Fun Fact: What your eyes see is completely different from what the real world actually looks like. Most people imagine that our eyes measure the world and display it like an LCD display inside our heads. In reality, the human visual system is extremely complex and does a tremendous amount of processing. And that’s why we need HDR Photography if we want pictures to look like the world that we see (which is not the actual world). If that’s confusing to you, read on. But be warned: You will never trust your eyes again.
1) The Adelson Checkerboard
Let’s start off with a bang. The Adelson Checkerboard is the coolest optical illusion of them all. First, just look at the image. When you mouse over the numbers below it will swap to a different image, so don’t do that until you read the full explanation.
So what’s the trick? The squares labeled A and B are both the same color. Really. No joke. Seriously, I’ll bet you $1,000. If you open them in Photoshop you will see that they both have the value 120. Now mouse over the numbers just below the image. As you move farther along the gradients get covered and you can see the colors for what they truly are. The other way to convince yourself that they are the same color is to draw a thick line between them which I do in the last images.
The moral of the story? Your eyes lie to you. When you look at the real world your eye de-emphasizes the soft global gradients (like the soft shadow) while emphasizing the strong local contrast gradients (like the tiles).
2) Hard/Soft Edges and Gradients
This illusion was new for me. I was at Siggraph in early August and saw the paper Edge-Aware Color Appearance.
The image (a) on the left is the original Siggraph Logo. If you perform a vanilla gaussian blur then you get image (b). But somehow image (b) appears lighter than image (a) even though they are the exact same color. Their paper discusses tweaking the image into (c) which is actually darker but perceptually looks the same as image (a). If you mouse over the second image you’ll see a rectangle that shows that the red of (a) and (b) is actually the same color. Yet another example of our eyes cheating the local gradients.
3) Edge Detection
Here is an image that I’ve seem many times. Instead of finding it I just made a new one.
Do you see how the very left end looks brighter than the very right end? Well, they’re actually the same color (127). The left half starts at 127 but as it gets closer to the center it ramps up to 140. The right side starts at 127 but as it gets to the center it ramps down down to 113. Your eye thinks that the harsh gradient from 140 to 113 is more important than the soft gradients that get you back to 127. In other words your eye biases your local gradients over global gradients.
4) Imagined Gradients
So far I’ve shown you cases of your eye thinking that local gradients are more important than global gradients. But it goes deeper than that. In the image below, you should see a large rectangle with a gradient and a wide-thin rectangle in the middle with a gradient going the opposite direction.
However, that gradient in the middle doesn’t actually exist. If you cover the wider gradient (which you can see by going from numbers 2 to 5) then you see that there is no gradient in the middle–it’s just a solid rectangle. That’s how seriously our eye takes sharp, local contrast. Our eye thinks that the local contrast between the inner and outer rectangles is so important that it creates a fake global gradient in the middle rectangle to make the local contrast stronger. The human eye doesn’t joke around with local contrast!
5) Color Differences
Ok, I’ve got one more for you. Credit goes to Henry LaBounta for showing me a variation of this one. On the upper-left I’ve got a grid of 9 colors, and a similar grid on the bottom-right. See any differences?
If you look closely you should see that the colors aren’t quite right. So which colors are lighter and which are darker? Which are more saturated and which are less saturated? Which ones are warmer and which ones are cooler? It’s pretty hard to tell. Now mouse over the second image. Now it is immediately clear how different those colors are. And telling which set is lighter/darker, saturated/desaturated, or warmer/cooler is a breeze.
Henry LaBounta’s point on this one was about color correction for CG artists. If you take 20 reference shots of concrete in different lighting conditions and try to match them for color by eye then you’ll never get it right. You’ll see some pros do a similar thing with a Color Checker. They do this so that if you want correct, neutral photos, you can just color correct the Color Checker and you’re done (instead of endless fiddling with sliders). The most common ones are from GretagMacbeth and X-Rite.
But there is a separate conclusion for HDR artists. If you see really bright and really dark areas in the real world then your eye will focus fudge the global difference between them but still focus on the local differences.
Conclusion: So what about HDR?
Hopefully I didn’t beat a dead horse too much here. You should now have a better idea of how your eyes lie to you. What does all this have to do with HDR Photography? Our eyes do a tremendous job of cheating strong lighting gradients. Here’s that LDR shot from the previous post (Why HDR?) using a standard digital camera.
The human visual system is truly incredible. Somehow our eye does this amazing thing where it cheats up the dark areas and cheats down the bright areas. There will be more discussion of that phenomena in later posts. Here’s the image showing from left to right, the original image, the image tweaked with NaturalHDR, and the image tweaked with a different HDR program to get a stronger, more processed look.
Maybe you think the middle one is great, or maybe you think it sucks. You could make a strong argument that the outside is still to bright and the cello case is still too dark. You’re free to tweak it differently for your shots. Also, I’m not going to claim that NaturalHDR is a revolutionary utopia of perfectness. But figuring out what that middle image should look like, be it NaturalHDR or some other option, is the most important problem in photography today.