speed to half a second, so that it could capture five divisions of trace movement.

As the front of the camera moves down the image of the dot on the sensor moves up, and as the front of the camera moves up, the image of the dot move downs. Having the dot move from left to right during the exposure would allow me to see during what portion of the shutter’s open period the vibrations occurred, and to see how long it took them to damp out.

I focused with the lens wide open. It was a joy compared to the Leica M240 and the Nikon D800E; the refresh rate of the display is rapid and there doesn’t appear to be any pixel subsampling. It was easy to achieve precise focus, something I can’t say about the other two cameras, or even the Nikon D4. Then I stopped the lens down to f/8. I set the ISO to 100. I set the intensity of the oscilloscope phosphor dot to a low level to make it small. As it turns out, I didn’t set it as low as in the previous ‘scope tests, so the blooming was worse than before, but I could see enough to tell pretty much what’s going on.

I made the first set of exposures using the self-timer with the camera oriented horizontally, so that the vertical changes in the line are the result of up-and-down motions of the camera. I brought the images into RawDigger – which already supports the a7R, although Photoshop CC and Lightroom do no at present – and exported them as gray-scale composite TIFFs so that there was no demosaicing. I adjusted the white and black points, and resampled the vertical dimension up 3:1.

Here’s the image:

There is a four or five pixel peak-to-peak vibration with a fundamental frequency of 10 Hz. This is about the same frequency as with the D800E tests shown here and here. That’s what you’d expect, since the lighter weight of the a7R compared to the D800E isn’t enough to make much in change the mass and moment of inertia around the tripod attachment point; the heavy lens dominates. The vibration reaches its maximum excursion about 50 milliseconds after the shutter opens. The amplitude of the vibration is about a fifth to a quarter of the mirror-slap vibration of the D800E. This is an apples-to-apples comparison since the pixels in the two cameras are the same size.

The difference is that you can finesse the D800E’s mirror slap vibration by using the delayed shutter feature, or, if three seconds is not long enough for you, the mirror-up setting. There are no equivalent settings on the a7R.

With the camera oriented vertically, so that side-to-side motions appear as vertical variations of the ‘scope trace, things look quite different:

There is less vertical deviation visible, meaning that the camera doesn’t shake from side to side when the shutter fires as much as it does up and down. However, there are variations in brightness with the same 10 Hz frequency of the vertical vibrations. This is caused by the fact that the camera is moving from side-to-side when it’s twisted into the portrait orientation, and the photographed trace is brighter when the camera is moving in the same direction as the ‘scope beam (from left to right), and dimmer when the camera is moving in the opposite direction as the ‘scope beam.

I’ll be doing some more testing, but my initial conclusion is that, if you’re going on safari in Africa, the a7R will not be your camera of choice if you routinely lock up the mirror with an SLR.