Level or not?

[Majeed Badizadegan]

The guy who took the photo says he uses the Canon 5D-III's onboard level all the time... so, let's look at another photo he took, and you can easily SEE the effect I spoke about in post #3.

http://www.flickr.com/photos/gordie_broon/8658179032/in/photostream/


In this photo linked to above, we can see the effect that the actual contours of the land surrounding a lake can have on a landscape photo's "horizon line".


So…where is the horizon "line" to be found? Shoreline geography can influence where the water/land junction actually is located in a photo. l

Where land meets water along the shoreline of miles-long, mountainous headlands will to vary from geometric perfection, and it will instead, reflect the real world's imperfection.

This case is pretty different. Everything is much closer and finding a "horizon" after the fact would be nearly impossible. So it's probably not a great example in that regard. But it's a good example to show contours in shore-lines and the imprefections of the world that make finding the correct horizon in post impossible in some circumstances. It's best practice to get it right in camera to avoid these situations, but we don't always have this luxury (especially when shooting handheld).

Now the question becomes, is that the case with the photo in the OP? No. There is a clear, straight edge at the top of the water. From where the water starts in middle-left frame to where it ends in middle-right frame is perfectly straight. This is critical, because now we know contoured shorelines aren't messing with our eyes. We know the water is level and straight. It is even across the frame. And thus it should be the primary reference for the horizon line, because it is the horizon!

 

[Majeed Badizadegan]

.

Are you not adequately entertained?

 

[Majeed Badizadegan]

I think Derrel's analysis is correct.

Part of what is happening is that the water horizon line looks too crisp, to the extent that I think he may have overcooked it with some sort of processing.

When it looks more or less "equally crisp" all the way along, we get no sense of increased distance, and it feels wrong. There's miles and miles of air between us and that farthest end of the visible water, and while the water horizon is slightly softened way out there, it's not softened ENOUGH. The entire water area feels fake and wrong to me, although it probably isn't.

I'm not sure what he did in post, if anything, to accomplish this effect, but I quite dislike it.

I hope for the conversation not to stray into anything too critical of the image, I don't want the author to feel like he's receiving unsolicited critique on his photo. I'm just mainly interested in a theoretical discussion here, whether my thinking on the leveling in general is correct or not.

The artist and I had an exchange on this topic, and it left me a bit puzzled that he'd put so much faith in electronic leveling as to completely disregard what actually looks right (to me).

 

[TCampbell]

I measured it (I have a transparent ruler that I sometimes slap onto my monitor).

It is level. (if it's not, you'd have to get into tiny fractions of a millimeter).

I've seen images where for whatever reason, the "brain" thinks something is off-level and I can see that tendency here. Actually what I noticed is that it almost looked like the water surface was curved... but since it does not extend all the way to the left, I can see how it would appear to be un-level. Measurements, however, show that it is level and whatever I'm seeing appears to be an optical illusion.

 

[amolitor]

Again, the water surface appears curved because the usual cues that would give us the correct picture, namely an increased sense of distance due to mist/atmospheric haze softening the water-horizon, are either not present or extremely weakened.

The water-horizon appears to us at first glace to be equidistance across the frame.

Thus, the visual curvature created by the approaching shoreline appears to NOT be due to an approaching shoreline, but by some other unsettling effect.

 

[Studio7Four]

There is a clear, straight edge at the top of the water. From where the water starts in middle-left frame to where it ends in middle-right frame is perfectly straight. This is critical, because now we know contoured shorelines aren't messing with our eyes. We know the water is level and straight. It is even across the frame. And thus it should be the primary reference for the horizon line, because it is the horizon!

Admittedly I do not have my full arrangement of software tools here at the moment, but your assertion that we are dealing with a perfectly straight water edge appears false. Draw a line from the tip of the land jutting in from the left to the point where the water leaves the frame on the right. There is a slight - almost imperceptible, but there - arc of water remaining above this line. Derrel is correct in that what you are seeing as a horizon is the contoured boundary between the water and the land. To be strictly correct, a horizon is the boundary between the earth and the sky, and only for a very small portion of this shot does the water reach to the sky (if it does at all, the distant land is so faded it's tough for me to tell). Had this been a shot of an ocean in which the water does reach to the sky you'd have a point.

Here's a thought experiment which might help clarify the issue. Image this same shot taken from 100 feet higher, or better yet 1000'. From that perspective you would expect to see a defined contour where the water meets the far land. What happens as the perspective approaches this, shot from near the water's surface, is that this contour is compressed into fewer pixels vertically. The relative distances are compressing that portion of the far shoreline to such a narrow band of pixels that your eye is being tricked into wanting it to be horizontal, but that's simply not the case.

Here's another thought experiment. Imagine the tree were replaced by a tripod with a plumb bob hanging from it, defining a truly vertical line in the frame. The shoreline will still be not-quite horizontal...would you tilt the image to define the relatively longer shoreline as horizontal at the sake of the plumb line?

 

[TCampbell]

There is a clear, straight edge at the top of the water. From where the water starts in middle-left frame to where it ends in middle-right frame is perfectly straight. This is critical, because now we know contoured shorelines aren't messing with our eyes. We know the water is level and straight. It is even across the frame. And thus it should be the primary reference for the horizon line, because it is the horizon!

Admittedly I do not have my full arrangement of software tools here at the moment, but your assertion that we are dealing with a perfectly straight water edge appears false. Draw a line from the tip of the land jutting in from the left to the point where the water leaves the frame on the right. There is a slight - almost imperceptible, but there - arc of water remaining above this line. Derrel is correct in that what you are seeing as a horizon is the contoured boundary between the water and the land. To be strictly correct, a horizon is the boundary between the earth and the sky, and only for a very small portion of this shot does the water reach to the sky (if it does at all, the distant land is so faded it's tough for me to tell). Had this been a shot of an ocean in which the water does reach to the sky you'd have a point.

Here's a thought experiment which might help clarify the issue. Image this same shot taken from 100 feet higher, or better yet 1000'. From that perspective you would expect to see a defined contour where the water meets the far land. What happens as the perspective approaches this, shot from near the water's surface, is that this contour is compressed into fewer pixels vertically. The relative distances are compressing that portion of the far shoreline to such a narrow band of pixels that your eye is being tricked into wanting it to be horizontal, but that's simply not the case.

Here's another thought experiment. Imagine the tree were replaced by a tripod with a plumb bob hanging from it, defining a truly vertical line in the frame. The shoreline will still be not-quite horizontal...would you tilt the image to define the relatively longer shoreline as horizontal at the sake of the plumb line?

I've sighted across the image from the side using a straight-edge. I'm convinced that the appearance of a curve is only an illusion... the line actually is straight and the water actually is level.

The water line does appear as a separation between the water and the sky (not the water and shore). The Earth curves... you can't actually see the distant shore as the camera is too low so all we see is the surface of the water.

Just for fun... here's the math:

The Earth's circumference is 24,901 miles. That means a distance of 1 mile has a curvature of .014º. The tan(.014) X 5280 = 1.33'. That means that unless the camera is at least 1.33' above the surface then what you'll see is the horizon line rather than the distant shore with a distance of just one mile. If the camera height is 4' above the surface then the farthest distant at which you can still see the shore would be 3 miles away... beyond that all you see is the horizon.

I don't know the true distances here, but that looks like more than 3 miles away.

 

[amolitor]

The horizon is, as a rule of thumb, something like 1.2 times the square root of the height of the eye in feet, in miles.

So if my eye is 5 feet above the water surface, the square root of 5 is about 2.24, and the horizon is 1.2 x 2.24 = 2.7 miles away. Roughly. It'll do for approximations.

What's interesting about this to me is that you can do this with simple trigonometry -- the egyptians could do this 3000 years ago -- and you can deduce that the world is round and roughly how big it is quite quickly and easily if you happen to be on a large body of water. Columbus, in particular, had tons of time in which he could have done this and deduced that China was actually about 10,000 miles further away than he thought it was (or at any rate told people it was so they'd fund his expedition).

 

[Majeed Badizadegan]

So what is mathematically correct doesn't necessarily look the best in a photo, is where I'm arriving.

I do believe the straightened version that I posted looks more appealing and all but eliminates the "optical illusion".

I did nothing but draw a straight edge along the straight body of water, had cs5 rotate the canvas and do the straightening. Drew a horizontal and vertical guide at 50% and screen capped for sharing here. No lens profile adjustments or anything else for that matter.

 

[runnah]

Your eye don't know no maths, do what makes it happy.

 

[Ilovemycam]

It looks crooked to me, the horizon line is about 1/16" lower from the top on the right side than the center/left, which for a photo about the size of a 5x7" is noticeable.

It's the type of photo that I can't help but start wondering right away what they did to it - and not in a good way because I find I'm noticing the processing more than the image.

And why people set their Flickr pages so others have access to them is beyond me; he's lucky that you're just editing it to show it as an example but it leaves it open for anyone to help themselves.

Hey, pix are for sharing and all our enjoyment

 

[Ilovemycam]

I think Derrel's analysis is correct.

Part of what is happening is that the water horizon line looks too crisp, to the extent that I think he may have overcooked it with some sort of processing.

When it looks more or less "equally crisp" all the way along, we get no sense of increased distance, and it feels wrong. There's miles and miles of air between us and that farthest end of the visible water, and while the water horizon is slightly softened way out there, it's not softened ENOUGH. The entire water area feels fake and wrong to me, although it probably isn't.

I'm not sure what he did in post, if anything, to accomplish this effect, but I quite dislike it.

I hope for the conversation not to stray into anything too critical of the image, I don't want the author to feel like he's receiving unsolicited critique on his photo. I'm just mainly interested in a theoretical discussion here, whether my thinking on the leveling in general is correct or not.

The artist and I had an exchange on this topic, and it left me a bit puzzled that he'd put so much faith in electronic leveling as to completely disregard what actually looks right (to me).

Screw elctronic leveling. Just do it on LR

The pix reminds me of a super wide shot where the horizon has a little bow to it.

 

[Buckster]

It looks crooked to me, the horizon line is about 1/16" lower from the top on the right side than the center/left, which for a photo about the size of a 5x7" is noticeable.

It's the type of photo that I can't help but start wondering right away what they did to it - and not in a good way because I find I'm noticing the processing more than the image.

And why people set their Flickr pages so others have access to them is beyond me; he's lucky that you're just editing it to show it as an example but it leaves it open for anyone to help themselves.

Hey, pix are for sharing and all our enjoyment

No, they're not; Not unless the copyright holder says they are.

It's really pretty simple: Photos are copyrighted. That means nobody has the right to copy unless the copyright holder has given express permission to do so. And he didn't allow his photos to be easily dragged off and onto the desktop from Flickr, by the way. It was likely screen-grabbed, not that it matters in the least - a copyright violation is illegal either way, whether it was easy or difficult to copy the image without the legal right to do so. I hope rotanimod got the copyright holder's permission first.

 

[pgriz]

If it ain't level, have another glass of wine. Because not all levels need to be horizontal.

 

[Tiller]

If it ain't level, have another glass of wine.

Well, there it is!