Martin that's actually a fairly esoteric subject, but the short answer is probably not to an appreciable degree, especially if you are outputting to a device with a fairly limited color space of its own, such as an inkjet printer. At some level the answer is yes, but that level is probably more of a metaphysical one than a practical one for most people. ;-)
NR is a JPG processing thing, unless you're talking about long-exposure noise reduction. That's why it is always better to shoot RAW and do your own processing. This is true of noise reduction, sharpening, saturation, etc... In camera JPG processing means you're sacrificing final creative control over your images to the camera's computer.
And since every image is different, it makes sense that one set of general trweak parameters (saturation, sharpening etc), is not nearly as effective as processing each image individually, with your artistic eye.
JohnJ80 wrote:
Very cool shots. Can you turn off the NR in the Leica? Leica is out for me because there is no underwater housing.
This, of course, begs the question - can you turn off the NR in the G7 or G9? Is NR applied to the RAW file?
Would be great to do the NR on my mac using Noise Ninja - has to be tons better than on the little microprocessor/DSP chip in the G7/9.
Patrick Cox wrote:
Ariel, Where did you find the F31 for $200?
I heard that BeachCamera used to have it for $239.99 and then a $30 MIR on top of that. I checked again just now and it's back up to $299 and a $30 MIR.
Ariel Bravy wrote:
I heard that BeachCamera used to have it for $239.99 and then a $30 MIR on top of that. I checked again just now and it's back up to $299 and a $30 MIR.
I got it for $210 used here on the B&S.
Buydig, a subsidiary of Beach Camera, has it for $269.99 minus a $30. mail in rebate. I ordered one today. I have bought a lot of stuff from them and they are very good - fast, free shipping, and for me, living in NY, no tax.
Nill Toulme wrote:
John, the 8-bit --> 16-bit comment was just by way of analogy. Just as you can't create data by converting an 8-bit file to a 16-bit file, you can't create missing colors by converting an sRGB file to Adobe RGB. You're just putting the same data in a larger container.
Patrick, if you're shooting RAW it doesn't matter. RAW is RAW. But if you're shooting jpg and intend to work in Adobe RBG, then you should set your camera to that space.
Nill
~~
www.toulme.net
Nill,
I do think that there is no difference or loss in converting from color spaces.
Adobe color space is a wider space than sRGB. The both use 8 or 16 bits. Going from an 8 bit Adobe to an 8 bit sRGB doesn't mean less colors it just means that the available space is still broken up into the same number of steps.
Think of it like this - the Adobe space is a stick that is 512 inches long. The sRGB space is a stick that is 256 inches long. Both sticks are marked of in 256 increments. The sRGB markings are one inch apart, the adobe markings are 2 inches apart. It isn't that information is lost it is just a different mapping. Since it is one to one you should be able to go back and forth with impunity (and no data loss).
The difference comes into how the out of gamut colors are mapped or compressed near the edges and what decisions are made on how to do that.
I'm not sure, but that is how I always understood it. Maybe someone else who knows more about this could correct me if I'm wrong.
Nill Toulme wrote:
Martin that's actually a fairly esoteric subject, but the short answer is probably not to an appreciable degree, especially if you are outputting to a device with a fairly limited color space of its own, such as an inkjet printer. At some level the answer is yes, but that level is probably more of a metaphysical one than a practical one for most people. ;-)
Nill
~~
www.toulme.net
be careful as many high end printers have wide gamuts exceeding AdobeRGB, like the Epsons with K3 inks, new canon iP5000 etc. Even the colour space of the camera exceeds AdobeRGB.
Mapping from a lrge colour space to a smaller one can lead to issues which is why we have different rendering intents to handle these cases. AdobeRGB is stronger in the blues and green and if you map to sRGB you have to decide what's the best way to map out of gamut colours. This choice will affect the look of the print. This is an enormous subject itself and worth a visit to the Printing forum at dpreview or find some excellent articles by Mike Chaney (QImage author) at:
JohnJ80 wrote:
but i think you can go back and forth with no data loss, correct?
Well perhaps if you start in the smaller space, i.e. sRGB, you can. But if you start in the larger space, Adobe, and convert to sRGB, you can't recreate what you lost by converting back to Adobe.
Similarly (and again strictly by way of analogy), nothing is "lost" by converting an 8-bit file to 16-bit, but nothing is gained other than a lot of empty bits. But if you start with a true 16-bit (or anything greater than 8-bit) file and convert it to 8-bit, then yes, information is necessarily lost, and you can't recover it by reconverting that 8-bit file back to 16-bit.
Another analogy would be jpg and tiff. Nothing is "lost" by converting the lossy jpg to tiff, but you do lose something when you go the other way. And if I understand correctly what PP just said, then this might be a more apt analogy, in that repeated iterations of back-and-forth could indeed have a cumulative deleterious effect. That point is beyond my meager knowledge though.
Well I feel good that I didn't buy my G7, I won it in a photo contest. The only thing I really don't like about it was that it had no RAW, so I gave it to my wife and she loves it.
I'll have to keep my eye on this one, I really do need a small pocket camera with RAW.
Nill Toulme wrote:
Well perhaps if you start in the smaller space, i.e. sRGB, you can. But if you start in the larger space, Adobe, and convert to sRGB, you can't recreate what you lost by converting back to Adobe.
Why? forget 8 or 16 bits and let's only consider 8 bits for the moment.
If I take one pixel of one of the colors from a RAW file that has the value of, say, 225. In sRGB that will map into a particular color. In Adobe that will map into another color. Both completely support 8 bits. I can move back and forth all day long and I won't loose anything because that pixel is still at 225 in both cases. If I deleted the pixel or changed its value, that would be a different thing entirely - but that isn't the case.
Similarly (and again strictly by way of analogy), nothing is "lost" by converting an 8-bit file to 16-bit, but nothing is gained other than a lot of empty bits. But if you start with a true 16-bit (or anything greater than 8-bit) file and convert it to 8-bit, then yes, information is necessarily lost, and you can't recover it by reconverting that 8-bit file back to 16-bit.
I think you are confusing converting from 8 bit to 16 bit. That is a different story. You can't map 64,536 values into 256 without losing data. In the changing color spaces, are presuming the same bit depth.
Another analogy would be jpg and tiff. Nothing is "lost" by converting the lossy jpg to tiff, but you do lose something when you go the other way. And if I understand correctly what PP just said, then this might be a more apt analogy, in that repeated iterations of back-and-forth could indeed have a cumulative deleterious effect. That point is beyond my meager knowledge though.
Correct, but that is a whole other issue and is most definitely not an analogy. JPG is a lossy compression scheme. Judgments are made as to what actual data can be dropped without severely impacting image quality - but data is lost. In Tiff it isn't These have nothing to do with color space and are not similar. There is no altering the data in changing from sRGB to Adobe or vice versa. The file is still the same.
Conversion from color space is a one to one mapping. The color may change or shift slightly but no data is lost and one can switch between them with impunity if I understand this correctly.
You may be right and I may be wrong. Maybe Pixel Perfect will come back and educate both of us.
In any event, the one thing that is certain (and relevant in the present discussion of the presence or absence of an Adobe RGB setting on the G9) is that it doesn't matter if you're shooting RAW, so long as you make your initial conversion into the appropriate color space.
JohnJ80 wrote:
If I take one pixel of one of the colors from a RAW file that has the value of, say, 225. In sRGB that will map into a particular color. In Adobe that will map into another color. Both completely support 8 bits. I can move back and forth all day long and I won't loose anything because that pixel is still at 225 in both cases.
You seem to ignore that there are much more colors e.g. in AdobeRGB than in sRGB. I.e. these additional colors have no equivalent representation in sRGB and therefore the conversion will be lossy. The loss may be minimal if the photo has a small gamut which fits entirely in the sRGB color space.
May I express my frustration with Canon and other manufacturers for never quite producing what (in my personal opinion) would be an ideal compact camera. It seems that the combination of small size, image-stabilized fast lens with zoom range at least to 28mm wide, manual controls, RAW capability and reasonable pixel density remains elusive. The G9 is clearly the closest thing on the market now. I would buy any one of the Panasonic DMC-FX55 in a heartbeat if it offered RAW capability, but it doesn't, and that's a deal-breaker.
Canon is really smart, because if the G9 had a slightly wider lens and slightly fewer pixels I wouldn't be thinking about which very expensive lenses I need for my 30D. That being said, I'm thinking about getting the G9 I'm frankly tired of lugging all that gear while travelling. I think what Canon needs to realize is that consumers at this level would not shy away from buying both products--one for travelling and the other for traditional use--even if the G9 feature set more fully overlapped what you could get with a dSLR/kit lens combo in terms of zoom range, speed, resolution, etc...
P.S. What do I need with 210mm zoom in a pocketcam? Needing telephoto capability is a GREAT reason to upgrade to a dSLR. Conversely, having a 24- or 28mm focal length in a small camera is indispensible.
knweiss wrote:
You seem to ignore that there are much more colors e.g. in AdobeRGB than in sRGB. I.e. these additional colors have no equivalent representation in sRGB and therefore the conversion will be lossy. The loss may be minimal if the photo has a small gamut which fits entirely in the sRGB color space.
Actually the number of colours in each space is the same. Think about it we use 8 or 16 bit data to represent an RGB value and thus we have have either 255^3 or 65534^3 possible values to fill the space. Adobe RGB is defined to have a wider gamut than is sRGB. That is to say, it uses the numbering scheme available to it to represent a wider range of colors. Not more colors of course since both are limited by the same physical characteristics dictated by three channels of 8 or 16 bits.
Colour space is thus quantized not continuous. If it were then yes aRGB would contain many more colours.
What aRGB does have then is colors that are further apart from one another than are those in sRGB. The colors must be more widely spaced, since the quantity available is the same as in sRGB but they have to cover a larger gamut.
And since they stretch to fill a larger gamut, specific sets of red, green and blue values represent different colors in aRGB than they do in sRGB.
But since I can set up a 1-to-1 correspondence between RGB values in the 2 spaces via some linear transformation then the mapping from ARGB to sRGB and back should be lossless. Every unique RGB value in aRGB has a unqiue counterpart in sRGB and thus the mapping is invertible. The sRGB values are closer together and the colours won't look the same, but nothings been lost.
To cover a larger gamut, specific values in aRGB end up being used for more highly saturated colors. Take red for example. aRGB can represent more saturated red colors yet both color spaces are limited to 255 (8 bit say) for the brightest red possible, 255 must get used for a more saturated red in aRGB than in sRGB. The highest numbers have to represent colors at the edges of the gamut and that edge is further out in aRGB.
Now this is why rendering intent is crucial; perceptual rendering will preserve all colours and not cause clipping; all aRGB colours will be mapped to unique values in the sRGB space, but the result can look washed out as everything is compressed to fit in the sRGB space. Relative rendering can cause clipping, but anything in gamut is not changed. Out of gamut colours are mapped to the nearest in gamut colour; many out of gamut colours can end up at the same point. This transformation though is lossy.
Pixel Perfect wrote:
But since I can set up a 1-to-1 correspondence between RGB values in the 2 spaces via some linear transformation then the mapping from ARGB to sRGB and back should be lossless. Every unique RGB value in aRGB has a unqiue counterpart in sRGB and thus the mapping is invertible. The sRGB values are closer together and the colours won't look the same, but nothings been lost.
To cover a larger gamut, specific values in aRGB end up being used for more highly saturated colors. Take red for example. aRGB can represent more saturated red colors yet both color spaces are limited to 255 (8 bit say) for the brightest red possible, 255 must get used for a more saturated red in aRGB than in sRGB. The highest numbers have to represent colors at the edges of the gamut and that edge is further out in aRGB.
Now this is why rendering intent is crucial; perceptual rendering will preserve all colours and not cause clipping; all aRGB colours will be mapped to unique values in the sRGB space, but the result can look washed out as everything is compressed to fit in the sRGB space. Relative rendering can cause clipping, but anything in gamut is not changed. Out of gamut colours are mapped to the nearest in gamut colour; many out of gamut colours can end up at the same point. This transformation though is lossy.
I don't know why this is in this thread, but I think you've already made the point that for any practical implementation moving from sRGB to aRGB and visa versa is a lossy process. You say you "can" create a 1 to 1 transformation that is invertible but that is not in fact how any photographically useful rendering intent does it. Such a rendering works great for printing saturated pie charts for presentations but is a disaster with photos. Instead a more lossy process that will not shift unsaturated colors is what should be used and is used in practice. See any book by Fraser or Margulis.
Very helpful, thanks. What I take from this (in part) is that I can convert a RAW to an Adobe RGB tif, then convert that file into an sRGB tif file without losing anything so long as rendering intent is preserved so that the mapping doesn't go astray. Yes?
But if I convert the RAW file initially into the sRGB space, and if it contained some colors that would have been within the Adobe RGB gamut but are outside of sRGB's, then those colors are lost, are they not? And converting that sRGB file to Adobe RGB will not restore them? Or do I still have that wrong?
