Tilt-shift query
  • User avatar
    Plays with Light
    Reporter
    Reporter
    Posts: 1354
    Joined: Sat Aug 10, 2013 10:34 pm
    Location: Lost In Thought
    Contact:

    Tilt-shift query

    by Plays with Light » Tue Jan 21, 2014 9:36 am

    I am a man on a mission to play this photography lark on a really tight budget, so I went and got a Samyang tilt-shift lens. It was a toss up between that and the original Canon TS-E 24mm, the independent tilt and shift of the Samyang and the extended warranty won me over. Image quality wise, they are on par with each other, the Samyang has a little more distortion, but PTLens cleans that up in post-production, no problems. If you are a bit rough with your gear, do not have excellent fine motor skills or have stubby fingers, it's not for you! I suffer from none of that, so this baby should see me through the next few years easily.

    Anyways, here are my first attempts from yesterday, while the twin boys attended a teenage science day at the local university, I got to grips learning how to use this thing. All captured at f/8. I did have a play with using f/3.5 to achieve a plane of focus along the ground and along a wall, but I really need to be tethered to my laptop, to achieve good focus, using the rear screen in liveview is not good enough for that sort of thing, for me.

    All of these are taken by shifting the lens to its maximum left or right, or up and down. So they are all three frame composites. I really wanted to push the lens and see the worst it had, which is written about with tilt-shifts so much and that you should never fully shift to 12mm, instead staying back a few millimetres to retain good imagery.

    ********
    In the second image, you will notice it makes the panels of glass across the rear of the building get progressively wider. WTF? It hasn't done this anywhere else when I took three images in landscape orientation and shifted them left and right to the maximum. Is this some bizarre thing that occurs with certain curves of buildings and not being on an exactly square or level basis with the sensor and the wall? Something else? The first and second frames for this panorama are fine and correct in proportion, the third, shifted fully to the right, has gone all weird though. None of the other captures from yesterday exhibited this anomaly...
    ********

    None of these images are anything exciting, I am just learning to use the beast.

    None of these images have any image correction applied on import or when combined in Photoshop, I did them as image repositions, with no geometric corrections in the panoramic module of photoshop. I really wanted to see how bad this lens could be.

    I did apply some other images in PTLens and used the 10 uses of its sample, to see how it's corrections went. It is well worth the $25 to buy and use on import, if you are a pixel peeper and wanting true lines like me!

    Thanks in advance for any assistance that any one can offer about the stretching of that second image. it really has me baffled. I am about to go take a whole series of captures of walls and analyse/set up my lens in PTLens for corrections to its distortion.

    Image
    SILC 01 by playswithlight, on Flickr

    Image
    SILC 02 by playswithlight, on Flickr

    Image
    SILC 03 by playswithlight, on Flickr

    Image
    Walkway by playswithlight, on Flickr
    Feedback and honest, constructive criticism is greatly appreciated.
  • W G
    Photojournalist
    Photojournalist
    Posts: 2125
    Joined: Wed Jul 22, 2009 10:25 am
    Location: Sydney
    Contact:

    Re: Tilt-shift query

    by W G » Tue Jan 21, 2014 9:47 am

    You have done incredibly well Alex,

    What you will need to get your head around is the GEOMETRY of the world in relation to the sensor with lens shift. You can actually use this phenomenon deliberately to vary the apparent scale of things. A bit complex for now, but we'll get to it in subsequent discussions.

    Best of luck with the new tool.
    Walter Glover

    "Photography was not a bastard left by science on the doorstep of art, but a legitimate child of the Western pictorial tradition." —Robert Galassi
  • User avatar
    Plays with Light
    Reporter
    Reporter
    Posts: 1354
    Joined: Sat Aug 10, 2013 10:34 pm
    Location: Lost In Thought
    Contact:

    Re: Tilt-shift query

    by Plays with Light » Tue Jan 21, 2014 10:07 am

    W G wrote:You have done incredibly well Alex,

    What you will need to get your head around is the GEOMETRY of the world in relation to the sensor with lens shift. You can actually use this phenomenon deliberately to vary the apparent scale of things. A bit complex for now, but we'll get to it in subsequent discussions.

    Best of luck with the new tool.


    Thank you, Walter. I have read some incredibly technical writings on the geometry of tilt-shifts and it made no sense to me at all. Will have a go and re-read it all again and see if it makes any more sense now that I want to know what's going on. I have a feeling I will learn it more through practical experience though, my learning style has changed over the years. I used to be an excellent academic, not these days! :roll:
    Feedback and honest, constructive criticism is greatly appreciated.
  • W G
    Photojournalist
    Photojournalist
    Posts: 2125
    Joined: Wed Jul 22, 2009 10:25 am
    Location: Sydney
    Contact:

    Re: Tilt-shift query

    by W G » Tue Jan 21, 2014 10:29 am

    Okay Alex,

    I have an exercise for you to do which might start to explain the geometry thing.

    Go and find a long low building like that glass thing-a-me with the chairs that you shot.

    Set the camera at an oblique angle so that the building fits within the frame with a bit of breathing space with the lens centred. Shoot.

    Now shift the lens to the full right, pan the camera to set the building centred in frame. Shoot.

    Then shift the lens to the full left, pan the camera to centre the building in frame. Shoot.

    Then look at the variation in aspect ratios of height to width of the building in the resulting images. Along with less spatial perspective, one will show the structure elongated. And it follows that the rendering with more acute spatial perspective will shorten the structure.

    Build on that to get to grips with what else is actually happening when you shift the lens.

    Hop to it,
    Walter Glover

    "Photography was not a bastard left by science on the doorstep of art, but a legitimate child of the Western pictorial tradition." —Robert Galassi
  • User avatar
    Doug
    Reporter
    Reporter
    Posts: 1831
    Joined: Wed Jul 04, 2012 8:49 pm
    Location: Adelaide
    Contact:

    Re: Tilt-shift query

    by Doug » Tue Jan 21, 2014 10:57 pm

    [quote="Plays with Light"]

    ********
    In the second image, you will notice it makes the panels of glass across the rear of the building get progressively wider. WTF? It hasn't done this anywhere else when I took three images in landscape orientation and shifted them left and right to the maximum. Is this some bizarre thing that occurs with certain curves of buildings and not being on an exactly square or level basis with the sensor and the wall? Something else? The first and second frames for this panorama are fine and correct in proportion, the third, shifted fully to the right, has gone all weird though. None of the other captures from yesterday exhibited this anomaly...
    ********

    I'll have a crack at this.
    This explanation of what you are seeing is only relevant if the glass in the image forms a convex shape and not a flat plane, if it is a flat plane then I'm taking rubbish :)

    Look at the first image for reference.
    On the side wall of the building are some louvers.
    The front and back ends of those louvers are in a vertical line because your lens has corrected for any vertical lines, corners and edges within the image.
    The wall at the back of the louvers(with the writing on it) is face on to the camera (well close enough to use as an example), so not only are its sides vertical but its top and bottom are horizontal.
    Though you know the louvers are parallel to each other there is just one that appears horizontal and that will be the one that is the same height as your camera lens.
    The rest of the louvers splay out because their ends are different distances from the camera and not on a plain you have corrected for.

    Now look at the second image and in particular the dividing line between the upper and lower pains of glass.
    There is only one pain of glass where that line is horizontal and that is the one just to the right of the tree trunk, so it is the only one directly face on to the camera, like the back wall in the first image was.

    Though the pains of glass either side of that one might only be 1 or 2 degrees different from the face on one, the outer edges of them are still slightly closer to the camera and therefore you get a perspective effect that is not controlled by the plain you are working with.

    Looking at the chairs in front of the windows for scale and imagining someone standing in front of one, then the dividing line of the glass looks easily higher than your camera height, which explains why the line goes up more and more with each pain further and further and at greater angles to the face on one.

    Another way to picture it is if the window dividing line continued in a full circle all the way round past the front of your camera. That circle would pass above your camera and would appear in photographic form as a wide ellipse shape with the furthermost part at the building being the bottom of the ellipse curve.

    If the divide between the glass was at your lens height the circle would come round and pass right in front of your lens. The front of the circle would be inline with the back of it and photographically it would appear as a straight line.

    Hope that makes some sense.

    Cool new toy. I'm jelious.
    Last edited by Doug on Tue Jan 21, 2014 11:41 pm, edited 1 time in total.
  • User avatar
    Plays with Light
    Reporter
    Reporter
    Posts: 1354
    Joined: Sat Aug 10, 2013 10:34 pm
    Location: Lost In Thought
    Contact:

    Re: Tilt-shift query

    by Plays with Light » Tue Jan 21, 2014 11:35 pm

    Doug wrote:I'll have a crack at this.
    This explanation of what you are seeing is only relevant if the glass in the image forms a convex shape and not a flat plane, if it is a flat plane then I'm taking rubbish :)

    <massive snip>

    Hope that makes some sense.

    Cool new toy. I'm jelious.


    I reckon you are on to something here, Doug! Thanks for that considered response. This is a bank of slightly concave windows (wrapping around me where I stood, so to speak). I'm wishing I had taken a few frames with no tilt or shift applied first as a comparison to refer back to. I'm in the city tomorrow, so I'll see if I can replicate it somewhere.
    Feedback and honest, constructive criticism is greatly appreciated.
  • User avatar
    Plays with Light
    Reporter
    Reporter
    Posts: 1354
    Joined: Sat Aug 10, 2013 10:34 pm
    Location: Lost In Thought
    Contact:

    Re: Tilt-shift query

    by Plays with Light » Wed Jan 22, 2014 7:48 am

    W G wrote:You can actually use this phenomenon deliberately to vary the apparent scale of things. A bit complex for now, but


    I think I get you now, Walter! The lens was physically closer to the enlarged portions of the glass panels and hence why they expanded in width! This probably could have happened with my 24-105mm lens too if I was at the widest end?

    I will try your exercise today while I am in town.
    Feedback and honest, constructive criticism is greatly appreciated.
  • User avatar
    Plays with Light
    Reporter
    Reporter
    Posts: 1354
    Joined: Sat Aug 10, 2013 10:34 pm
    Location: Lost In Thought
    Contact:

    Re: Tilt-shift query

    by Plays with Light » Fri Jan 24, 2014 5:43 pm

    W G wrote:Okay Alex,

    I have an exercise for you to do which might start to explain the geometry thing.

    <snip>

    Hop to it,


    I haven't forgotten about this, I've done it and just need to assemble the images. Have a family guest from interstate here and am too busy as a consequence. Will get to it over the next couple of days.

    I will say, it did what I thought it would do, no strange behaviours there. Pictures to come...
    Feedback and honest, constructive criticism is greatly appreciated.
  • User avatar
    Plays with Light
    Reporter
    Reporter
    Posts: 1354
    Joined: Sat Aug 10, 2013 10:34 pm
    Location: Lost In Thought
    Contact:

    Re: Tilt-shift query

    by Plays with Light » Sat Jan 25, 2014 1:49 pm

    *** I have not applied any corrective measures to these images, they are SOOC in regard of barreling, etc...***

    W G wrote:Okay Alex,

    I have an exercise for you to do which might start to explain the geometry thing.

    Go and find a long low building like that glass thing-a-me with the chairs that you shot.

    Set the camera at an oblique angle so that the building fits within the frame with a bit of breathing space with the lens centred. Shoot.


    _MG_9507_HDR.jpg


    This is pretty much the view I am used to with my 24-105mm lens at its widest. Not exactly as I see the scene with my eye, but not exaggerated too wildly to make it displeasing.

    W G wrote:Now shift the lens to the full right, pan the camera to set the building centred in frame. Shoot.


    _MG_9510_HDR.jpg


    This is almost as my eye sees the scene! There's not just a perspective shift, the proportions have all changed up quite a bit too!

    W G wrote:Then shift the lens to the full left, pan the camera to centre the building in frame. Shoot.


    _MG_9513_HDR.jpg


    This one was a real shocker when I looked through the viewfinder, the left hand side of this building now almost hits you in the eye! This is a favourite for architects renderings, to pull as much dimension and interest from a dead-pan cube as is humanly possible...

    W G wrote:Then look at the variation in aspect ratios of height to width of the building in the resulting images. Along with less spatial perspective, one will show the structure elongated. And it follows that the rendering with more acute spatial perspective will shorten the structure.

    Build on that to get to grips with what else is actually happening when you shift the lens.

    Hop to it,


    I like that this scene has the lanes marked out, so I can see even more of what's going on with each shift. There's a lot of spatial change occurring throughout these three variations. I could see this coming into play nicely at say, a billabong, waterfall or the like. Completely changing the proportions of the place to exaggerate certain aspects of the place.

    Thanks for the suggestion to do this, Walter. It is most interesting to observe and make note of this change.
    Feedback and honest, constructive criticism is greatly appreciated.

Who is online

Users browsing this forum: No registered users and 1 guest