Sony's lossy RAW compression?

[Avotius]

Reading reviews of Sony's full frame mirrorless cameras I have come upon people saying Sony is using non optional lossy RAW compression, which in my mind really seems weird for raw images.

I was wondering if anyone here using any of the A7 series cameras has ever encountered a problem in real life professional settings where the posterization resulting from the compression has found its way into your images or any other issues you have found.

Thanks.

 

[craygc]

5n & A7II - never seen an issue

 

[YYV_146]

The long answer is that you won't see compression unless you're shooting subjects with extreme contrast differences. If you clip the sky to white but still have -3 or -4 in the shadows, for example, the recovered shadows will show artifacts because the algorithm can't figure out what to do with the exposure difference. This is also true for something like single-exposure star trails, since the stars are at least 6-7 stops brighter than the background, and the difference is amplified by a long exposure.

But in real life these conditions are rare and far in between. You can for example use Sony's time lapse app or a smart remote and do image stacking, as proper star trails should be made, and the issues goes away. As for shooting into the sun, my experience is that artifacts don't show up if there is some level of information in the highlights. This means that shooting at base iso or under-exposing 2-3 stops is usually a good practice, since Sony sensors are ISO-less anyways.

The short answer: It's not a problem. Moire on the A7r is a much bigger issue, but that I've accepted as a trade-off for pixel-level resolution.

 

[goo0h]

Moire on the A7r is a much bigger issue, but that I've accepted as a trade-off for pixel-level resolution.

do you think moire will be even more of an issue with the a7R II with it's even more densely packed sensor?

 

[YYV_146]

do you think moire will be even more of an issue with the a7R II with it's even more densely packed sensor?

Theoretically the denser sensor should have less issues since it takes a finer texture to reach moire levels. If photographing a piece of fabric where the fine patterns are close to the pixel pitch of, say, the A7s causes moire, then the A7rii image should be fine since the pixel wells are much smaller.

I mostly notice it on a A7 with the AA filter removed when shooting brick building from a distance. That seems to be the one pattern that never fails to create moire on any AA-less camera...

 

[Spanik]

Maybe not now but as display technology will advance you'll be wondering why you didn't care for it now.

 

[pvdhaar]

I've had no posterization issues with the way ARW files are compressed for normal photography. However, when doing astrophotography there is a problem. Actually, there are two.

First off, with astro, you're trying to see in the dark.. The detail that you want to pull out of the image is close to the background level, and therefore when there's a bright pixel in the same block, that detail gets mapped onto the background level. In a nearby block that doesn't contain such a bright pixel, the pixels close to background level are emphasized. This is visible as blockiness in the image.

This can be overcome to some degree by stacking multiple images with shorter exposure. Because of earth's rotation, bright stars move around between exposures, and fall into different blocks. The stacking (there's SW that maps the images like DeepSkyStacker or GRIP), averages out the blocking and improves the end result.

However, there's a secondary problem, and that is that the posterization isn't only in the luminance, but also in the color domain. Cosine roll-off of the lens causes circular color bands (magenta-green-magenta-green) with blocky outlines when cranking up the contrast at the lowest light levels.. this happens even with stacking, because the light roll-off of the lens is constant..

My solution is to do astro with a Nikon (in my case a D90).. and all the rest with the Sony.

 

[sevo]

The only subject where I believe I may have seen any effect from it was when reproducing rasterized newsprint. But that is a difficult subject in any case, and the culprit may just as well have been the AA filter or thick cover glass on the A7.

 

[willie_901]

SONY does filter raw files. This in itself is not necessarily bad. What is disappointing is the filtering can not be disabled.

The data shows SONY's raw filtering has a small, but measurable effect on images. In my view the filtering is not an issue for practically all photographers. While the filtering has caused a bit of controversy, it's not a show stopper.

Here is a link to a long series of data-driven articles that evaluate the SONY A7RII. The topics covered include lens filter effects, IQ with adapted wide-angle lenses, sensor performance in a variety of situations (long exposures) and ISO invariance (shadow region performance). Some of these measurements are directly related to the topics discussed in this thread.

 

[uhoh7]

Working a ton with both, I can say, for me, Sony Raws are very lame compared to my M9 RAWs (uncompressed).

The Leica raws respond very well to pushing the shadows, but the same moves on the Sony files are far less effective and create much more noise. In fact nearly everything you do to a sony raw in terms of exposure adjustment makes noise in Lightroom.

Now, when I edit the Sony files, often the first thing I do is change the camera calibration, which does not make noise, then I adjust gingerly.

Sony seems to have no concept of a proper RAW. So many complaints they even give the issue lip-service, but the problem has been driving us crazy since the Nex cameras were introduced in 2010, and they do nothing.

 

[goo0h]

speaking of Sony raw, I found this to be an interesting discussion:

http://www.dearsusan.net/2015/08/23/393-in-the-a7rii-has-sony-secretely-tweaked-its-recipe-for-raw/