Luminence value for web output

[digitaldog]

Well one solution to the OP's question (Does anyone have any thoughts about what to set  your luminance target for your display in preparing files for websites?) is this: take a pile of postIt notes, write a group of values on 10 from say 100-200 on each, tape them to the wall, throw a dart and the value the dart hits is the answer. 

We've got basically three groups. One is very sophisticated about color, photography and imaging (the vast majority of the LuLa audience). They are not using CRT's left over from the last century, they work with color managed applications including web browsers. And they care how they and others see their images.

The 2nd group isn't anything like the first. Many probably are working with 15 year old CRTs that today might output 75cd/m2 or they have to buy the cheapest LCD they can find. They will not calibrate because they have no idea that that is or if they do, they will not spend the money to do so. Futz with the OSD controls, maybe. They are unaware their browsers are not color managed. As long as there is any degree of color emitting off the display, they are basically somewhat of happy.

There's a 3rd group that's very much like the 2nd but whatever they see, it's wrong and they don't like it and will tell you how bad your images are. This group is helpless.

So basically the question is, aim for group 1, or aim for group 2-3 and suffer no matter what dart hits the PostIt note value?

One solution for group 1 and some group 2 stragglers who really do want to navigate to group 1 is to place a black to white step ramp somewhere on the web page and tell people who don't know otherwise that they should be able to make out say the 21 steps and if they see blown out whites in several patches, or blocked up blacks, they are incorrectly viewing the web images. A URL to a site that will inform them if their browser is color manages is useful too*. IOW, education. People either care about the color they see and strive to get it, they care but don't know how to achieve it or they don't give a crap and will complain no matter what you do.

Even if we agree on some ISO spec for cd/m2, it does no good for group 2-3. I'm fairly certain that within group 1, just here on LuLa, there is a range of calibrations for cd/m2, probably between 120 - 160 cm/m2, some lower because they wish to emulate the CRT's of the past. Might make an interesting poll. I calibrate to 150cd/m2 because that setting produces the closest visual match to my viewing booth. Yet when I travel about the web using my color managed browser, I cannot recall seeing 'quality imagery' (not the offering of group 2-3) that doesn't look AOK. The images look fine.

As to Peter's link, if Karl states it, you can probably take it to the bank. There are few people around who both understand this topic more and have brought to market two of the finest display products in imaging history (Radius PressView, Sony Artisan) and invented ColorMatch RGB. And it helps he really is a color scientist. Whether that equates to 200cd/m2 (seems a bit high) or not would make an interesting experiment for group 1 and do nothing to help group 2-3.

Frankly I do not think this a hill worth dying on. If your only concern is output to the internet and you feel 200cd/m2 is a good middle of the road, and you understand that you simply can't satisfy all the people who will view your images with or without color management, go for it. Or just throw that dart and see where it lands. In the grand scheme of things, probably doesn't matter. How many people viewing your images on the web are even using a browser that understand the numbers it spits out?   

*This is probably enough info for this group to understand: http://www.color.org/version4html.xalter

[D Fosse]

So following the logic of the above quote, when prints correspond to an avg. 250:1 dyn. range,
and assuming that ca. 125 cd/m2 on screen work for print-image-matching,
it makes a black luminance of 0.5 cd/m2,

and with a 400:1 monitor dyn. range,
it finally points to a white luminance of 200 cd/m2.

There's a basic flaw here. The dynamic range limitation in print vs. screen is the black point, not the white. Matte paper has lower contrast than glossy because the black point is higher.

To extend the dynamic range from 250:1 to 400:1, all you need to do is let the monitor go down to its native black point, which is usually in the 0.3 range (or lower).

A couple of calibration targets with different black points will take care of all this, and you can still have the same white point and overall luminance.

Great post by Andrew, summing it up nicely. Personally I just care about group 1 and ignore 2 and 3.

[Simon Garrett]

Great post by Andrew, summing it up nicely. Personally I just care about group 1 and ignore 2 and 3.

I agree. 

The 3rd group are beyond help so there's not much point worrying about them.

Group 2 are the "fat, dumb and happy" group.  No point providing images that will be "right" on their monitors, as every monitor will be different.  However, people get used to how images look on their monitor.  They adapt to what they see. 

An interesting recent example is this.   Colour photometry expert Taylor Swift sees only blue and black in the image.  She notes she's "confused and scared", which I would be if my monitor were that bad. 

By the way, I get a bit nervous about the v4 test page on the ICC web site, as it doesn't show what people think (or even quite what the site says).  It just shows whether the browser being used makes an attempt to use v2 and v4 profiles.  It doesn't show whether the user's monitor is calibrated and profiled, and it doesn't show whether the browser is fully colour managed.  For example it apparently shows that Internet Explorer is fully colour managed, which it isn't. 

[digitaldog]

By the way, I get a bit nervous about the v4 test page on the ICC web site, as it doesn't show what people think (or even quite what the site says).  It just shows whether the browser being used makes an attempt to use v2 and v4 profiles.  It doesn't show whether the user's monitor is calibrated and profiled, and it doesn't show whether the browser is fully colour managed.

Yes, it could be more clear and it isn't the best designed page for group 2. I also like another URL on the topic but group 2's heads will explode:

http://www.gballard.net/psd/go_live_page_profile/embeddedJPEGprofiles.html

The answer again is education. Like the step ramp, instructions have to be written for the group 2 users who hope to migrate into group 1. They will need hand-holding. I suppose if we left these two pages as our choice for informing them, we could add some text explaining they go to this URL and if big top image looks like the last and smaller image (multiple color renderings), they have a problem. I mean for some, we'll need to explain something as simple as gray ramp path 1 and 2 (or 20-21) have to show a separation of tone. We can't assume they even understand what "your #2 dark patch is blocked up" means without full explanation.
If someone wants to build a page for group 2 that's more self explanatory, be useful.

Funny timing, I got a call today from a friend who stated when he views his product packaging on his web page on the Mac, it's fine but looks awful on any mobile devices. He knows to send his web guy sRGB files. I went to his site, downloaded the image, it was in ProPhoto RGB! Whoever was working on his web page somehow uploaded the wrong JPEG.