Are 40D raw files compressed

[peter_popp1]

Hi all -

 

I was out taking some shots of the moon last night, and noticed for the first time that my raw files are not

all the same size out of my 40D. The moon shots came back at about 8.4 or 8.5 mb, whereas all of my

daytime shots come back around 14 mb, with some slight variation. A nighttime shot of the moon (with

all of that black background) is a prime candidate for compression, but I guess I thought that CR2 files

weren't compressed. What gives?

 

Peter

[KenPapai]

The 40D's CR2 file sizes will vary as you see, self-evidently -- your axiomatic findings are proof enough. Yes.

[joe_fikes]

It seems clear that some compression is going on. That's true with raw files for older camera models as well. Some forms of compression are "lossless" in that the original data can be recovered exactly while other compression schemes are "lossy" (e.g. jpeg). If you can stand something less than perfect reconstruction of the original data, then lossy compression can do a better job of reducing the file size. But if you want perfect reconstruction, you can still get some (modest) file size reduction through lossless compression.

 

Joe

[steve_dunn2]

<p>Yes, just like all other Canon RAW files (at least, all for remotely modern bodies), they are losslessly compressed.</p>

[J.W. Wall]

Depends on how much data is required to reproduce the scene?

[bobatkins]

Yes. The Canon RAW format is a lossless compressed format.

 

The more "information" in an image (i.e. the more detail - or noise), the larger the compressed file will be.

[conraderb]

Peter - raw file size will vary, depending on a few things, most noticeably, IMHO, ISO. I can get around 100 shots at ISO 1600 on my 20D with a 1 gig card, and around 125 shots at ISO 100.

[mendel_leisk]

Undoubtably the reason high ISO files are bigger is due to increased noise: it just doesn't compress as well.

[colinsouthern]

All Canon RAWs are compressed.

 

If you have a quick think about it - if they weren't, and you were shooting with a 40D then (with 14 bit A/D conversion) you'd need 2 Bytes per pixel multiplied by 10,000,000 pixels = 20,000,000 bytes + EXIF data.

 

In other words each shot would be exactly the same size, and approx 20MB - which they aren't :)

 

Cheers,

 

Colin

[vineetmodi]

Can somebody tell me the difference between "RAW" and "s-RAW" in the 40 D? I haven't used it yet, but I just saw that feature in the showroom. I was told that the only difference between the two is the Size and nothing else, but I wasn't fully convinced. Why would canon give the option of two Raw formats in a camera if the quality of both are exactly the same?

[peter_foiles2]

Vineet-

 

You were given inaccurate information. Information is discarded in s-Raw and that is why it is smaller.

[vineetmodi]

Thanks Peter, I also did some google search right now and got more info. So basically its the pixels that get lost in sRaw isn't it? The "quality" is the same, but I can't blow up the picture and take out like 8' by 10' prints out of sRAW. Its a good small size file for small prints and computer screens etc.

[robin_sibson1]

sRAW is still a bit of a mystery (in terms of Why? as well as What?) but at least it is clear that it is much more information rich than a RAW file with half the (linear) resolution, and may well support high-quality prints in small sizes quite successfully.

 

RAW and sRAW files are encoded into a compressed form losslessly, as other posters have pointed out. The important point is that because this encoding is lossless, the only people who need to know anything about it are programmers writing code to pack and unpack it. Whereas with a JPEG, which uses lossy encoding, the photographer needs to understand what is being thrown away so as to be able to take a decision about whether it is acceptable.

[robin_sibson1]

Just to make my comment clearer, what I am saying is that information IS thrown away as between RAW and sRAW, but the subsequent encoding of sRAW is lossless.

[daniel_nordling1]

Judging by the actuall size of the sRAW files, giving just a quarter of the pixels but still half the size of a full size RAW file, I would guess that the sRAW has full color pixels as opposed to the Bayer coded pixels of a full size RAW. This means that the color resolution is actually the same for sRAW and full size RAW, with the difference being the luminance (or black and white) resolution.

 

This would suggest that the sRAW:s would stand uppsizing better than the actuall pixelcount hints, that is we have better per pixel image quality in a sRAW. This is still a teoretical resoning, while the limited support for sRAW in raw-converters has stoped me from doing any real tests.

 

/Daniel