High ISO Compared: Sony A7S vs. A7R vs. Canon EOS 5D III
Sony's A7-series lineup has garnered quite a lot of interest amongst camera enthusiasts, striking an appealing balance between image quality and size. Both the A7 and A7R bodies, as well as the native FE lenses Sony has recently introduced, are remarkably small and lightweight - especially when you consider the quality of images they're capable of producing.
The A7S is Sony's newest entry in its full-frame mirrorless lineup. But where the 'R' in A7R stood for resolution, the 'S' in the 12MP A7S stands for sensitivity. Furthermore, with its full-sensor readout and 4K video capabilities, the A7S is as much a camera for video as it is a camera for stills. And let's face it: there are very few videographers and photographers who wouldn't be thrilled by a camera offering higher sensitivity and impressive, low noise, high ISO performance.
Now, this purported increase in sensitivity comes at a cost: in the face of cameras with ever-increasing megapixels, the A7S comes in at a rather paltry 12MP. Lower resolution and higher sensitivity are certainly not unheard of - the Canon 1DX and Nikon D4S both top out at relatively modest resolutions of 18 MP and 16MP, respectively (though this is partly done in the name of speed). But is the resolution of a bygone era (Canon EOS 5D anyone?) a worthy tradeoff for the still image and video quality enhancements? Sony certainly seems to think so.
We just got our hands on an A7S, and have been busy putting it up against its competitors to see what advantages it offers and, quite frankly, to see if the claims regarding its low light performance are accurate. Particularly, some of us here have been curious as to whether not the increased ISO performance is significant enough to warrant the resolution tradeoff when the A7S is put up against its older sibling - the A7R.
Why make this resolution trade-off at all? Leaving aside video considerations (such as simplifying full sensor readout) for now, lower resolution sensors can increase pixel-level performance (because bigger pixels capture more light), but it's typically total light gathering area across the entire sensor that is a major determinant of ISO performance, all else being equal. So, to see whether the A7S offers anything beyond the pixel-level benefit its lower resolution would lead you to expect, the higher resolution image is normalized to the resolution of the lower resolution camera. Ultimately, for the A7S to make sense to stills as well as video shooters, Sony's engineers need to have exploited some of the other advantages that well-designed larger pixels can potentially bring *1.
We put the A7S up against the higher resolution A7R to see if the A7S offered any significant high ISO advantages over the A7R when the output of the A7R was downsized to that of the A7S. Furthermore, we pitted the A7S against a professional DSLR not too far outside the price range of the A7S. And so, we bring you this real-world comparison between the Sony A7S, Sony A7R, and Canon EOS 5D Mark III.
We shot a night scene that included a range of tones from deep shadows to bright highlights to get a comprehensive idea of noise performance of these cameras at various ISOs.
We waited until midnight to ensure minimal changes in ambient light during the course of our shoot (summer days are long in Seattle). In order to level the playing field for all three the cameras we did a few things:
- Used the same lens (Canon 24-70 f/4L IS) for all cameras. A Metabones Smart Adapter III was used to fit the Canon lens on to the Sony bodies.
- Aperture and shutter speed were matched across all cameras for any particular ISO setting.
- RAW files were converted in ACR 8.5 to give relatively consistent rendering across cameras. This was done by manually selecting white balance per camera, and adjusting Shadow Tint as necessary at higher ISOs in order to avoid magenta-tinted blacks. Sharpening and noise reduction were left to ACR defaults (Sharpening: 25 | Luminance NR: 0 | Color NR: 25).
Do note, however, that (1) nobody in their right mind would boost high ISO JPEGs in this manner (it's preferable to digitally boost Raws over JPEGs), and (2) when we did boost the high ISO Raw files in ACR, significant noise resulted. This itself speaks to the value of the higher ISO modes on the A7S (if you need them, that is), but also indicates that a more careful balancing act of noise reduction, sharpening, and exposure boosting would be more appropriate to obtain the 'simulated' higher ISO values than by simple (digital) exposure boosting.
Without further ado, let's get to the comparisons. Below and on the next page you'll find a variant of our typical studio scene widget. Have a play with it (further instructions after the widget), then view some of our specific thoughts on the comparison on the next page. By default we've set the widget below to compare the A7S vs A7R, but you can compare either camera to the Canon EOS 5D Mark III as well. And, as always, all files are available for download if you would like to tinker yourself.
Using the Widget
The widget above and on the next page is a variant of our typical studio scene widget. At the top is a drop down menu that allows you to select 'Normalized (12 MP)' or 'Native Resolution'. 'Native Resolution' indicates that images retain the maximum resolution the camera is capable of, whereas for the normalized analysis, images were downsized to 12 MP (using the bicubic resampling method).
As usual, we also have the Full, Print, and Web buttons at the top right of the widget. Full will maintain whatever you've selected in the top, center drop-down menu (12 MP for the normalized analysis, and full sensor resolution for the native resolution analysis). Print will downsize all images to 8 MP, while Web downsizes all images to 5 MP.
*1 For example, lowering cumulative sensor read noise, increasing effective sensor efficiency, and other factors that - to prevent this article sprawling - we won't elaborate upon further here.