Sometimes when I click to enlarge an anaglyph the red/cyan colours of the thumbnail change to yellow/blue, making the normal specs unsuitable for viewing it. This happens to Stu's latest one this morning, for example, though only a minority of his anaglyphs are thus affected. It occasionally happens with other posters too, but I can't spot any systematic pattern that would suggest a cause, much less a remedy. Any suggestions?

In one of the other threads you mentioned your screen is set to very high brightness (something you should fix if you want to see the images here as they were intended to be seen). Maybe it magnifies small color differences between color images that have embedded color profiles (which are other than default sRGB) in them and the automatically generated thumbnails by the forum software which discard that profile. Can you point to one of those changing attachments?

No, it's not that. The two colours change totally between the thumbnail and the full size versions. I saw it most recently today when viewing Stu's latest outcrop/thingy picture via the college computer system. Now I'm at home on a separate computer (the one with the bright screen) and that same anaglyph is behaving fine with no odd colours on enlargement.

I think I've seen it happen on my home computer too on past occasions, but I can't be absolutely sure of that.

Your computer is haunted, that's the only explanation. Give Yvette Fielding a ring, she'll sort you out.

Sounds like you're almost sure it only happens on your work computer, which makes me wonder about a faulty monitor or video card. Next time it happens, would you be able to try the same image on another nearby system?

For reference I just checked all of Stu's recent anaglyphs and I don't see any change in colour between the thumbnails and the full size image with my default browser (Chrome 4 bleeding edge developer release..). This is on a Windows 7 laptop for what its worth.

However I just checked using IE8 on the same system and what do you know I see the symptom:

This post: http://www.unmannedspaceflight.com/index.p...st&p=153158 has the problem.
While this post: http://www.unmannedspaceflight.com/index.p...st&p=153175 does not.

Both are in the Post Marquette Island thread.

The metadata in the "non buggy" jpg has the following entries "Program Name: Adobe Photoshop elements & Color Representation: sRGB" that are not present in the "buggy" version. It appears that IE8's default handling of jpg's differs from sRGB while Chrome (and MS Paint on Windows 7 for that matter) doesn't.

Interesting behavior.
Edited to add:
There's some more information on something that seems to be related in this blog post with a fairly wild example - if the bike is Green then your browser applies the embedded color profile, if it's blue it ignores it and if its red it's using the Microsoft specific WCS color profile. On my machine Chrome and IE8 display this as Blue, Safari and Firefox 3.6 display it as Green so there is some difference between this example and stu's images but I suspect it is still related.

That fits. The college system uses IE(don't know which version) and my home computer Firefox.

You gotta love IE... Never mind the fact sRGB is the de facto standard on the web and elsewhere and is assumed by default.

So, does this mean that the internet is littered with images that appear differently coloured on different browsers? How is one to know which images are suspect? I'm thinking of a graph or diagram with lines in different colours that are referred to in accompanying text by the colour names - this could play havoc with the the meaning of an article or paper. How prevalent is it? (Might help if I knew what a 'buggy' was - presumably not 'something with a bug'.)

Can you fix it for them Joe?

I just fired up IE 8 and did a grab with the (ahem...Vista ) snipping tool

Attaching both for the sake of shear interest. Really unusual. Another reason not to select your painting and decorating colors via the Internet.

EDIT: Firefox on the left, IE on the right.

This is really weird and bad since it really makes it impossible to completely trust the colors one sees in images using IE. It's easy to get around this problem though by saving the full size image and then viewing it using something else (i.e. not IE).

Firefox displays these images correctly.

ngunn: "buggy" is precisely that - an incorrect behavior that I have no good explanation for..

I knew there was a problem with browsers handling ICC profiles a number of years back but I had thought that Firefox had sorted that out and had naively expected that all the recent browsers had followed suit. I haven't looked into this in years but I've just been digging around and it seems that the problem is better in some ways but its now more complicated than it used to be. Not only are there browsers that don't support them but there are multiple versions now and even good old Firefox has some issues:

Check out this link:

The only browser that I can find that handles this example completely is Safari V4 (the picture is reasonably consistent across all 4 quadrants). IE8 and Chrome fail this in the same way and only appear to render (reasonably) correctly for the Adobe ICC V2 color profile. Firefox 3.5 & 3.6 renders OK on three of the tests but fails on the v4 YCC-RGB color profile.

However in my testing of Stu's Anaglyphs:
Chrome V4, Firefox 3.6 & Safari 4 all display the "correct" colours, IE8 is the only one that fails. However it is failing for a case where the color profile required should be sRGB which IE8 appears to be able to apply in the ICC test example above. I think what is happening there is that this is the default behavior for a jpg file that it thinks is error-free. What I think is happening in this case is that IE8 doesn't like the fact the this image has no EXIF meta-data, or possibly there is EXIF data there and IE8 then fails in some fashion. What it's doing is completely dropping the red channel which leads to the color change we see, I've found some evidence of others reporting similar issues with IE8 for files that have had all EXIF metadata stripped out.

I suspect that files built\saved using sRGB should be rendering reasonably consistently (within the limits of our Monitors to accurately display the required color gamut) on all modern browsers but IE8 fails quite badly when it hits files it finds "difficult".

Oh hell, I didn't break the Internet did I?!?!?!



Just wondering if it's a result of sometimes initially making a 3D image with Stereophotomaker, altering its brightness/contrast etc in Photoshop and sometimes cropping it with Paintshop. Could that mix-match of progs affect this?

Stu, you know if you broke it, you bought it....

Gotta be an IE8 problem. I've seen the same effect on many anaglyphs by various people here.

Didn't your mother teach you to not prepare pork and chicken on the same cutting board? I suspect it's something like that. You gave the Internet image poisoning.

Working with internet content on a daily basis, the best approach is to NEVER use any color profile, other than the default sRGB (if you really need to have it). It's the only why to get the same RGB values to display the same across a JPG, text, site background, PNG or GIF icons, etc. All other setting will give you variations on what you actually see. That's how things work in the real world !

Good grief. (Edit - not directed at 4th Rock's comment, but at the root cause being an MSIE bug.)

The color variation problem is not limited to IE. Firefox also doesn't support all the color space corrections it should do.
On the other hand, operating systems also have different profiles. Windows and Mac have traditionally different gamma values.

In practice, all devices should display/use sRGB or at least a good approximation of it. Therefore, an untagged JPG image will be handled as sRGB, and look "normal". Specifieng sRGB will make no diference.

Wide color spaces might take advantage of LCD capability in displaying more rich colors, but in practice don't work across an open medium such as the internet. Inside Photoshop, it's a diferent matter.

I sometimes use specific color spaces for certain camera/filter combinations, such as Voyager's OGV. The system converts data from those filters to sRGB in realtime, the same as it does for CMYK for examplo. But the final image must be converted to the default color space.