edit: I put together the pieces! yay! :D
But, my understanding is that for an image displayed 1-1 pixel-to-screen there is no intermediate math going on. So, there shouldn't be any opportunity for non-gamma-correct calculations to screw up the output.
Here's a nice example of non-gamma-correct browsers producing incorrect results after scaling an image: http://filmicgames.com/archives/354 also http://www.4p8.com/eric.brasseur/gamma.html But, in the SO question there is no scaling. So, I'm still confused about the difference in visual results between different applications directly displaying the image...
So obviously my various image viewers aren't reading the gamma correction to show the pear, however I'm more curious why I don't see the apple correctly if the image has been intended to show the apple without gamma/color correction?
My monitor is color calibrated, so shouldn't I just see the apple and not the pear?
It's a mix of Tux and the GNU logo.
The strange thing is that Firefox shows Tux, and Chrome shows the GNU logo. Very strange considering both Firefox and Chrome show the pear. My understanding is that the GNU logo (the 'lighter' image) should be shown if the gamma correction is applied. Does anyone know what's going on here?
EDIT: By removing the iCCP color profile chunk, the image displays properly in Firefox.
For anyone interested in making one:
First, I put the two images on separate layers in photoshop. I changed the output levels on Tux to be from 0 to 210, darkening it slightly. I changed the output levels on the GNU logo to be from 215 to 255, making it very bright. Then I used a layer mask on the top layer with a grid pattern to mix the two images. The saved image was a mix of the relatively normal Tux pixels and the bright GNU pixels.
The final touch was editing the gAMA value in the PNG. Using the free program, TweakPNG, I set it to something very low, like in the pear image: 0.02. And that was it!
http://i.imgur.com/hTauClg.jpg (screenshot of what I see in firefox)
I apologize if this is a dumb question.
The reason the screenshot doesn't appear different in Chrome is because no gamma correction is attached to it.
Anyway, check this new link:
This is the same image with the color profile removed. It should show the GNU logo in both Firefox in Chrome, but show Tux in browsers which ignore the gamma correction.
I apologize if my explanations are confusing... I'm very new to this stuff.
In this case black and white fields of a checkerboard correspond to pear and apple pixels, so you are left with one of the images.
Off topic: how are you finding windows 8?
On my desktop I initially would use the metro stuff here and there but after few a few months the novelty has worn off and I very rarely venture into the metro stuff. When used that way Windows 8 is just Windows 7+. I think one of the greatest things about Windows 8, is the promise of Windows Blue, Microsoft's commitment to pushing out major upgrades.
In summary, on a tablet Windows 8 is excellent, on a desktop it's at times between ok to pretty good.
Edit: here is a screenshot from my tablet of some of the printers that I've connected to in my travels http://i.minus.com/it1xr6gbOnAFx.png
(My copy of GIMP is a bit outdated, but that shouldn't be a big deal for a basic feature like this, right?)
"GET A REAL BROWSER"
You know the Bob Boss uses IE so you send a link to a "specially crafted" web page that uses an image like this to show the boss a completely smart solution (place high value object in this location inside vault) and ask him if it is OK. Now just get Bob to forward the link to the web page to Alise who uses some other, known web browser and she sees a map with a location outside the vault.
Just a thought: for now if you own the server anyway you can just do browser sniffing and send two completely different images..
Regression tests are fun!
This goes back to the late 1990s; the page mentions MSIE for Solaris.