Perspective distortion - an example

Perspective distortion always seems to be a topic that gets keenly discussed around here, so here's a classic example with 3 shots I took when I was quite new to my superzoom camera.

I was standing on the platform and the locomotive was coming towards me quite slowly, so not knowing which might give the best picture, I took three pictures, maintaining the same subject size by shortening focal length (zooming out), thus counteracting the reducing camera to subject distance.

This was a BIG mistake; upon review, when flicking through them in succession, I never could decide which one looked 'right'. Note in particular, the apparent length of the boiler. :rolleyes:

At 185mm (equivalent);
http://www.pbase.com/dhumphries/imag...6/original.jpg

At 117mm (equivalent);
http://www.pbase.com/dhumphries/imag...9/original.jpg

At 79mm (equivalent);
http://www.pbase.com/dhumphries/imag...6/original.jpg

Also note that these are all in the "telephoto" area and nowhere near the full zoom range I have! If you look at the distance the loco travelled, it really isn't that far.

So, let that be a warning to you; think ahead and compose the shot in your mind so you get one good result, rather than 'machine gun' the subject. :D

Anyone got any thoughts?

Dave,

I personally would have picked a spot and let the engine reach that before firing. That way distracting elements could be minimised, and your focus would be spot on. I would say that zooming out as an object approaches, even if they are slow moving, can make for a less than perfectly composed or focused shot, as the act of zooming is one more thing to think about.

I m not sure that 'perspective distortion' is a problem per se. Yes the length of the boiler does shorten, and on this engine it is a relatively long one, but is that really an issue? All telephoto lens will foreshorten by their very nature, I suppose it depends on what kind of shot you are trying to achieve here.

I entirely agree that using a splattergun approach doesn't usually work, with any moving object like this. The first thing to look at / focal point is usually the front, with the number being a good first check to make sure that you have stopped the action successfully, followed by overall depth of field covering the main object, and be it car, bike, train whatever, the same applies. The clarity of the boiler and pipework on the second and third shots is testimony to this fact.

Compositionally, with the first shot, the carriages in the background distract as does the sign to the right and the depth of field on the engine doesn't seem 'right.' (Driver is out of focus as is a lot of the engine.) Second and third are a lot better in this respect, although on the final one the engine has progressed too far, losing the far buffer behind the carriage.

Such subjects are often dismissed, but there is usually a lot of things going on, and bringing them together successfully does demand a lot of practice. I certainly would not try to complicate that unnecessarily.

Ian

Quote:

---

Originally Posted by shreds

I personally would have picked a spot and let the engine reach that before firing. That way distracting elements could be minimised, and your focus would be spot on.

---

Hi Ian,

Yup, that is exactly what I am saying;

Quote:

---

So, let that be a warning to you; think ahead and compose the shot in your mind so you get one good result

---

For all the reasons you say, I have never used any of these 3 images, they're all "duffers". the only way they could prove beneficial was;
a) to demonstrate the wierd effects of perspective compression/expansion, and
b) say to people, don't do this

This was an example of how not to do it and demonstrating why, it was taken in the first 3 months of owning the camera. Your comments re-enforce the message, which is great! (If slightly mis-directed)

Thanks,

I don't see that there is much wrong with the middle one that a touch of crop at each edge to remove distractions won't cure

The best example of perspective distortion was many years ago (about 25 from memory). The scene was of a man and a woman each sitting on their respective ends of a bench. The shot was taken at a slight angle from one end, and as the lenses changed from about 20mm to 300mm the the distance couple were apart changed from complete strangers to an intimate couple.
A little hard to describe but I think you'll get the picture.

I am reminded of an alternative exercise with counter-intuitive result:

Taking a scene from EXACTLY the same viewpoint with w/a then zoom. The zoomed image turns out to be a crop of the w/a with no difference to the shape of things. I got to this wondering how the level of detail compared in a shot taken with w/a (and the comparitive advantage of greater DOF) with a set like a panorama take using tele lens. The panorama thus built takes in exactly the same as the single w/a shot but with more detail.(but maybe if there is a close foreround object it will not be in focus)

The apparent change of shape of the boiler (or closenesss of couple) is caused by changes of angle and shooting position, not the FL.

Quote:

---

Originally Posted by crisscross

Taking a scene from EXACTLY the same viewpoint with w/a then zoom. The zoomed image turns out to be a crop of the w/a with no difference to the shape of things.

---

Hi Chris,

Good post!

There seems to be a lot of mis-information about perspective distortion and wide-angle lenses floating about - I was thinking of shooting something along the lines of what you mentioned in an attempt to debunk some of it a little (don't think I'll bother now) :)

Wide-angle lenses don't cause perspective distortion - the simply allow a photographer to get closer to the subject whilst maintaining the required FoV ... and THAT causes perspective distortion.

Hi Guys,

It seems I've done it again; i.e. expressed myself very badly.

Due to the discussions in other threads on this topic I was well aware it was the shortening of subject to camera distance that was the primary* cause of the "perspective distortion" in this example.
* and indeed the angle of view of boiler is also contributing, which I didn't mention - thanks Chris.

It was the distance aspect I was trying to demonstrate, but failed miserably. :(
I got too hung up on focal lengths and gave the wrong message entirely :eek:

As Ian (shreds) says, it is not really a distortion at all, it is our perception that is flawed.

In respect of wide angle lenses, Colin's last point is absolutely 100% true.

As Chris discusses, a PP crop can give exactly the same visual effect as telephoto zoom, this will include foreshortening effects if you are looking at something a long way off. Again, this is because the camera to subject distance has effectively increased, since it is clearly not focal length.

On the topic of using a lot of telephoto captures to build a wide angle panorama with loads more more detail, has anyone here seen the Gigapxl project? Interesting stuff (but I haven't read it all)

In fact, I have made such a hash of the original post, I was considering starting over, but maybe I should learn my lesson and just shut up!

Alternatively, if anyone else wishes to write the words and use my pics?

It occurs to me I would have been better served labelling each picture with camera to subject distance rather than focal length, but sadly the EXIF data let me down there :o
Can anyone work it out by maths from the images?

Apologies for everyone's time spent here

No worries Dave, you have got us thinking on the subject you wanted to air.

I think there is a related and even more fraught one of the extent that the human eye-brain auto-corrects for distortions its memory and experience can take into account and what one does about it with a camera to emulate. It can deal with the shape of a boiler. It is a more serious landscape problem-subject where distant hills and mountains alway come out very flat in images compared to how one perceives them. I led a walk from our patch yesterday and groaned inwardly when at the best distance viewpoint several camera came out - and I know they will go home and be very disappointed.

Quote:

---

Originally Posted by crisscross

I led a walk from our patch yesterday and groaned inwardly when at the best distance viewpoint several camera came out - and I know they will go home and be very disappointed.

---

Oddly enough, I went to a bit of a local viewpoint myself yesterday, but having had the disappointment previously with wide angle shots, I took a succession of semi-zoomed in shots (only hand held and auto exposure, so not much good for stitching together), but the aim was to spend time later poring over the detail my eyes couldn't resolve, at leisure and in more comfort. It worked! I also did some seriously telephoto shots of one or two areas that looked interesting. This was all with less than ideal lighting, so a return trip is on the cards in better conditions.

So maybe next time "the cameras come out", is it possible to advise them not to take the wideangle view, but zoom in a bit on the detail in one or two places. (With IS on or they could be spoiled by camera shake). Maybe even take an example print or two with you to show them good and bad.

Just an idea,

Quote:

---

Originally Posted by Dave Humphries

So maybe next time "the cameras come out", is it possible to advise them not to take the wideangle view, but zoom in a bit on the detail in one or two places.

---

Essentially, all one needs to do in this circumstance is shoot with a full frame equivlent focal length of around 50mm (the "normal perspective") :)

For a while I tried out increasing the height of mountain (English apology for mountain that is) shots relative to width when extracting .jpg, but didn't find it convincing.

In the case cited Dave, the shoot tends to be straight into the sun (English apology for sun), so you tend to get a just a slightly wavy boundary between over-exposed sky and underexposed land unless you work at it.

Hi Colin,

I was thinking it should be more than 50mm; to compensate for the way the brain studies a given point in detail to the exclusion of what's around it, a sort of mental zoom.

So I'd suggest 80 - 120mm equiv (which even "3x zoom" compacts or kit lenses can do), just so when they get home, there's an equivalent to what they remember seeing, not something wider, even if it is natural.

But that's just my physco-babble,

Quote:

---

Originally Posted by crisscross

~ the shoot tends to be straight into the sun (English apology for sun), so you tend to get a just a slightly wavy boundary between over-exposed sky and underexposed land unless you work at it.

---

You are just going to have to find a nice, north facing viewpoint! :)

Yeah, hadn't considered that complication, oh well, it made sense at the time (or so I thought) :o

Cheers, Chris

Quote:

---

Originally Posted by Dave Humphries

Hi Colin,

I was thinking it should be more than 50mm; to compensate for the way the brain studies a given point in detail to the exclusion of what's around it, a sort of mental zoom.

---

Hi Dave,

For full frame 35mm cameras, 50mm is considered to be the length that best mimicks the perspective that's seen by the human eye; other that that, if folks are after a different perspective then "the world's your oyster" I guess :)