April 10, 2017

The Sony A6000 and Raw files

I shoot exclusively in RAW and am always surprised when I find a truly serious photographer who doesn't (but they are out there). There are some surprising issues with Sony RAW files.

Lossy RAW compression

There has been quite a furor over this, and rightly so. Sony uses a lossy compression algorithm to generate their raw files. Lots of people misunderstand this, and it isn't surprising unless they are software specialists. Some people argue that they want uncompressed raw files, which simply shows that they are idiots. What they (and you and I) want is a lossless compression in the raw file. This is what Canon and Nikon use, and Sony should too.

Apparently Sony continues with this choice of policy in the a6300 and a6500 cameras. And a policy choice is what it is, any of these cameras could easily perform lossless compression, it is nothing more than a software choice, and in fact performing lossless compression is simpler than performing lossy compression.

Does it matter? Probably not. But we are purists, and we want everything in our raw files. A quick look at file sizes (and comparing to RAW files generated by my Canon 5Dii with virtually identical numbers of pixels) shows that the Sony files are uniformly about 25M in size, and the Canon files range from 25-30M. So Sony is neither buying us much in terms of compression, nor are they likely throwing very much away. Just for the record uncompressed RAW files are about 50M for a 24Mpx camera.

There is all kinds of information about this online if you care to read about it.

A person has to speculate that Sony is deliberately crippling their APS-C camera line. Especially since after all the furor they continue with this policy in the A6300 and A6500 cameras. Even more curious is Sony's response to provide uncompressed RAW, not losslessly compressed raw in their full frame camera line. This is truly bizarre, and almost seems like a "sour grapes" decision by their engineers. What else could it be? The Sony engineers aren't stupid. I can imagine them grousing about the ungrateful customers and we'll show them and all of that sort of thing.

If you read the second of the above articles above, you may be convinced that indeed it just doesn't make a difference, the Sony Engineers are smarter than we thought, and all of this is a big fuss about nothing. Though why they don't just say "the customer is always right" and give us lossless compression like we ask for (and don't need) I will never know. Maybe they are just convinced that there is no point to it and are too lazy to code up anything other than uncompressed files. Well who knows?

12 or 14 bits

Oddly, it is surprisingly tricky to get an answer to this question. One article made the claim that the A6000 will drop back to using 12 bit conversion when it is forced to do fast image capture, and the article above partially confirms this. The article above claims that conversion is done first as 14 bits, then mapped to 11 bits using a compression curve. Then an additionally lossy step performs compression as they describe.

This all may work out "pretty well", but it is definitely not what we expect from a top tier camera.

Feedback? Questions? Drop me a line!

Tom's Digital Photography Info / tom@mmto.org