The question then has to be asked – does the ‘Translucent’ mirror reduce the image quality? It does reduce the light reaching the sensor, but the ISO settings of the A55 appear to be adjusted to allow for this. If you shoot 1/125th at f/16 at ISO 100 with the A55, it is similar to the same settings on the A580 (a conventional SLR with the same sensor), not half a stop darker. That means the sensor is being given a touch of extra gain, but comparison of the two cameras’ files does not show any apparent penalty.
After many tests and many images, I can only say that fitting an uncoated UV filter to a lens on the A580 would be more likely to degrade the image or cause flare marks, than using the A55 instead of the A580. The angle of the mirror does not encourage internal reflection, and when a flare patch appears from extremely strong sources in one particular region of the frame, it seems to be a result of a reflection from the AF module.
A studio test of the flare pattern. The three bands appear to be produced by a combination of the mirror and AF module’s reflective lenses, nothing to do with the mirror glass as such, and result from three lines of LEDs in the light source. The small chromatic flare patches are caused by the lens, not the camera.
The actual result of this flare pattern, caused by internal reflections. It takes extreme conditions to show it, and under the same conditions, most cameras and lenses produce some kind of flare – just not quite like this.
This flare only occurs when a very bright source is positioned slightly to the right and higher than the centre of the image. Placing the same source to the left and below, right and below, or left and above does not produce the flare. As it happens, the critical position is about where you would place the sun for a sunset shot following rule-of-thirds composition with the sun towards the right side of the shot. Many users could get this flare, but in practice a typical sunset lacks the contrast to produce it. The light source has to be much brighter relative to the exposure.
The mirror can produce a secondary or ghosted image, offset by a few pixels and lower than the actual image. This is only visible with sharply rendered point sources on a black field, like stars, or sharply focused edges of very bright areas like a window. If the image is even slightly defocused, the secondary image is not detectable; if the contrast is anything other than extreme, it is invisible. Because of the angle of the mirror (which is not 45°, but slightly flatter-on relative to the sensor plane) wider angle lenses will show the ghost towards the bottom of the shot and centre (horizontal composition) but not towards the top, because the mirror is more or less at 90° to the image forming ray bundle in this part of the image.
Here is a shot taken with the 18-55mm SAM kit lens at 45mm focal length and f/6.3, only a stop down from wide open. There’s a very bright pinpoint catchlight in the Bengal cat’s eye, and those whiskers range from perfectly sharp to slightly defocused. Will there be some image-destroying secondary catchlight? Has the contrast been wrecked by that mirror? Will this cheap lens, high resolution, and this unacceptable new design spoil the result?
You can judge for yourself. Note the chromatic effects (tiny coloured dots on the cat’s nose, aliasing on on whisker). Note also the level of sharpening used in ACR 6.3, and the level of luminance NR. See the detail in the colours of the cat’s eye. Now step back – this is a 100% screen shot not from the native size of the file, but from a 25 megapixel supersized export from ACR. This is making the A55 give me an image bigger than the Alpha 900, using a kit lens nearly wide open. This is one very sharp camera, to the point of colour moiré (I’ve emphasized this by not using any Chroma NR, and using double the normal level of sharpening).
The practical effect of these factors is less than the effect of forgetting to fit a lens hood, or buying a cheap filter instead of a high quality multicoated one. Because of the EVF system, and the depth of field preview, you can test for flare patches. This is not always possible with optical finders, as they can create apparent flare which is not present in the picture. It’s caused within the prism or mirror-prism and the eyepiece optics. The A55/33 have a relatively clean optical path, viewing the EVF display which will show any flare. If flare occurs within the eyepiece of the EVF, it won’t look like image flare. So you can shield your lens, adjust your composition, as required to avoid any unwanted effects.
In many hundreds of finished pictures, I have only ‘lost’ shots when I have deliberately tried to find extreme subjects to produce flare. It has never happened in any normal shot. The same goes for the secondary ghosted image; the only time I’ve seen it is when trying to get it. No doubt both will occasionally spoil a picture for someone – one picture in a thousand, perhaps. But the SLT design will enable, or rescue, another fifty pictures in that thousand which would have worse with a traditional SLR. I tend to judge my gear by success-rate, by how many frames are sent to the trash when editing and how many are kept. On this basis, the A55 is probably the best Alpha I’ve owned.
What I found about the image quality of the A55 is simple enough. It has an auto ISO setting from 100 to 1600. With earlier generations of Alpha, the ability to limit this range was vital. You might pick just auto 200 to 400, or 200 to 800, to ensure 1600 was never used. The A55’s ISO 1600 raw quality is so good that with just the normal amount of noise reduction in Adobe Camera Raw (25 Luminance, 50 Colour, with auto detail levels) creates a grain-free, highly detailed result.
The A55 does not depart from ISO 100 in a hurry when set to Auto ISO. Shooting outdoors on a day of mixed light with very mixed subjects (from snow to dark road surfaces) the camera stuck rigidly to ISO 100 until I took an interior, which sent it straight to 1600.
After a few experiences like this, I have started to set the A55 to Auto ISO, something I do not do with any other camera apart from the NEX-5. When the A580 arrived, I realised this would perform in a similar way, and again have used Auto ISO without worries. Even when using Canon cameras with a wide auto ISO range, I’ve never encountered such clean ISO 1600 daylight results. People associate high ISO with night shooting, but actually it’s so good to be able to use something twice as fast as ‘press’ colour negative film used to be, with the same grain as ISO 100 film used to show in a finished print.
This is no exaggeration. If you are still using a camera from just a few years ago, such as the Nikon D200, Canon 40D, Sony Alpha 350 or even the excellent Alpha 700 then you have no idea just how good the latest 16 megapixel CMOS is. Raw processor improvements have made all the older cameras perform better, but these new sensors are perfectly matched to the latest Bayer conversions in software like Adobe Camera Raw 6.3 and Lightroom 3.3.
Hand-held, camera on ground, screen composition, 1/60th at f/8, ISO 6400, SAM 18-55mm.
What you get from ISO 6400 like this, without any special attempts to reduce the noise. Great crisp looking grain in a 100% clip. And where there is sharp detail, it’s very sharp.
As for the higher range still, the results remain proportionally superior. Even at ISO 6400 you can make a large print which shows no sign of excessive noise, or detail-mashing noise reduction. At 12,800 you certainly know that it’s a boosted setting. Not so long ago, 3200 would have been considered acceptable with the same level of noise and NR.
Using the A580 and A55 together for a short period, with various lenses, I felt that both cameras with the 16.2 megapixel sensor are sufficiently demanding to make the 18-55mm SAM kit lens only just adequate. But apart from a slight feeling that the A580 is set up with a slightly stronger anti aliasing filter, leading to a softer look to very fine detail, I could find no reason to prefer one camera over the other. There’s a lot of talk about the SLT design giving lower contrast. Given the huge range of contrast adjustments and picture looks I apply to images, any tiny difference in the optical contrast hardly matters. Just choosing one lens instead of another makes more difference than using the A55 instead of the A580, or any other camera.
Indeed, if you want low contrast, try the 16mm f/2.8 Sony SEL lens on the NEX-5. That’s a lens with a mere 5 elements and a light path with includes nothing but a very small amount of air. The microcontrast and overall contrast alike are lower than you’re ever likely to get from the A55 no matter what lens you fit. Do I find it a problem? No, I love it. It handles difficult light superbly and gives me an extra half stop in dynamic range. Just like old Leica lenses, it lifts the shadows for you. Contrast can always be boosted in post-processing. It’s very difficult to reduce. I remember the first time I used a Contax RTS with a 35mm f/2.8 Zeiss; I had to drop one paper grade to get the same look to my prints, and Kodachrome 64 was just too contrasty to have any shadow detail, it demanded the gentler response of Kodachrome 25.
Well, the A55 is my new Kodachrome 25 – dead sharp, but gentle with tone and colour.