sRGB or Adobe 1998?

[gordo]

This question pervades photography forums - without seeming to reach a conclusion: Should photographers edit their images in sRGB or aRGB (Adobe 1998).
Here's one argument: "Because most people will not be previewing images on a profiled monitor in an ICC aware application, sRGB is the closest space to most monitor's native space. Without complete color management of all users that view the image, sRGB will be your best bet for getting others to preview your image somewhat like it does in an ICC aware application."

Here's another: "Keep in mind, you can always go smaller, but once you do, you can't go back up. For ex, if you have a large bucket of water and you pour it into a smaller bucket, you will loose some of your water. If you pour the water back into the big bucket, you don't get all your water back; you have a big bucket with a little water in it. Almost every source I have read recommends Adobe RGB for in camera and post processing - certainly for print publication."

What do recommend to customers/photographers - sRGB or aRGB, and why?

thx, gordo (not affiliated with Kodak)

my photos: Gordon Pritchard's Photography
my blog: Quality In Print

 

[meddington]

I recommend AdobeRGB over sRGB simply because I work in the print industry. sRGB clips a large portion of most CMYK spaces. AdobeRGB encompasses most of the gamut of the majority of print devices currently available, and the trend in monitors seems to be wider gamut as well. Another one to throw into the mix is ProPhotoRGB to completely take advantage of a printers gamut, as well as archive images for future use (as gamut continue to broaden). Perhaps a bit overkill currently, and necessitates the need for 16 bit, but an option for those planning for long term archiving.

 

[Lukas Engqvist]

I would recommend ECI RGB, not only because I live in europe but because of the whitepoint being that of prepress industry (5000 K) and also a better fit at clear yellows. We print alot of yellow machines and so ECI gives us better reproduction. Adobe RGB is very close to ECI RGB, and if it was just nature photos I'd be fine with AdobeRGB…Â*I will probably be creating device links for all three (sRGB, AdobeRGB and ECI RGB to be able to benifit from the RIPs 16 bit colour conversions)

 

[mattbeals]

Not a color management expert here. There have been similar posts on the ColorSync forum about this. It would appear that this question will never be concretely answered since there are so many varied opinions from the experts. It seems that for archival that Pro photo would be best. For screen viewing, generally speaking, sRGB. And for print if you aren't printing to ISO or GRACoL then Adobe 1998 would be best. If you are printing ISO or GRACoL then ECI would be the best fit.

So shoot as big as you can, shoot in RAW and away you go! Since most "prosumer" cameras shoot in sRGB it's pretty safe to assume that at one point or another, usually on capture, that sRGB is where the image is. If you've got the ability to shoot in RAW then do that. The argument can be made that you just convert everything not already, for example, ECI RGB and then convert to CMYK.

Personally I edit in the native color space of the image and then convert to CMYK and tweak that.

 

[meddington]

I would recommend ECI RGB, not only because I live in europe but because of the whitepoint being that of prepress industry (5000 K)

Hi Lukas

What would you say would be the advantages of the ECIs 5000K white point.AdobeRGB used to have conversion issues when using abs Col rendering, but that has been rectified in one of the CS releases.


>And for print if you aren't printing to ISO or GRACoL then Adobe 1998 would be best. If you are printing ISO or GRACoL then ECI would be the best fit>

I recommend adobergb over eci often because its available on most peoples workstation already, and the slight clipping that occurs with the gracol and ISO profiles very rarely manifests into a problem. But ill agree that ecirgb is better suited for higher gamut print output.

 

[Lukas Engqvist]

I'd be very cautious about editing CMYK

I'd be very cautious about editing CMYK

The argument can be made that you just convert everything not already, for example, ECI RGB and then convert to CMYK.

Personally I edit in the native color space of the image and then convert to CMYK and tweak that.

I would be very cautious about tweaking in CMYK... mainly since this risks getting TIC/TAC too high. If you tweak an image after converting to CMYK you have not made good use of the CMYK profile. (old school scanner operators will here say that this is easier and better, but that is only true in as much as the chroma profs were better than ICC based digital proofs). I would argue that the only thing you may do with an image after converting it to CMYK is to add sharpness, and this is more necessary for newsprint than coated papers as the lpi that is used today will render the fine details as long as basic sharpening is done in the image.

Perhaps that is why I prefer ECI RGB. Since a tagged image will convert correctly in a colour managed workflow. (I do still use gamut warining in RGB mode to make me aware of overly saturated areas, if I would convert to CMYK I would never regain the lost information anyway)
White point is important to get the lighter colours correct it is as important as white balance and is the basis for comparing light near neutrals.
If the Adobe RGB was ok we wouldn't get the green sunsets… with device links this problem can be fixed, but since it is more difficult to get a client to adapt to a device link workflow than install the ECI RGB profile I would say the ECI RGB will give better results.

 

[meddington]

Hi lukas

I still don't follow how the white point of an RGB editing space will have an impact on print work. After all, one generally would not use abs Col rendering when converting from an RGB space.

As far as the green sunsets, can you describe the workflow or provide a sample showing this issue? Again, I'm not sure how often this would come about in print as adobergb's gamut is sufficient for ISO and gracol in the yellow area. I could see how an image untagged' or tagged with eci RGB would have this effect upon assigning adobergb to it, but this isn't a proper workflow either.

I do agree that editing in RGB has many advantages over cmyk, but I wouldn't say that editing cmyk automatically leads to inferior results. I would say that this is likely still a predominant manner of image editing and there's certainly high-end printing achieved this way.

 

[mattbeals]

Tweaking in CMYK is different than editing in CMYK. I'm not doing anything where I'm going to be causing ink density to go too high. Sometimes at the last minute you have to tweak, and that's usually in CMYK. With the proof preview and gamut warning, etc in PhotoShop you can make all those changes before hand with certainty (or at least a high degree of certainty). Ideally we'd all be working in an RGB workflow. Looking at the greater industry, not just print providers but content creators and printers, we aren't there yet. Are we closer? Yes, but not far enough along for it to work for the masses. Editing in CMYK virtually ingrained in our culture because we weren't able to express or communicate in RGB for so long. Even now it's still difficult for many, if not most, people.

 

[Lukas Engqvist]

The white point will have an effect even in a colour managed flow because of the difference from the white. In exactly the same way as paper colour or lighting will affect near neutrals.
For the neutral spine to be maintained there must be a distortion in the relative difference of the near neutral colours. This is only apparent in cases when fading to white but with designs including many feathered objects this cannot be totally ignored.

Having said that with a good CMYK ICC using a separation that bases neutrals on black even in lighter tints gives a more tollerant workflow (in this respect). This is also more of a problem if you have a coverage of tints less than 5% and was therefore not recognised in a film workflow. One way to avoid the issue is to clip lower % screen. It will give a harder edge to fades and feathers though.

Also closest to pure yellow that can be acheived with relative conversion in Adobe RGB is 2C 97Y ECI RGB is not perfect with it's 1C and 98Y. In ECI RGB 100% cyan is in Gamut, Adobe RGB allows only 100C 5M.
But ECI RGB can get 97Y 1M in gamut where as Adobe RGB's closest match is 97Y 5M if you do not want Cyan impurities. (which is why converting with device link would be the best option, but has not been truly practical until PDF print engine due to transparency issues with RGB sent to RIP)

 

[meddington]

>The white point will have an effect even in a colour managed flow because of the difference from the white. In exactly the same way as paper colour or lighting will affect near neutrals.For the neutral spine to be maintained there must be a distortion in the relative difference of the near neutral colours>

Fair enough lukas, Ill accept your explanation, though I haven't been able to reproduce the white point issue that you mention.. Also, in many cases, eci or adobe RGB is an intermediate space between device spaces and should maintain the original intent in this regard. Moreover, the white point of the original photo capture as shot is most often variable (unless under fixed studio lighting), so a given white point changes throughout the workflow.

> Also closest to pure yellow that can be acheived with relative conversion in Adobe RGB is 2C 97Y ECI RGB is not perfect with it's 1C and 98Y. In ECI RGB 100% cyan is in Gamut, Adobe RGB allows only 100C 5M. But ECI RGB can get 97Y 1M in gamut where as Adobe RGB's closest match is 97Y 5M if you do not want Cyan impurities. (which is why converting with device link would be the best option, but has not been truly practical until PDF print engine due to transparency issues with RGB sent to RIP)>

Those are good points, and are well taken, however the values you mention are specific to only one profile/rendering intent: ISO coated/rel col. Gamut mapping characteristics of the cmyk profile will also contribute to this as well. For example the ISO coated profile available from Basiccolor (based on the same Fogra39 data) will allow for 100Y 0C from eci RGB.

Again, good points made for ECIrgb over adobergb, but these purity scenarios aren't prevalent in the majority of images. And as shown by your examples, some impurities will be involved in nearly every workflow. One could try to minimize these issues with device links or profile edits, or different gamut mapping options. One could also tweak in cmyk.

Mike