Adobe RGB or sRGB???

[Bill, 70's military B&W]

I'm sure this topic has been beat to death but I have to ask "which should I use?"
I just bought an Epson 3880 and I use Canon equipment.
I know this is LF Forum and I'm asking about 35mm, but I'd really like to hear others opinions. Many use both.

I have tried both, and just ended up with problems... mainly because the world seems to revolve around sRGB. I send e-mails and the sRGB shots seem more realistic on computers, mine and theirs, also MPIX seems to want sRGB, other printing shops too. I know I have my own printer now and I can do either, so I'm asking for experiences and results from others.
Thanks,
Bill

[Larry Gebhardt]

Ideally you would shoot RAW and convert the image into the widest color space that your output can support. For your new printer you may want to consider using Prophoto RGB. When you shoot RAW it doesn't matter what color space the camera is set to. You will still want to put images on the web in sRGB, but it's easy to convert the image to that before publishing.

If you are shooting JPG, use Adobe RGB on your camera. I would only use sRGB on the camera if I knew the only use fore the images would be web display (like for pictures for an ebay auction).

I use Adobe Lightroom for most tasks around my DSLR images. One of the main reasons is it handles color spaces very well (converting when it makes sense). I highly recommend it.

[David A. Goldfarb]

As a general rule, I use Adobe RGB for print output, sRGB for the web. On my camera, I use Adobe RGB, and use the "Save for Web" option in CS5 usually to resize and convert to sRGB.

[paulr]

Ideally you would shoot RAW and convert the image into the widest color space that your output can support.

You should convert to the widest RGB color space available regardless of your output device. This is currently Prophoto RGB, as Larry suggests.

The reason is that any colors lost in your conversion are lost forever. The clipped colors will often occur outside the gammut of your current monitor, so you won't even have control over the rendering. Best to hang onto as much information as possible.

It's not a problem that a space like ProPhoto is bigger than your printerr's gamut. You can always use your printer's profile to soft-proof, and then use it as your printing destination space. Or if you're converting for the web or email, soft proof and convert the low res version to sRgb. There are no disadvantages to working this way. Some have pointed out that with a big color space, the editing increments are larger. But if you're working in 16 bit mode, this won't be an issue. The increments will be minute even in Prophoto.

[Andy Eads]

Bill, sRGB was created to accommodate "everymans" home computer monitor. If memory serves me, it was proposed by HP and Microsoft many years back for the then new internet. So it is really a display profile not a print profile. Adobe RGB was created to be a capture profile bridging the gap for graphic arts and printing professionals and encompassing a gamut that cameras and scanners of the time could achieve. Photopro RGB was a Kodak invention that aimed to be a high end capture space that permitted color adjustments with the fewest shifts elsewhere in the gamut, especially if using more than 8 bits per color. In RAW, each camera has its own gamut but the manufacturer will not likely let you at that profile. Larry's advice makes good sense but, as always, your personal use will drive your decisions. The book "Real World Color Management" is the most comprehensive treatment I know of. You may want to check it out.
Andy

[Gary Tarbert]

For what it's worth digital camera Raw adobe RGB,scanning Prophoto RGB , Photoshop workflow prophoto RGB , Anything converted for web or email converted to sRGB .Never for print output Cheers Gary

[pherold]

Steve Upton had an article a while back that spoke to both sides of this question. While most of us here are "Input Centric" there's a place in the world where some can be "Output Centric":
http://www.colorwiki.com/wiki/Color_Management_Myths_26-28 (http://www.colorwiki.com/wiki/Color_Management_Myths_26-28#Myth_27:_Why_would_anyone_ever_want_to_choose_a_working_space_that_is_larger_than_you_can_print.3F)

[bob carnie]

Really nice description Pat.



Steve Upton had an article a while back that spoke to both sides of this question. While most of us here are "Input Centric" there's a place in the world where some can be "Output Centric":
http://www.colorwiki.com/wiki/Color_Management_Myths_26-28 (http://www.colorwiki.com/wiki/Color_Management_Myths_26-28#Myth_27:_Why_would_anyone_ever_want_to_choose_a_working_space_that_is_larger_than_you_can_print.3F)

[Brian Ellis]

What Gary said. Use Pro Photo where you can, RGB where you can't. sRGB only for the web.

[Greg Miller]

Conventional wisdom says use the widest gamut color space (ProPhoto RGB). But for the purist that is not always the best option. A 16 bit color space always has the exact same amount of discrete colors available. So sRGB, Adobe RGB, and ProPhoto RGB all have the exact same total number of discrete colors. Since ProPhoto RGB has a wider gamut than sRGB, Prophoto RGB's colors must be spaced more widely. That is great for an image that contains colors not contained within sRGB, but is actually is a detriment for low gamut images (think foggy scene). So the absolute best color space depends on the image and the colors the image contains. You can argue that the difference is not visible to humans in a 16 bit work flow, but for a bunch of people who obsess about gear and subtle tonal gradations, it is at least worth thinking about.

[onnect17]

Wherever the device you're using to manipulate the image, most likely the monitor, is capable of. If you display can handle 97 or 98% of AdobeRGB, then you should be OK with that space. Otherwise you are safer using sRGB.

[sanking]

Wherever the device you're using to manipulate the image, most likely the monitor, is capable of. If you display can handle 97 or 98% of AdobeRGB, then you should be OK with that space. Otherwise you are safer using sRGB.

How does one determine what percentage of AdobeRGB a monitor can display?

Sandy

[Mark Stahlke]

Sandy,
I use Apple's ColorSync Utility. The ghosted part is my monitor profile, the color part is sRGB.

87343

[onnect17]

It's mostly based on area within a tricolor vector based graph. At least NEC uses that system. Check --> http://www.necdisplay.com/documents/ColorBrochures/PA_Series_SpectraView_SpecBrochure1.pdf

Here's a shot of the screen. The yellow tringle represents the AdobeRGB. Actually the display goes beyond the AdobeRGB IN the "reds"

87344

[Greg Miller]

Wherever the device you're using to manipulate the image, most likely the monitor, is capable of. If you display can handle 97 or 98% of AdobeRGB, then you should be OK with that space. Otherwise you are safer using sRGB.

The compromise being clipping wide color gamut images today to accommodate a low quality monitor or today's monitor technology. Then in a few years monitor technology improves (or you replace the low quality monitor) to a point where it could have displayed those out of gamut colors but you no longer have them available. Many people will have a hard time accepting that compromise. Especially considering what technology will allow in 5, 10, 20, 30 ... years.

[paulr]

The compromise being clipping wide color gamut images today to accommodate a low quality monitor or today's monitor technology. Then in a few years monitor technology improves (or you replace the low quality monitor) to a point where it could have displayed those out of gamut colors but you no longer have them available.

Exactly.

And there's no downside to using a larger color space, assuming you're working with 16 bit images. Use the wide space, use a color managed workflow and calibrated monitor, and then soft proof to your final intent, whether it's to sRGB for the web or to the profile for the printer you're using.

Easy peasy.

[Preston]

Exactly.

And there's no downside to using a larger color space, assuming you're working with 16 bit images. Use the wide space, use a color managed workflow and calibrated monitor, and then soft proof to your final intent, whether it's to sRGB for the web or to the profile for the printer you're using.

Easy peasy.

+1 That's exactly what I do. I use Adobe RGB and let Photo Shop manage color for printing using the appropriate paper profile, and sRGB for the Web. btw: I have an NEC wide-gamut monitor.

--P

[Bill, 70's military B&W]

To EVERYONE... Thank you!!! I need to explore Prophoto RGB, truth is I have never even heard of it.
So it's aRGB for the camera and shoot in Camera RAW, scan in Prophoto, print with the Epson 3880 using Prophoto. Does Photoshop use Prophoto?
Thanks, Bill

[Jim Andrada]

There's always LAB color - I've read a couple of books that make some reasonable sounding claims about it that (like a lot of things lately) I can't quite remember. Oh well, better to forget things that happened than to remember things that didn't I guess. At least I'm not quite to the point where, when I go to a party, I have to wear my "Hi, I'm Jim" badge upside down so I can remind myself who I am.

[Bill, 70's military B&W]

LAB Color... a friend of mine has shown me how to use LAB color to isolate the B&W channel, then sharpen the B&W only, recombine the channels and you have a sharp image without the edges of the colors becoming distorted. Seems to work. Any comments? The sharpening part of digital photography kicks my butt.
Thanks again, Bill

[Greg Miller]

LAB Color... a friend of mine has shown me how to use LAB color to isolate the B&W channel, then sharpen the B&W only, recombine the channels and you have a sharp image without the edges of the colors becoming distorted. Seems to work. Any comments? The sharpening part of digital photography kicks my butt.
Thanks again, Bill

You don't need LAB to do that. In any RGB color space, just change the Blending Mode of the layer that you want to sharpen form "Normal" to "Luminosity". This sharpens the gray scale (but not hue or saturation). Pretty much the same thing as sharpening only the L channel in LAB and without the ill effects of switching to LAB and back.

[Jim Michael]

Exactly.

And there's no downside to using a larger color space, assuming you're working with 16 bit images. Use the wide space, use a color managed workflow and calibrated monitor, and then soft proof to your final intent, whether it's to sRGB for the web or to the profile for the printer you're using.

Easy peasy.

When the image is ultimately mapped into an 8 bit device why don't you have an increased risk of color banding in a smooth color gradation when using the wider gamut?

[paulr]

When the image is ultimately mapped into an 8 bit device why don't you have an increased risk of color banding in a smooth color gradation when using the wider gamut?

You don't as long as you're doing all your potentially destructive edits—including profile conversion—in 16 bit mode. You've got 256 times as many gradients working in 16 bit, so any loss, banding etc., ends up way below the resolution that exists in the final 8 bit mode, which itself has enough gradients to appear completely smooth to the eye with the dynamic range available on any print or monitor.

If you were editing in 8 bit mode, then yes, it would be wise to use a color space that's only as wide as you need.

[Jim Michael]

The reason I asked is that I noticed something odd when making a color slope adjustment in LAB mode on a 16 bit file. It was an aerial image with a smooth sky gradient. I was increasing the slope on the A & B channels to mitigate the effects of atmospheric haze, but it induced banding.

[paulr]

Hmmm. I don't know why that would be. Not much experience with LAB though.

[Mark Stahlke]

The reason I asked is that I noticed something odd when making a color slope adjustment in LAB mode on a 16 bit file. It was an aerial image with a smooth sky gradient. I was increasing the slope on the A & B channels to mitigate the effects of atmospheric haze, but it induced banding.

Jim,
Did the banding show on a print?
Most monitors are only 6 or 8 bit. A 16 bit image could appear to have banding on the display that's not in the file.

[Jim Michael]

Jim,
Did the banding show on a print?
Most monitors are only 6 or 8 bit. A 16 bit image could appear to have banding on the display that's not in the file.

That was sort of the point. If the output device is 8 bit whether print or display then it would seem logical that there would be a higher banding risk when the gamut is wider. Although the issue I was seeing wasn't necessarily a gamut issue, it just occurred as a result of increasing the color slope (e.g. convert to LAB mode, curves menu, drag top to left and bottom to right equivalent amount on A&B channels).

[Mark Stahlke]

That was sort of the point. If the output device is 8 bit whether print or display then it would seem logical that there would be a higher banding risk when the gamut is wider. Although the issue I was seeing wasn't necessarily a gamut issue, it just occurred as a result of increasing the color slope (e.g. convert to LAB mode, curves menu, drag top to left and bottom to right equivalent amount on A&B channels).OK. I get it. My printer and software do 16 bit printing so I assumed...

[paulr]

Another possibility ... if you're using a monitor that doesn't have hardware calibration, then all color management happens by way of the video card adjusting its output. This adjustment always happens in the form of reduction of the size of the color space (since it's not possible to increase it). A monitor that's normally 8 bits per channel can easily be reduced by a few bits if a lot of correction has to be applied. This can lead to banding under all kinds of typical circumstances.

[bob carnie]

Jim _ I work in LAB quite a bit , how far left and right were you draging the A & B.
Also Large moves in Lab can indeed be destructive, It is the largest colour space and if you go too far you are going beyond the monitors capabilitys of showing the colours LAB wants you to see.

I will overamp sometimes to spread the colours far apart from each other then with the opacity on the layers pallette bring it down to a more logical look. Then I use this to send back to RGB for some pretty good BW conversions.


That was sort of the point. If the output device is 8 bit whether print or display then it would seem logical that there would be a higher banding risk when the gamut is wider. Although the issue I was seeing wasn't necessarily a gamut issue, it just occurred as a result of increasing the color slope (e.g. convert to LAB mode, curves menu, drag top to left and bottom to right equivalent amount on A&B channels).

[Jim Michael]

I wouldn't say it was an excessive correction, say maybe 1-1.5 cm on each end. But thanks for the analysis, that helps.

[bob carnie]

There are lines that you can drag too, after about two lines you are in dangerous territory unless you want to use opacity to lower the effect

I will never bring in client work that is from small camera files, also I would not do colour adjustments in LAB after sharpening.

otherwise I not sure what would be causing visual banding on screen.


I wouldn't say it was an excessive correction, say maybe 1-1.5 cm on each end. But thanks for the analysis, that helps.

[Ohnostudio]

To Bill, the OP,

One thing that may help you understand things a little better is that everything device has a profile. Your monitor and printer have them, and even projectors do. The profiles help the devices talk to each other for best match purposes. Even specific papers can have a profile. Since you have that large 3880 some learning would be beneficial so that you don't blow through ink :)

Keep an eye on this page at XRite

http://www.xritephoto.com/ph_learning.aspx?action=webinarsarchive

Any day now they should be publishing a webinar that was aired on or about Jan 17. It is a good one. While it does of course promote XRite products, it also gives you enough ammunition to approach color management solutions sensibly.

Personally I don't even have a printer at home anymore except for a laser mono. All of my own stuff goes to a lab or is delivery by digital file.

[Bill, 70's military B&W]

Ohnostudio
Thanks for the link, I'll be watching for that webinar. It is a good site with a lot of information. I added it to my favorites.

Bill

[oysteroid]

With respect to the banding on monitors that isn't in the file that was mentioned, I'll also point something out that I wish I had known before I selected a video card. The DVI interface does not support 16 bit per channel color. You have to use a card that has a DisplayPort port. But that isn't the end of it. Consumer level gaming video cards, even those with DisplayPort, don't usually have drivers that support the higher bit depths. You have to get the Quadro or FirePro type cards. I have a monitor that supports high bit depths, but a card that only has DVI ports, so I can't use DisplayPort and the banding in grayscale images onscreen is horrible. Wish I had known. I will probably buy a different video card one of these days.

[oysteroid]

Actually, it seems I was wrong about my monitor. It only supports up to 8bits per channel, even with DisplayPort. Apparently, even the ones that support deeper color only support 10bits per channel. So maybe banding in grayscale images onscreen is mostly unavoidable except with super high-end gear.

And if I understand things correctly, if you are displaying pure grayscale images where the red, green, and blue values being sent to the display are all equal, with 8bits per channel, you only get 8bit grayscale. 2^8=256. So you only have 256 levels of gray. With 10 bit color, you get 1024, which is better. But you still might see banding in long gradients without noise or dithering.

[Mark Barendt]

Steve Upton had an article a while back that spoke to both sides of this question. While most of us here are "Input Centric" there's a place in the world where some can be "Output Centric":
http://www.colorwiki.com/wiki/Color_Management_Myths_26-28 (http://www.colorwiki.com/wiki/Color_Management_Myths_26-28#Myth_27:_Why_would_anyone_ever_want_to_choose_a_working_space_that_is_larger_than_you_can_print.3F)

Thanks for the link.

This input vs output centricity has been with us a lot longer than digital, essentially it's shooters vs printers.

Adams tried to address the problem wholistically with visualization and the zone system, the paper and the lighting where the print was to be viewed were the ultimate arbiters in his world, it was the only practical way to see and share his results. IMO the basic intent of the zone system is to make printing easier. I think it is safe to say Adams was output centric.

One thing that helped me immensely when deciding on color space was knowing how and where a given photo would be used. Like knowing I'm shooting to print on grade 2 paper, it gives me a target.

A second helpful thing was an understanding that the color space I work in doesn't really matter, there is no need for fidelity here. In the digital world all the color we see is contrived, it is purely the result of fully adjustable parameters, and it is completely subjective. When I proof my output, if I don't like the color or its luminance or whatever, I can change it and I can tweak the system so that that decision can be reflected on all my images going forward.

The third thing that really helped was my decision that I don't care about updating my old photos to future tech. That is simply a personal thing, I'm just not into reinterpreting my old negatives. I'd much rather go shoot something new.