Regarding checking the effect of sRGB vs Adobe RGB vs Pro Photo RGB on a given raw file - you can do this very easily. Open the raw file in Camera Raw and change the color space via the link in the bottom.
Notice how colors can get out of gamut when you go from Pro Photo RGB to sRGB for instance. Just activate the highlight and shadow warnings and play around exposure and color adjustments. Also pay attention to how the histogram change.
Fred Miranda wrote:
In-camera color profile only affects JPEG files. If shooting RAW, it does not matter if you have sRGB or AbobeRGB selected.
Agree. Technically, Sony's raw compression scheme is way beyond my understanding but if it must somehow use a gamma curve to decide what get's more compression vs less compression (in camera through processing), is that not in some sort of color space itself? If it's not a specific color space per se, it must somehow effect the color channels (I'm sure this is somehow related to the last post by theSuede).
Stoffer wrote:
Regarding checking the effect of sRGB vs Adobe RGB vs Pro Photo RGB on a given raw file - you can do this very easily. Open the raw file in Camera Raw and change the color space via the link in the bottom.
Notice how colors can get out of gamut when you go from Pro Photo RGB to sRGB for instance. Just activate the highlight and shadow warnings and play around exposure and color adjustments. Also pay attention to how the histogram change.
I know that very well see it daily when processing in C1 the file changes as you set diffrent profiles( color space) but when you import to ACR it imports to what you have set. This is before it even hits the raw converters. If there tagging the raw SRGB in camera in the raw file than that file is screwed out of the gate.
This goes beyond the logic of what all assume to be correct.
Tariq Gibran wrote:
Agree. Technically, Sony's raw compression scheme is way beyond my understanding but if it must somehow use a gamma curve to decide what get's more compression vs less compression (in camera through processing), is that not in some sort of color space itself? If it's not a specific color space per se, it must somehow effect the color channels (I'm sure this is somehow related to the last post by theSuede).
Your on to something. This is deep in the file before it even hits the SD card.
GMPhotography wrote:
I know that but when you import to ACR it imports to what you have set. This is before it even hits the raw converters. If there tagging the raw SRGB in camera in the raw file than that file is screwed out of the gate.
This goes beyond the logic of what all assume to be correct.
You need to use my tricks while you're still in ACR. Don't hit Open Image.
Anyway, I find it highly unlikely that Sony could force sRGB on the raw conversion in ACR even if they wanted. It doesn't make sense as the raw format is entirely free of color space formats. But if you had that suspicion, I'm saying that you might be able to check that out with my trick in ACR.
If the raw file was really mapped to sRGB, it shouldn't change if you choose Pro Photo RGB.
GMPhotography wrote:
I'm playing devils advocate here. I'm questioning what they are doing or not doing. As I said it's a thought given they are doing this compression algorithm to start with.
Lossy compression of RAW files is not unprecedented, Nikon offer a lossy compression option, I believe RED's RECODE RAW video format is also lossy compressed.
RAW compression has some clear advantages - e.g. reduced filesizes. Sony's lossy algorithm has some clear advantages, it ensures the RAW data is a constant size regardless of image content, it's highly parallelisable, and very fast (there are no floating point operations required, just simple lookups and addition/subtraction).
Attempting to applying colour space transforms makes no sense conveys zero advantage to anyone.
GMPhotography wrote:
Is there a program or something to confirm that.
There are many ways you could confirm it, but I'd say it's not worth the effort.
GMPhotography wrote:
what if Sonys SRGB tag is sticking to the raws on import. I see it this way they are doing something no one does with there raws , why would this area be any diffrent. They obviously are doing weird stuff that makes little sense
It's entirely up to the RAW converter how it attempts to interpret the RAW data, and any metadata attached to the RAW files. I can say with certainty that ACR/Lightroom doesn't pay any attention to any colour space metadata tags and always uses the user specified space.
Thanks Matt you know where I was going with this question. I was wondering if they where doing something really stupid here.
Example Im a Sony engineer lets say . Im told to put in SRBG and Adobe RGB in the camera to tag the jpeg files for in camera processing to come out in that Color space. Ill do one better ill tag the raw too since the end user put in SRGB in the menu well Ill just tag the raw too as i assume that is what they want.
See where Im really going with this. Obviously I completely simplified that to make a better understanding of how possibly they did something and not realized it.
Okay Im done back to work. Fun question for the morning and got us all thinking
p.16 #10 · Uncompressed 14 bit for A7sii and A7rii
Tariq Gibran wrote:
Technically, Sony's raw compression scheme is way beyond my understanding but if it must somehow use a gamma curve to decide what get's more compression vs less compression (in camera through processing), is that not in some sort of color space itself? If it's not a specific color space per se, it must somehow effect the color channels (I'm sure this is somehow related to the last post by theSuede).
Colour spaces are like co-ordinate systems, they allow you to interpret a set of numbers. So I could say my secret volcano base is located at [1232.343, -2343.34] but that tells you nothing unless you know whether I'm using Latitude and Longitude (and which geodesic), or Eastings and Northings, for example.
Sony's lossy compression algorithm converts data values from a linear 14-bit space into a nonlinear 11-bit space. So it's a space, yes, but it's non a colour space, since it's one dimensional. Anyway this is irrelevant as the decompression algorithm reverses the operation, producing linear values again.
Some data is lost in the process, and could in very specific circumstances lead to posterisation, but I think the blue skies issue is the result of downstream colour conversions that have nothing to do with a) the RAW file itself, or b) any metadata attached to the RAW file.
So why does this issue apparently (I've not seen this proven) not affect other cameras? All cameras respond to colours differently due to different frequency responses in the CFA dyes, and so require different downstream colour manipulations to get to a certain end result. I think this is most likely the cause - if there is indeed actually an issue.
p.16 #11 · Uncompressed 14 bit for A7sii and A7rii
GMPhotography wrote:
Thanks Matt you know where I was going with this question. I was wondering if they where doing something really stupid here.
Example Im a Sony engineer lets say . Im told to put in SRBG and Adobe RGB in the camera to tag the jpeg files for in camera processing to come out in that Color space. Ill do one better ill tag the raw too
I don't see why that would be a particularly stupid thing to do...
However, like I said RAW converters are free to do whatever they like. ACR/Lightroom/dcraw definitely ignore any such tags and let the user choose the colour space. I'm going to go out on a limb and say the other major RAW converters do the same.
p.16 #12 · Uncompressed 14 bit for A7sii and A7rii
Matt Grum wrote:
Colour spaces are like co-ordinate systems, they allow you to interpret a set of numbers. So I could say my secret volcano base is located at [1232.343, -2343.34] but that tells you nothing unless you know whether I'm using Latitude and Longitude (and which geodesic), or Eastings and Northings, for example.
Sony's lossy compression algorithm converts data values from a linear 14-bit space into a nonlinear 11-bit space. So it's a space, yes, but it's non a colour space, since it's one dimensional. Anyway this is irrelevant as the decompression algorithm reverses the operation, producing linear values again.
Some data is lost in the process, and could in very specific circumstances lead to posterisation, but I think the blue skies issue is the result of downstream colour conversions that have nothing to do with a) the RAW file itself, or b) any metadata attached to the RAW file.
So why does this issue apparently (I've not seen this proven) not affect other cameras? All cameras respond to colours differently due to different frequency responses in the CFA dyes, and so require different downstream colour manipulations to get to a certain end result. I think this is most likely the cause - if there is indeed actually an issue....Show more →
Another, probably technically inept question... If Sony is using an in-camera process/ processor to create a gamma curve, that is then used to decide what gets compressed more vs less based on that curve which is then subsequently turned into a second gamma curve during the actual raw conversion (based on raw edits as well as initial changes by say ACR/ Lightroom when opening), could that not lead to more posterization effects? Said another way, is there an extra one (or possibly two) gamma curve transformation/ manipulation occurring that does not occur with non-lossy raw formats?
p.16 #13 · Uncompressed 14 bit for A7sii and A7rii
Tariq Gibran wrote:
Another, probably technically inept question... If Sony is using an in-camera process/ processor to create a gamma curve, that is then used to decide what gets compressed more vs less based on that curve which is then subsequently turned into a second gamma curve during the actual raw conversion (based on raw edits as well as initial changes by say ACR/ Lightroom when opening), could that not lead to more posterization effects? Said another way, is there an extra one (or possibly two) gamma curve transformation/ manipulation occurring that does not occur with non-lossy raw formats?
I'm not quite sure what you're asking, the curve doesn't decide what gets compressed - the curve does the compression.
So in the first stage of the compression 14-bit values from the ADC are converted into 11-bit values using a curve. This curve can map two different 14-bit values to the same 11-bit value, hence some information is lost. The 11-bit values are then stored relative to the min/max values in each 16 pixel block (potentially losing more information).
The opposite curve is used as the last stage decompression. So as far as Lightroom is concerned the data is linear and always has been (but some values are missing when two or more 14-bit values were mapped onto the same 11-bit value).
p.16 #14 · Uncompressed 14 bit for A7sii and A7rii
Matt Grum wrote:
I'm not quite sure what you're asking, the curve doesn't decide what gets compressed - the curve does the compression.
I think what he means is that the shape of the curve determines which part of the original linear range sees the most lossy compression (i.e. loss of information bits). The low values of the 14 bit range lose about one bit while the high values of the 14 bit range lose about five bits. In that sense the curve is "deciding" which parts of the range get compressed more than others.
Of course the shape of the curve is chosen so that most of the "lost" information was actually just photon shot noise, but the point is that the curve shape is "deciding" or perhaps better stated "defining" what ranges of the original data are more or less compressed.
p.16 #15 · Uncompressed 14 bit for A7sii and A7rii
kwalsh wrote:
I think what he means is that the shape of the curve determines which part of the original linear range sees the most lossy compression (i.e. loss of information bits). The low values of the 14 bit range lose about one bit while the high values of the 14 bit range lose about five bits. In that sense the curve is "deciding" which parts of the range get compressed more than others.
Of course the shape of the curve is chosen so that most of the "lost" information was actually just photon shot noise, but the point is that the curve shape is "deciding" or perhaps better stated "defining" what ranges of the original data are more or less compressed....Show more →
p.16 #16 · Uncompressed 14 bit for A7sii and A7rii
GMPhotography wrote:
Yes totally understand and that and how it totally should be BUT what if Sony is tagging the raw. Given there compression stuff I'm questioning this.
If you are taking two shots, one for sRGB in-camera setting, the other for Adobe RGB, varying only that and nothing else between the shots, raw dumps from both shots are the same.
p.16 #17 · Uncompressed 14 bit for A7sii and A7rii
I don't mean to hijack this thread, but may I ask a simpleton's question: So far I have done my editing on my laptop for a lack of better options. Hence, I left the camera setting at sRGB. Since moving, I got myself a nice 24" display that pretty much covers the AdobeRGB range. If I understand what Matt wrote above correctly, then it doesn't matter if I set the camera to AdobeRGB or sRGB, because Photoshop ignores the tags and let's the user set the color profile in post? Or is there any benefit in setting the camera to AdobeRGB after all?
