Oh it's much worse than that, at least nowadays, for some people. It's not something that you can turn off, everything is going to be bad for a long time and it will get worse before it gets better. 255:0:0 means the reddest red. But my wide gamut monitor's reds are like 50% redder than yours (and next year's wide gamut monitor's will be 50% redder still). So of course it looks different if your web browser just sends #123456 to my monitor and to yours. The same RGB triples are being mapped over a bigger color space. Graphic formats have provisions for metadata so perhaps your 12:34:56.png is tagged as meaning 12:34:56 in the sRBG color space and the browser dutifully maps that to whatever RGB triple would look the closest in your monitor's color space. The background is probably untagged, or tagged differently, or IE8 defaults images to sRBG but doesn't for backgrounds. One way or another this is Microsoft's fault. For their web browser or OS or whatever. Being as big as they are they should have strong-armed all the monitor and video card vendors to support 10 bit color channel resolution and then assumed that every color on screen is in sRBG space unless specified otherwise by apps like Photoshop that actually know what they're doing.
On Wed, Aug 19, 2009 at 2:23 AM, Peter Pentchev <r...@ringlet.net> wrote: > On Wed, Aug 19, 2009 at 11:40:05AM +0300, Yossi Kreinin wrote: > > Alan Amaya wrote: > > > On Tue, Aug 18, 2009 at 5:43 AM, Yossi Kreinin > > > > > > The color issue is likely due to color correction / gamut mapping. Any > > > recent non-professional wide-gamut monitor will soon leave you wishing > > > it wasn't. What is #44ffaa after all, but some hex digits which never > > > look the same. A photograph though, that should look real. > > > > > > > But why doesn't #44ffaa display *the same* in the page background and > > the embedded PNG? I don't really notice small differences in the signal, > > but a non-zero derivative where there should be a zero derivative kinda > > sticks out. Also, why do other browsers work just fine? > > Because somewhere in the process of: > - fetching the webpage source > - parsing the webpage source > - rendering the webpage to a virtual "window" > - mapping this virtual "window" to your virtual "desktop" > - combining the palettes of the "window" and the "desktop" > - rendering the combined... thingy... to your physical display device > - obtaining a screenshot of either the "window" or the "desktop" > - storing that screenshot's image into your virtual "clipboard" > - fetching the image from the virtual "clipboard" > - storing it into a virtual "image file" with its own format and > limitations > - mapping this virtual "image file" onto the next webpage > - combining the palettes of the "image file" and the webpage > > ...something, somewhere, has decided that one of those virtual thingies > does not have enough colors in its palette to display all the webpage's > colors exactly. Thus, it needs to remap some colors to some similar, > but not quite the same colors, and your #44ffaa is no longer a strict > specification, more like... I don't know, a piece of friendly advice, > maybe? > > Now, as to why a single browser would do some of those things differently > than all other browsers - sorry. I don't know. Feel free to hate it. > Ah, you already do. Good. > > And yes, I am fully aware that your graphical environment is probably > one of those newfangled big-memory technicolor gimmicks with something > like 32K or 64K or 16M colors, which no sane person would ever need > (256 colors should be enough for everyone, right?). And yes, I am fully > aware that this means that no color/palette remapping should ever be > needed. And yes, I am fully aware that you probably knew all this. > Okay, as a compensation for your time spent parsing this message, as > a special offer you may send 1 (one) hate message to hates-software@, > limited to 150 kilobytes, on a topic of your choice. Offer valid only > until end of current week, no rebates. > > G'luck, > Peter > > -- > Peter Pentchev r...@ringlet.net r...@space.bg r...@freebsd.org > PGP key: http://people.FreeBSD.org/~roam/roam.key.asc > Key fingerprint FDBA FD79 C26F 3C51 C95E DF9E ED18 B68D 1619 4553 > If I were you, who would be reading this sentence? >
