Canon's M and S RAW files

[Slobodan Blagojevic]

Ok, M and S files offer smaller size, with all the advantages it might have (less storage space, faster transfer, rendering, etc.). However, is there any benefit in terms of image quality, e.g. noise?

Bart?

[dwswager]

Ok, M and S files offer smaller size, with all the advantages it might have (less storage space, faster transfer, rendering, etc.). However, is there any benefit in terms of image quality, e.g. noise?

Bart?

Considering that the S-Raw is 1/4 the pixel count and M-Raw about 2/3 the pixel count of a regular Full resolution raw file,  one would expect to see some downsampling advantage.  This sounds like a fascinating topic.  I suspect, good image specific post processing would achieve better results, but...waiting for this thread to develop.

[Bart_van_der_Wolf]

Ok, M and S files offer smaller size, with all the advantages it might have (less storage space, faster transfer, rendering, etc.). However, is there any benefit in terms of image quality, e.g. noise?

Bart?

Hi Slobodan,

Although I've not shot or studied my Canon files in M or S compressed form, I do not think that they have any other benefit than smaller size in camera.

From what I remember reading about it, the files may be (partly) demosaiced and basically linear gamma RGB or another color coordinate system at a decimated pixel size, hence smaller despite being demosaiced pixels. However, the decimation is probably not aimed at improving the signal to noise ratio, although it could be a side effect of it. But as always, even if the S/N ratio is improved, the loss of resolution is also a given, and as such offers no benefit (if it even gets as good) compared to downsampling.

To be more sure though, I'd have to look into the matter and shoot some examples myself. The Raw converter one uses may also play a role, so the initial test would probably be with Canon's DPP application.

Cheers,
Bart

[Craig Lamson]

Hi Slobodan,

Although I've not shot or studied my Canon files in M or S compressed form, I do not think that they have any other benefit than smaller size in camera.

From what I remember reading about it, the files may be (partly) demosaiced and basically linear gamma RGB or another color coordinate system at a decimated pixel size, hence smaller despite being demosaiced pixels. However, the decimation is probably not aimed at improving the signal to noise ratio, although it could be a side effect of it. But as always, even if the S/N ratio is improved, the loss of resolution is also a given, and as such offers no benefit (if it even gets as good) compared to downsampling.

To be more sure though, I'd have to look into the matter and shoot some examples myself. The Raw converter one uses may also play a role, so the initial test would probably be with Canon's DPP application.

Cheers,
Bart

I just shot a job using my 5Ds using some full Rez, some mraw and some sraw.  Many images I make for clients only get delivered at 8x12.   So far I'm not seeing any real issues with the smaller raws, but I have  yet to shoot the same image as a full Rez downsampled to both smaller raws to compare quality.

I must say using the smaller raws makes more sense....for me...than switching cameras or storing the larger raw files when I don't need them.

[AlterEgo]

M and S files

http://blog.lexa.ru/2013/07/14/pro_sraw.html

[Bart_van_der_Wolf]

http://blog.lexa.ru/2013/07/14/pro_sraw.html

Ah, yes, thanks. Now it's coming back from my recollection. They are a 'lossy' YCrCb space encoded and decimated formats. So color accuracy suffers, but it of course depends on the subject matter if that is noticeable (especially without a better, downsampled reference).

It remains unclear how the subsampling of color is done, but most likely much less sophisticated than a good downsampling would achieve, perhaps they just simply average neighboring pixels, instead of taking a weighted average like with downsampling. Also the decimation method used for Luminosity (Y) is unclear.

Again, mRaw and sRaw can be perfectly usable for some applications, and may be the only way to store huge numbers of files on a limited (in camera) storage medium. But arguably better image quality can be had from proper downsampling of the full size Raw conversion. MRaw and sRaw are (only) about file size reduction, not highest quality.

Personally I would not use it as a means to reduce noise (if that even works adequately), because it's much more effective to do noise reduction at the full size and do proper down-sampling, and the noise vs detail trade-off benefits from having more pixels to base it on (even if the final output size is relatively small).

Cheers,
Bart

[Ghibby]

From my use and brief testing M and S raw are not that useful on the 5Ds, especially M raw which appears significantly less sharp than full res and even more so S raw which is really sharp. Both are also fractionally cropped. Undoubtedly they are demosaiced in my opinion. Lloyd Chambers makes a very good analysis on his DAP subscription site. My findings fully support his more thorough analysis.

If it's for in camera storage I would use sRaw to get out of a tricky storage shortage issue at a push but would not touch mRaw, storage space saving is minimal for the quality loss. Storage being so cheap now I can't see this ever really being a problem with say 128 or 256 gb on board.

I have had good results with lossy DNG format. This is obviously only useful for computer storage not cam storage. It's smaller / similar to sRaw in file size but full res. It's demosaiced but with a floating point conversion method I think, preserving a great deal of file information. Quality is great, 98% of original full Raw. No change to dynamic range or colour either. My worry is its future compatibility, ties you to Adobe as well. I do not use it for critical or important work but for more transitory images and non critical work it more than halves the storage required. Worthy of investigation in my opinion if you have not tried it. I was highly sceptical of it when I first tested it but it's proved genuinely useful a few times and image quality is a non issue.

Ben

[AlterEgo]

I have had good results with lossy DNG format. This is obviously only useful for computer storage not cam storage. It's smaller / similar to sRaw in file size but full res. It's demosaiced but with a floating point conversion method I think, preserving a great deal of file information. Quality is great, 98% of original full Raw. No change to dynamic range or colour either.

http://chromasoft.blogspot.com/2012/01/lightrooms-new-lossy-dng-compression.html

[Ghibby]

Hi AlterEgo,

That is a very interesting article. Certainly more than reinforced my thoughts on using lossy DNG for anything other than basic and unimportant work. As I said in my post I would never use it for critical work. Same applies to M and S raw.

However the question becomes how does this compare to M and S raw for canon shooters. Canon is clearly doing something with these small raw formats, is it better or worse than Adobe's Lossy DNG?

If I have time I think I will do some testing of my own soon.

Ben

[nemophoto]

I've shot modest amount of MRaw files with my 5Ds and have been very satisfied with the results. The quality of the files seems good, though I haven't done a significant amount of high ISO shooting with that setting. In a way, it makes more sense to shoot full RAW and then downsample the image afterward if you are concerned about grain/noise. I few shots I did that were underexposed about a stop or a litle more, showed about the same amount of noise as a full res RAW, which would lead me to believe there's no real qualitative improvement shooting medium RAW for that reason. You reap more benefits with full RAW, do your conversion, then downsample if you need to.

From my use and brief testing M and S raw are not that useful on the 5Ds, especially M raw which appears significantly less sharp than full res and even more so S raw which is really sharp. Both are also fractionally cropped. Undoubtedly they are demosaiced in my opinion. Lloyd Chambers makes a very good analysis on his DAP subscription site. My findings fully support his more thorough analysis.

Ben

I think the softness you see, Ben, is more due to the lower res aspect of the shot (50MP vs 28MP vs 12MP) versus a lack of sharpness BECAUSE you used MRAW and SRAW.

[Ghibby]

I think the softness you see, Ben, is more due to the lower res aspect of the shot (50MP vs 28MP vs 12MP) versus a lack of sharpness BECAUSE you used MRAW and SRAW.

I don' think this is the case, on a per pixel basis at least.  I think what is happening is that the M and S raw images are being taken from a crop of the full frame of 8640 x 5760 (the full res is 8688 x 5792).The very slight crop I see in these formats is the loss of 48 pixels horizontally and 32 vertically, in practice this is very minor.

For the M RAW this results in a 1 : 1.3333 reduction ratio (from the 8640 x 5760 cropped frame). By contrast the S RAW is a precise 1:2 reduction (4320 x 2880). The reason it is sharper on a per pixel basis is that this reduction is clearly favourable as each pixel is formed from 4 of the original pixels. 

There is no question in my mind that the level of per pixel sharpness in S Raw is better than M Raw and even a file exported from at the same dimensions from the full res raw requires considerable sharpening to equal the sharpness of S-RAW. 

It would also appear that Canon could do something akin to phase one's sensor plus technology with S RAW. If it was implemented in a different way a massive reduction in noise and increase in DR for high ISO could be achieved from this setting in principal. A bit of innovation would be nice!

Regards
Ben