PDA

View Full Version : Why does 'tilt-shifting' make photo subjects look like miniatures?


choie
02-06-2011, 05:08 PM
Note: I know this might be a Cafe Society thread. Sorry in advance if I guessed wrong by placing it here...

I've only just read about tilt-shifting photographic effects today -- yeah, I'm behind the times as usual -- and now I'm really curious about the whole thing.

A very dumbed-down explanation for those few as ignorant as myself (if such scary creatures there be!): Tilt-shifting is an effect used by photographers or mimicked in post-processing image-editing software to create a "miniature" effect on their subject. To me, as a layman, I'd explain it as using a very sharp focus on the subject while using very soft focus (or blurring?) on the background and foreground & other elements in the picture.

Of course, it's far more technically complex than that, and it has to do with changing the lens's rotation vs. the subject, yadda yadda. Here's the tech explanation on Wikipedia (http://en.wikipedia.org/wiki/Tilt-shift_photography). Truth is, I'm not so interested in the how as in the why.

But before we go on, and since a picture is worth a thousand words -- probably more in my case -- let's look at some cool examples:

(Credit where credit's due: the first four are from BoredPanda.com)

1. Are these Matchbox cars? Nope, real ones. (http://lh6.ggpht.com/_gKQKwLZ8XUs/TGw9AQ8YDiI/AAAAAAAADkY/Tw7LJhU0hxA/s800/Tilt-Shift-1.jpg)

2. Some kid's train set? Nuh-uh. (http://lh4.ggpht.com/_gKQKwLZ8XUs/TGw9tUYmEYI/AAAAAAAADlI/3CtDwpudo_U/s800/Tilt-Shift-13.jpg)

3. I would so want this toy Parliament set if it really existed! (http://lh4.ggpht.com/_gKQKwLZ8XUs/TGw93Viib7I/AAAAAAAADlc/LBzw2RaMN_E/s800/Tilt-Shift-18.jpg)

4. It even works with Van Gogh! (http://bp.uuuploads.com/van-gogh-tilt-shift/tilt-shift-van-gogh-the-harvest.jpg)

5. Here's a straightforward set of "before" and "after" photos (http://acidcow.com/pics/3471-tilt_shift_photographs_before_and_after_6_pics.html) from AcidCow.com.

6. And finally, 50 Beautiful Examples of Tilt-Shift Photography (http://www.smashingmagazine.com/2008/11/16/beautiful-examples-of-tilt-shift-photography/) from wonderful Smashing Magazine.

So my question is: I've read the explanations of how the artists/photographers achieved the effect. I barely understand 'em, but I kinda get it. But what I don't get is why it works. Why do our eyes perceive these objects as "mini" rather than simply very-sharply focused full-size objects?

(Geeze, you'd think after watching the eighty-billion hours of behind-the-scenes material on the LOTR trilogy I'd know all there is to know about how cameras fool the eye vis-a-vis miniatures... :D)

minor7flat5
02-06-2011, 05:19 PM
Boy, every time I come out from my cave where I live, I see something I never heard of before.

There are two things that are clearly happening here:

1) They use blurring effect above and below the center of the photo to imitate a very shallow depth of field. This makes the photo look just like an up-close portrait photo where the background is intentionally blurred by opening the aperture wide. It also looks very similar to macro shots of flowers, where everything in front of and beyond the photo is greatly blurred.

2) They punch up the saturation, making the colors look fake and toylike.

Chronos
02-06-2011, 05:29 PM
Ordinarily, when you take a picture of a life-sized parking lot or Parliament building or whatever, your distance from the scene is a much larger than the focal length of your lens, so everything in the scene is at effectively infinite distance, and so everything can be in focus at once. By contrast, when you're taking pictures of miniatures, your camera is usually very close to the scene, compared to the focal length, and there's a big difference between something being (say) 2f away and 3f away. So you can only focus on a small part of the scene, and things in front of or behind that are out of focus.

tellyworth
02-06-2011, 05:35 PM
Tilt-shift doesn't make subjects look like miniatures.

It makes things look like photographs of miniatures.

Macro photography - extreme close-ups of small things - has a very narrow depth of field, and thus produces selective focus much like you see in those tilt-shift images. Since the only time you'd otherwise see those visual cues (extremely selective focus with a mid-to-wide angle field of view) is in photographs of miniature models, your brain wants to interpret them as miniatures.

Your eye does the same thing, when you move close to a small object. In a photograph we don't have binocular vision to judge size and distance, so we're limited to cues like depth of field.

choie
02-06-2011, 06:30 PM
Thanks for the responses, all.

So ... let me see if I have this right. You guys are saying that in 'regular' photos of, say, the Big Ben / Houses of Parliament, usually everything would be in focus (except, presumably, stuff that's farther away, but I'd assume the transition from sharp focus to blurry would be fairly gradual). Our eyes are so used to this form of image composition that this is what we expect.

In contrast, when someone's shooting a picture of something very close, the photographer has to use a 'narrow depth of field', which means (for dummies like me) super-sharp close-up that makes everything other than the immediate object blurry.

So.... the effect of the tilt-shifted pics is that when I'm looking at Big Ben being super-sharp with everything else blurry, my brain is telling me, "this object must be very close to the camera." And if Big Ben looks only three inches tall while being "close" to the camera, this must mean that this is not the full-sized Big Ben, but rather a miniature version.

Am I close?

Lumpy
02-06-2011, 08:16 PM
There's another effect in play besides depth of focus. When photographing a small object close up, the lens is wide enough that it's capturing light from a considerable angle on either side of the direct line of sight between the center of the lens and the object being photographed. IOW, it's as if the final image was a composite view of the object taken from widely differing angles. The effect can be duplicated for large objects by adjusting the aperture/focal length ratio, or multiple exposures from shifted angles.

BigT
02-06-2011, 10:24 PM
The tilting itself does also help, too, because, in real life, you are more likely to encounter a toy in the shifted position than the object in question. The only way to get anywhere close is to tilt your head, and your senses of balance and proprioception compensate so the object does not appear tilted.

HubZilla
02-06-2011, 10:30 PM
I made one (https://picasaweb.google.com/grantjap/GoGoGodzilla#5550493930822111298) from my trip to Tokyo.

Projammer
02-07-2011, 01:01 AM
Another thing that I notice is that even in shots where there should be action, there's no motion blurring of the main subjects. That makes them look posed and staged.

Bryan Ekers
02-07-2011, 01:19 AM
I figure if you're looking at distant buildings/cars/people, the difference between focusing on an object 500 meters away vs. one that's 520 meters away is trivial. However, when examining something smaller, focusing on an object 8 inches away and then trying to focus on one 16 inches away requires a significance adjustment, even though the absolute difference is much smaller - 8 inches vs. 20 meters.

Popular Mechanics (http://www.popularmechanics.com/technology/how-to/tips/how-to-take-fake-miniature-photos) article (well, blurb, really) on techniques and how/why they work.

Dr. Strangelove
02-07-2011, 03:38 AM
However, when examining something smaller, focusing on an object 8 inches away and then trying to focus on one 16 inches away requires a significance adjustment, even though the absolute difference is much smaller - 8 inches vs. 20 meters.

It's worse than that, actually. Even in *relative* terms, focus usually drops off more rapidly for close objects than distant ones.

For a given lens configuration (aperture, etc.) there is a measurement called hyperfocal distance. Roughly speaking, this is the distance away from the camera such that when focused on, *all* objects farther away than that point will also be in focus.

For typical configurations, this distance is on the order of a few meters. And so for typical outdoor shots, where all the objects are farther away than this, we expect that everything will be in focus. Even the moon and stars remain in focus, despite being essentially infinitely farther away than anything in the foreground.

The opposite is true of miniature/macro photography. Because the distances involved are well within the hyperfocal range, the focus drops off quickly outside of the focal plane--there is a narrow depth of field. A difference of 10% in depth may mean the difference between entirely in focus vs. entirely out of focus.

Dr. Strangelove
02-07-2011, 03:45 AM
Another thing that I notice is that even in shots where there should be action, there's no motion blurring of the main subjects. That makes them look posed and staged.

This is even more evident in video. A related trick is to decrease the frame rate to the 5-10 fps range. Again, the idea is to make it look like cheap stop-motion animation.

Francis Vaughan
02-07-2011, 05:23 AM
A few minor points.

Tilt and shift are two different things. Shift is movement of the lens such that the optical axis remains normal to (i.e. at 90 degrees to) the film plane.
Tilt is the movement of the lens such that the optical axis is tilted away from the normal to the film plane. Shift is used to alter the perspective of the result. Architectural work is the common use, where it is possible to control the perspecive of the resulting image so that vertical lines in the object remain vertical in the final picture. Digital processing of images has left such control pretty much a thing of the past.

Tilt however means that you can control the plane of focus. Usually tilting is used to allow the photographer to place objects at different distances all in focus in the image. Clearly this is heavily constrained, the optical axis can only be tilted in one direction. So, at best, it is possible to tilt the lens so that a set of objects arranged on a line can be placed all in focus. Or it is possible to create a couple of zones of in-focus - for instance the foreground across the lower part of the picture and a middle distance component - and the rest of the picture out of focus. Usually such control is available only on view cameras, or some larger medium format cameras. So the use cases are restricted to studio shots and landscapes.

Clearly the opposite is possible - which is where the OP comes in. It is possible to deliberately tilt the lens so that only a small part of the object in the picture is in focus - much less than would be expected. And to place it so that it crosses the object in the picture at a range of distances.

Depth of field issues come about because when we take pictures of small things, we don't use a correspondingly small camera. If you had a 1/72 scale model, and could take pictures of it with a 1/72 scale camera, with 1/72 scale film grain, or 1/72 scale pixels, the apparent depth of field would stay the same. (This ignores diffraction effects - perhaps we just scale wavelength too.) The critical problem is that we can't scale the resolution of the film we use. We could build a tiny camera - with everything 1/72 sized - but since the film remains 1:1 in terms of technology the image would have 1/72 the resolution, and the picture would be useless. Depth of field is a somewhat arbitrary construct that is derived from the film resolution. Lens systems can only focus one distance correctly at one time. Anything else is blurred. However the amount of blur changes. We can geometrically define the size of the circle of confusion that appears on the film (or image device) for any point that is at the wrong distance. The circle gets bigger the further from the right distance you are. If you have film or an imaging device of a given resolution, clearly if the circle of confusion is smaller than the smallest size you can usefully resolve, the source effectively remains in focus. So a range of distances are in focus. Smaller appatures result in smaller circles of confusion, and so the range of distances (i.e. the depth) of objects in the field of view that appear to be usefully in focus increases. The whole analysis scales up, but because our film resolution does not change, we see effects where small focal lengths have greater depth of field - because they cover less film area from the same object, and longer focal lengths have less depth of field, because they use more film area. Larger format cameras appear to have worse depth of field than smaller cameras - simply because they offer greater resolution.

Machine Elf
02-07-2011, 08:05 AM
This is even more evident in video. A related trick is to decrease the frame rate to the 5-10 fps range. Again, the idea is to make it look like cheap stop-motion animation.

Here's one of the best examples I've seen. (http://www.vimeo.com/3156959)

Mangetout
02-07-2011, 08:22 AM
Shift is used to alter the perspective of the result.I think the perspective adjustment in these photos is a significant part of their reason for looking like models - yet this seems to be almost ignored in favour of depth of field.

When we look out of a window, we're looking at a big scene - perspective is a really significant factor - the ratio between the depth of the scene vs our distance from the nearest object is comparatively high- whereas with a model, it's all there right in front of us - the ratio between the depth of the scene and the distance from the nearest part is low - making everything appear more or less the same size, regardless of its place in the picture.

Chronos
02-07-2011, 03:32 PM
Quoth choie:
So ... let me see if I have this right. You guys are saying that in 'regular' photos of, say, the Big Ben / Houses of Parliament, usually everything would be in focus (except, presumably, stuff that's farther away, but I'd assume the transition from sharp focus to blurry would be fairly gradual). Our eyes are so used to this form of image composition that this is what we expect.

In contrast, when someone's shooting a picture of something very close, the photographer has to use a 'narrow depth of field', which means (for dummies like me) super-sharp close-up that makes everything other than the immediate object blurry. Two mistakes, here. First, in a distant scene, everything that's far away will be in focus, no matter how far it is, since it's all effectively at infinite distance, and you can't be further away than infinity. Second, when you're focused on something close, not only will things further away be blurry, but things that are even closer than your subject will also be blurry.

pulykamell
02-07-2011, 10:16 PM
I think the perspective adjustment in these photos is a significant part of their reason for looking like models - yet this seems to be almost ignored in favour of depth of field.

The effect you're seeing in those photos is pretty much all tilt, not shift. I've worked with tilt-shift lenses before (mostly for architecture and keeping verticals and horizontals properly straight) and when I played around with doing some of the "miniature-looking" effects, it was a function of the tilt of the lens and having the plane of focus not being parallel to the plane of the film (or sensor). Shift doesn't really contribute anything at all to the "miniature" feeling of the photos. Shift just sort of allows you to move your horizon (or vertical center) up and down without actually having to tilt your camera up or to the sides (which then ruins your true verticals or horizontals.)

MrFloppy
02-07-2011, 10:23 PM
I just bought a Nikon P7000 and it has a 'miniature' preset which apparently does this effect in the camera. I'll have to try it now.

Colophon
02-08-2011, 09:57 AM
But before we go on, and since a picture is worth a thousand words -- probably more in my case -- let's look at some cool examples:

(Credit where credit's due: the first four are from BoredPanda.com)

1. Are these Matchbox cars? Nope, real ones. (http://lh6.ggpht.com/_gKQKwLZ8XUs/TGw9AQ8YDiI/AAAAAAAADkY/Tw7LJhU0hxA/s800/Tilt-Shift-1.jpg)

2. Some kid's train set? Nuh-uh. (http://lh4.ggpht.com/_gKQKwLZ8XUs/TGw9tUYmEYI/AAAAAAAADlI/3CtDwpudo_U/s800/Tilt-Shift-13.jpg)

3. I would so want this toy Parliament set if it really existed! (http://lh4.ggpht.com/_gKQKwLZ8XUs/TGw93Viib7I/AAAAAAAADlc/LBzw2RaMN_E/s800/Tilt-Shift-18.jpg)

4. It even works with Van Gogh! (http://bp.uuuploads.com/van-gogh-tilt-shift/tilt-shift-van-gogh-the-harvest.jpg)

Neat examples. I've wondered exactly the same thing before - why simply narrowing the depth of field should have such a dramatic effect. On the cars and train, it's almost as if you can see the cheap moulding of a child's toy, and yet they are real. Perception is weird. :)

Edit: Looking at the "before and after" set in this link (http://acidcow.com/pics/3471-tilt_shift_photographs_before_and_after_6_pics.html), it seems that boosting the colour saturation is a big part of it, too.

pulykamell
02-08-2011, 11:48 AM
Edit: Looking at the "before and after" set in this link (http://acidcow.com/pics/3471-tilt_shift_photographs_before_and_after_6_pics.html), it seems that boosting the colour saturation is a big part of it, too.

Wow, that's some horrific color over-saturation going on there. It's not necessary for the "tilt-shift" miniature effect (the effect does work in black-and-white and with more natural color palettes--for example, the Vincent Laforet work in link #6 isn't what I would call over-saturated), but I do agree that exaggerating the colors does help push the toy/miniature illusion along.

Quercus
02-08-2011, 12:27 PM
I think another part of the illusion is having an abnormally low level of detail in the picture, so it seems like a toy or model to our eyes.

On a technical level, making things out of focus gets this automatically over most of the picture. The oversaturation also tends to make small details harder to see and notice. I haven't looked at any full-sized versions of these, but I bet the illusion is less in a higher-resolution version (unless of course, the photographer intentionally lowered the resolution...)

And then the most successful of the shots are ones that, through accident or careful shot selection, don't have much detail in the small in-focus central portion.

curious11
02-08-2011, 12:58 PM
Based on the above descriptions of the technique I theorized that by isolating a portion of the "in focus" section of the image that it would look normal. And it does. I placed one hand over the other hand to leave a small opening between my thumbs and forfingers and looked at the little girl running in the first picture and she doesn't look like a toy.

scr4
02-08-2011, 01:41 PM
Edit: Looking at the "before and after" set in this link (http://acidcow.com/pics/3471-tilt_shift_photographs_before_and_after_6_pics.html), it seems that boosting the colour saturation is a big part of it, too.

And on these examples, it looks like there's no change in perspective (i.e. no distortion to modify perspective). It's done mainly (or totally) by narrowing the depth of field and color saturation.


By the way, are there examples of doing this in reverse to make models look realistic? I imagine some of these techniques mentioned here are routinely used for special effects.

naita
02-08-2011, 04:55 PM
I'm still not convinced none of those pictures are models. Look at the "grass" in the second example with the train. That's exactly the texture of the green granulate people use to make "grass" in models. Why would real grass look like that after some basic transformations?

pulykamell
02-08-2011, 05:40 PM
And on these examples, it looks like there's no change in perspective (i.e. no distortion to modify perspective). It's done mainly (or totally) by narrowing the depth of field and color saturation.


Yes, those examples are fake tilt effects (horribly overdone, IMHO) in Photoshop. They are not true tilt-shift. They are trying to replicate the feel of tilt by narrowing the depth of field along an oblique axis, but don't achieve exactly the same result. Still, it is that depth of field that mostly contributes to the model-like feel of the image.

Naita--that photo looks like classic tilt-shift to me. Note you can even see the people in the train. Look at the first three photos here (http://www.smashingmagazine.com/2008/11/16/beautiful-examples-of-tilt-shift-photography/) (the last link in the OP). Do any of those look fake to you? You're going to have to take my word on it, but that photographer is a very good friend of mine, and I've seen the whole sequence of shots from those series, and I assure you, it's not fake.

John DiFool
02-08-2011, 05:47 PM
I've got another aspect to discuss: the shadows seem to look wonky in some of these. The kinds of shadows you get outdoors on full-sized objects aren't going to look the same as the shadows on miniatures indoors, right? Someone is playing games with that too I'd say.

pulykamell
02-08-2011, 05:59 PM
I've got another aspect to discuss: the shadows seem to look wonky in some of these. The kinds of shadows you get outdoors on full-sized objects aren't going to look the same as the shadows on miniatures indoors, right? Someone is playing games with that too I'd say.

Which ones?

Dr. Strangelove
02-08-2011, 06:21 PM
By the way, are there examples of doing this in reverse to make models look realistic? I imagine some of these techniques mentioned here are routinely used for special effects.

It's difficult. While blurring is easy, deblurring is not. It's just not really possible to take an image with a narrow depth of field and expand it.

However, there is now software that can take a stack of images of the same object, with with different focal planes, and merge them. See here:
http://en.wikipedia.org/wiki/Focus_stacking

In principle, one could use this technique to make miniatures look less miniature. However, I haven't yet seen a good demonstration of this.

There are a few diorama-makers out there that do an amazing job, though, like this guy:
http://www.flickr.com/photos/24796741@N05/sets/72157604247242338/with/2346008881/

I'm not sure how, but he keeps pretty much everything in focus (he must use a tiny aperture, accepting a certain amount of diffraction blurring). The black-and-white ones are more convincing, to my eyes, probably due to the saturation effect mentioned.

The Second Stone
02-09-2011, 02:24 AM
Tilt and shift when used put the plane that is focused different that just parallel and flat on an opposite wall. To see the tilt effect hold a flashlight beam against a wall and tilt it. It will be an ellipse. This is used to correct distortion and make architectural photos have the buildings square. Usually. When these effects are used when the lens is wide open (gathering the most light with a lower f stop) there is very little of the subject that is in sharp focus. When only a center is in sharp focus and all the edges are blurry, we interpret that as a small object because it usually is a person or smaller for that kind of picture. Close the f stop way down to 64, like Ansel Adams did with his tilt/shift photos and everything is sharp. You can give mountains shapes that they don't have and they still look real if you don't overdo it. The mind interprets really sharp photos as real life large scenes. The mind also fills in colors in black and white photography.

When you look at an art photo you are no more seeing the real thing than looking at a musical score is listening to the real thing. It is an interpretation. And in this instance the camera has been used in such a way as to interpret the scenes for you as miniature scenes.

pulykamell
02-09-2011, 02:58 AM
Tilt and shift when used put the plane that is focused different that just parallel and flat on an opposite wall. To see the tilt effect hold a flashlight beam against a wall and tilt it. It will be an ellipse. This is used to correct distortion and make architectural photos have the buildings square.

Shift is used for perspective control, not tilt. Tilt is used to change the plane of focus. (Unless I'm misunderstanding you.) Shift keeps the plane of focus the same. Tilt will not correct converging verticals and keep buildings square.

Colophon
02-09-2011, 09:09 AM
I'm still a bit puzzled by why shift lenses allow perspective correction.

Why does shifting the lens up an inch or so (http://en.wikipedia.org/wiki/File:New-35mm-left.jpg) allow you to get the top of a tall building in shot any more than simply leaving the lens in position and raising the camera by an inch?

Francis Vaughan
02-09-2011, 10:03 AM
The easy way to think about shifts is this. In order to have vertical lines in the subject appear as vertical lines on the film plane, the film plane must be parallel to the vertical lines - i.e. the film plane must itself be vertical. If you tilt the camera to allow it to frame the top of the building the film plane is no longer vertical.

So what you need is a much wider angle lens, or for the same focal length lens, a much bigger film plane, so that when you keep the camera horizontal (and thus the film plane vertical) you include a wider scene, and thus include the top of the building. Now you don't have a bigger film camera. But what you can do, is keeping the lens stationary, you can move what area of film you do have down relative to the lens, where it will intersect the image of the top of the building you desire.

You will note that this implies that the lens is actually capable of covering a much larger area of film than a conventional lens for that camera format needs to do. Indeed this is true, shift lenses are significantly more expensive, and contain a lot more glass, because they are in effect designed to cover a much larger image area than the format of the camera would suggest.

So to summarise, a shift lens is allowing the film to sample a sub-section of the image that the lens forms. The easy way to use a shift lens is to keep the camera dead level, and then think, not in terms of shifting the lens, but of shifting the film behind the lens. Of course, except for very close up work, whether you shift the camera relative to the lens or the lens relative the the camera makes no difference.

Colophon
02-09-2011, 10:27 AM
Thanks Francis, that makes sense.

The Second Stone
02-09-2011, 12:47 PM
Shift is used for perspective control, not tilt. Tilt is used to change the plane of focus. (Unless I'm misunderstanding you.) Shift keeps the plane of focus the same. Tilt will not correct converging verticals and keep buildings square.

I stand corrected. Never used one myself.

choie
02-09-2011, 12:58 PM
Just wanted to thank everyone for the answers so far, especially for the additional photo & film examples. (The video linked by Machine Elf (http://www.vimeo.com/3156959) is especially amazing.) Lots of the tech stuff is going over my head, but I'm trying to take it all in. :)