On the left is a view from a non-colour managed app (Image Viewer), and on the right ‘geegie’, fully colour managed.
It’s the same pic, just opened in separate viewers.
I used ProPhoto as it’s working profile and saved as a 16-bit integer tiff. The png above is a screen dump cropped with Gimp2.8.
Now most people won’t have colour managed apps, so they’d be seeing the disgusting mess to the left. How can this be changed so that whomever will see the same as the pic on the right?
Any help would be much appreciated, just go easy on the technical jargon please.
There is literally no way to solve this 100%. When you publish to the web, you’ve relinquished control of many of the viewing variables that lead to a pleasurable viewing experience.
Further, I don’t think most people care about having color managed things; they care about having an image that looks good to them, which mostly means too much contrast and saturation.
I’ve been turning more to printing things, then showing people in person. I actually like that a lot more.
I understand your point, but doesn’t that just make you feel partially satified with your hobby/profession? I mean, most people will first see your pics on the Internet, and if they look as disgusting as the image above left, they probably wouldn’t be interested in seeing the colour-managed (above right), printed version, would they?
Also, how do those photographers on say, 500px, get on? Their images look great, both colour managed or not.
I make art because it is something I need to do for myself. If I make something that I am happy with, then that is all I need; any additional positive feelings I get from sharing my work with others is purely a bonus and is not at all necessary for me to continue to do what I do.
What’s interesting is that when I change the system colour management from the recently calibrated version to the monitos default, the colours in Image Viewer become like those of Geeqie. So what’s happening there? Should I continue to use the default monitor profile?
The Image Viewer one still has an abundance of noise, whereas Geeqie has some but not a lot.
Note: Geeqie is not yet “fully color managed”. It does not support rendering intents, meaning anyone who uses a profile with a perceptual rendering intent in Geeqie will see complete rubbish. This has been fixed in git but has not been merged into master yet.
Anyone with a github account should add their voice to that issue so that it gets merged, its extremely important.
Welcome to just a glimpse into cross-OS/cross-browser web development hell…
There are many longer answers, but this basically just boils down to a relatively simple solution (that you may not like): if you want to make sure the majority of folks are able to see your image the way you want, then you have to edit for the lowest common denominator.
This means sRGB if you are targeting computer viewing. Or at the very least, adding a step to your processing to create an sRGB version you’re happy with?
Usually, the thing driving this is how far along the browsers are with color management (as I assume that’s how most folks viewing things on the internet will be seeing the images).
Regardless of whatever colour management you do or don’t use, export your for-web images in sRGB, converting them to sRGB from ProPhoto, AdobeRGB or whatever if necessary.
+1. I understand that the advantages of AdobeRGB in very colour-rich photos are far outweighed by the time-consuming issues it often throws up for non-expert users, as well as the more general issues related to sharing non-sRGB images on the web. sRGB may not be the best in terms of absolute gamut, but it does allow one to get on with image-making and easily covers the needs of hobbyists such as myself.
I really think you’d find things much, much easier if you make your whole workflow sRGB, and then ensure that your OS and monitor are set to use sRGB and adjust your monitor image controls so that things (websites, etc.) appear sane and balanced. That alone should get rid of issues such as the purple.
as all others said this colour management in the end is more or less just related to where those images are to be displayed.
I can’t tell about you @Fotonut but from my images 99% are just displayed on several types of screens, smart phones, notebooks, computes and tv screens. For all those images I use a sRGB for the end result which also always is a JPG. I said “a” sRGB just because from what I’ve learned so far sRGB isn’t always sRGB. I’m sitting at work but as I remember it’s an ECI sRGB ICM I’m using.
So the typical workflow for those images is: developing from original - this depends with which cam it was taken (dslr or mobile or …) So if it is RAW and the raw developer gives a pleasant result the image will be exported as JPG with sRGB - and done. If I want to tweak a little more with a image editing software able to read 16bit images the interchange image will be exported as 16 bit tif using ProPhoto and from the editing software the end result will then be exported as JPG using sRGB. The 16 bit tif will be deleted then.
For some images, those I’m not sure how they will look at a not calibrated monitor, I use a old (8++ years) notebook with a Linux on it - I think it’s Lubuntu but not sure. This old one is’nt calibrated at all and the software there does not care about colour correction or what ever, so this I use to check some images and if I like what I see there then it will be put for viewing.
For all the other images which I want to be printed this workflows differs by using other colour profiles but this depends on the printer or better the printing company I will call for that. For some i.e. a calendar or a photo book it’s sometimes also not that bad idea to use sRGB but as I said it depends on the company, some also give you a profile.
I’m not a professional but so far this also did work for some ordered photo sessions I’ve done. All my clients were able to get their photos in that colours they wanted.
as @paperdigits said after all it’s about enjoyment and not about being colour correct in all ways. This differs if you have to work for professional agencies but then it’s a job and not a hobby and as long it’s a hobby and as long as you like the result anything is correct well at least I think.
I totally agree with what @paperdigits, @patdavid and @DavidOliver have said: the digital entertainment technology is, simply speaking, not yet artist-friendly. There are too many variables still playing a role in the creation of the final image on screen:
most viewers (not artists…) do not know anything about color management: they do not use color-managed applications and, if they do, their monitors are anyhow not calibrated
even if they do everything correctly, their devices probably cover a gamut even smaller than sRGB
So, my recommendation is the following:
do your edits in a wide-gamut colorspace (Rec.2020 or ACEScg)
save a master copy as a floating-point TIFF in the same working colorspace, or an even larger one (ACES for example)
for web display, go from your master copy to sRGB, and make sure the sRGB profile is embedded in the final image. That’s the best you can do.
When going from a larger colorspace to sRGB, you have two choices:
do a straight conversion and clip all the out-of-gamut colors. This is the easiest solution, but might lead to some loss of details in the clipped areas. Do not try to use the “perceptual” rendering intent when converting to sRGB: the sRGB profile you’ll be using most likely do not support this intent.
reduce the saturation (globally and/or selectively for certain colors) until all the image fits inside the sRGB gamut. This can be a rather lengthy process, tailored to each image, so you probably want to do that only for some selected one…
Note that you will not see the colors outside the sRGB gamut, because your display is most likely an sRGB device as well… the best way to judge if some colors fall outside of the sRGB gamut is to make use of the gamut warning functionality of your editor, if it is implemented.
I’ve been very much enjoying darktable’s soft proofing feature, which shows you the difference between two profiles in the context of your current working image.
Thanks @DavidOliver, with regard to the purple haze, I discovered that was actually a bad edit. I had added magenta for effect, but failed to section it where I needed it. I could definitely see the haze after a while myself, particularly when zoomed in. I guess I hadn’t paid too much attention to it prior to posting.
This is excellent advice. Some years ago I had to selectively reduce the saturation of the cyclist’s jacket to stop it looking truly awful when exporting to sRGB:
The reds looks glorious in ProPhotoRGB. In sRGB they look so dull, but such is life
Well, the big issue is that you actually never saw the the reds in ProPhoto, only some over-saturated reds that got clipped to your display’s gamut (probably sRGB or very close).
In fact, the most instructive thing you can do to check how good the colors are reproduced in a certain working colorspace is to look at the individual RGB channels.
I’m pretty sure that the red jacket still had some details in the blue and green channel when edited in ProPhoto, while all those details had been sadly lost after the conversion to sRGB.
Looking at individual RGB channels is the best way to visually judge what information is present in a wide-gamut colorspace while using a small-gamut output device (like your display).
Point taken. However it was edited on a screen with 92% of the NTSC space, and the difference when the photo was converted from ProPhotoRGB to sRGB was very noticeable on that screen. The red jacket looked so much nicer in the wider gamut colour space.
I’m assuming that a 92% NTSC gamut is basically a 100% Adobe RGB display… since Adobe RGB was an erroneous copy of the NTSC color space.
Adobe RGB only adds additional gamut versus sRGB in green and cyan, it has exactly the same red and blue primaries.
So really, the something is wrong with your processing if it looks worse in sRGB. Perhaps when you’re displaying it in sRGB, it gets clipped in an ugly manner, whereas when you’re displaying the ProPhotoRGB one, the perceptual conversion is desaturating the most extreme colors so as to prevent clipping.
Regarding out-of-gamut colors, I found that this CC curve works well for me to prevent color clipping when working in ProPhoto and saving to sRGB, so I set it as the default CC curve in my defauly Pentax K10D profile:
Regarding @Fotonut’s original issue: if you share the raw file and PP3 then we can stand a higher chance of finding the problem.
Or at the very least, adding a step to your processing to create an sRGB version you’re happy with?
well at least I think.
