Basic Tilt-Shift Simplified

[Don Burkett]

Bob, thanks for the invite, something I'd be interested in but really am too new to the tilt shift concept to feel worthy of putting together a tutorial. My mission though is to learn the principals well enough that I can simplify them for others. Whether that takes a month or longer will depend in how much time I can get to play with it. Hopefully, we'll get enough contributions in this thread I can gather material and accelerate the learning process. []


For example, Johnathan remarked that focus on distance tilt to closest will sometimes require multiple cycles between the objects to get it perfect. He's right of course and for certain scenes it requires more fine tuning than others. But, I haven't figured out the rule yet. In other words does it require more tuning if the nearest object is really close or moderately close. Does the height of the closest object make the difference or it it the height of the distant object. Maybe it's just the difference in heights or the the distance between them. These are the kinds of things I need to figure out.

[Don Burkett]

Ok, so I thought I'd take a whack at documenting last night's experimentation.


Set up: Canon 7d; TS-e 24mm L II. ISO 200
<ul class="unIndentedList"][*]
Lens is 11" in front of the table
approximate 2" above table height.[*]
The 'Rook' is 16" from the front of the
lens, the rear pawn is 26" from the lens.[*]
The pieces are approximately 1"
circumference at the base and spaced 1" apart.[*]
The raw files were converted as jpegs @ 800x in
DPP using the Standard Picture Style, Saturation -1, Sharpening at 3, Noise
Reduction set to 0, White Balance set to tungsten.[*]
A diffuser was hand held on each shot, which
resulted in some exposure differences.
These were mostly adjusted for, but can still be detected. I'm calling this out because, tilting the lens will affect meter readings,
but that is not being tested for in this exercise.[*]
Note that a couple of the chess pieces are
crooked. This is not lens distortion.[*]
Also note,
this shifting and focusing seems a lot easier at landscape distances than when
objects are closer together.[*]
Any observations that I think you will find
useful are in italics.[/list]





First shot is focused on the rook @ f/3.5 no tilt or
shift. Observe the depth of field. At
the presented size, sharpness seems to be maintained through the first 2 chess
pieces and break up at the third. At
100% resolution, the sharpness breaks up at the 2nd piece.


[img]/resized-image.ashx/__size/800x0/__key/CommunityServer-Discussions-Components-Files/12/2475.Image001.JPG[/img]


Second shot is focused on the rear pawn @ f/3.5 with the
lens tilted to the left (horizonal) just under 2&deg;. This brings all the chess pieces into
acceptable focus at the presented size.
At 100% resolution the rook is slightly out of focus. So a tad bit more fine tuning could have been
done. But, the desired affect was
achieved as plane of focused was tilted at enough of an angle to bring all the
chess pieces into focus. In the future though I think I will use my
Hoodman Hoodloupe for 3" lcd screens to ensure my focus is better adjusted
before shooting.


[img]/resized-image.ashx/__size/800x0/__key/CommunityServer-Discussions-Components-Files/12/4810.Image002.JPG[/img]


Comparing the 2 shots, there are a couple of other useful
observations to keep in mind.

1. Tilting the lens to the left,
   effectively moved the image frame 3/8" to the right.
2. The DOF in front of the ruler was
   dramatically altered (thinner) in the tilted image.

Note to self: Tilting
the lens, affects the DOF but don't have enough understanding yet to provide
any simple guidelines. I do believe it
creates a wedge of sorts but this
is one aspect I need to study. Was
hoping that using the blanket would give me enough texture to track the DOF
patterns. To some degree that worked, but since the DOF will logically run in
the direction of the angle of the lens a different set up is needed. Especially when the lens is tilted at a
vertical angle vs. a horizontal angle.


Third Shot, I wanted to compare the difference of using
f/3.5 vs. something like a f/7.1 (ok, exactly f/7.1). Clearly if you compare Image 3 to Image 2 the chess pieces look crisper, the edge of the
ruler is brought into focus and clearly there's more depth of field.


[img]/resized-image.ashx/__size/800x0/__key/CommunityServer-Discussions-Components-Files/12/0753.Image003.JPG[/img]





Fourth Shot;for fun, to better visual the difference between the two shots, I put them into layers and set the layer type todifference. The "difference" layer blend is a subtraction based on color, not sharpness soit's not a literal subtraction of
pixels that are the same. Further I know changing the aperture and the
varying exposure create differences of their own, However, working on a general theory, that where the image is sharper, there is a difference in pixel color . The reality is, even if the theory is right, you can't tell using this method, which image is contributing to the difference. But it made for a fun image.


[img]/resized-image.ashx/__size/800x0/__key/CommunityServer-Discussions-Components-Files/12/1715.Image004.jpg[/img]





Fifth Shot; Creating a negative of Image 4 gives a visual of
the differences between f/3.5 and f/7.1 and not withstanding the caveats I
mentioned abovegenerally I think you can interpret where increased aperture contributes to the image especially when you reference back to the actual images..


[img]/resized-image.ashx/__size/800x0/__key/CommunityServer-Discussions-Components-Files/12/1033.Difference-02.jpg[/img]


If this is something that is helpful, I will continue to post as I experiment. Thanks for looking.

[Jonathan Huyer]

Don -


That's an interesting test --- thanks for organizing and sharing that. You made a comment that tilting the lens affects exposure --- this is true (as it is for shifting), and can be easily corrected by simply shooting in live view mode. Also you are correct that the DOF is wedge-shaped when you tilt. See this article for an illustration of this effect:


http://www.cambridgeincolour.com/tutorials/tilt-shift-lenses2.htm


Jonathan

[Don Burkett]

Thanks Jonathan, Cambridge in Color is a great site, should have gone there already. Still, I enjoy the process of discovery. Think I