Lens Compression, Fact or Fiction?
As I write this, Storm Ciara has begun and we are hunkered down hoping for the best.
This shot was taken yesterday.
London Skyline from Richmond Park
Quintessentially, this image was the calm before the storm. It was a beautiful, cold winter's day in London.
Jo and I brought forward our weekend walk, due to the imminent storm and ventured up to King Henry's Mound and its surrounding vistas, to get a decent view of London.
In past weekends I have been testing my mobile phone's camera, but this weekend I decided to actually take the kit, so that I could test the Sony FE 70-200 GM 2.8 in anger.
If I'm honest I thought I'd be using this lens more. A 70-200 is sold to us as the "item of desire" and like many of us I dutifully purchased it as the lens I would probably use most. As it turns out I usually reach for the GM 16-35mm or the GM 85mm portrait lens. I know I'm the one that makes the choice. But given that choice it usually stays in the backpack.
The reason I chose to post this image is not so much about its inherent aesthetic or claimed quality, but as a point of interest regarding image compression*.
For those of you not familiar with the concept of image compression it goes something like this...
Longer lenses compress images; standard lenses, not so much, and wide angle lenses don't compress at all.
A simple Google search will produce a hundred references that tell you this is a myth.
They will all take you through the steps to simulate this effect and then explain why it's the lens-to-subject distance that's important, not the length of the lens that you use. They will go on to show you the effect, and then they'll tell you that if you don't move the lens (in relation to the subject) you will get the same perspective. I get that and I agree with them.
So that's it then *Image compression 101.
Wait, stop... the world is full of pedants. I know because I am a fully paid up member of the very club.
It turns out that the interwibble is full of conflicting names for image compression. Image compression is often referred to as Lens Compression, Perspective Distortion and a raft of other terms. I have realised that I have opened up a can of worms and exposed my own limitations in understanding advanced optics...
Every article I have read discusses the obvious example described above. That being the difference between wide angle, standard and telephoto lenses. The examples deal with what is seen in the field of view. They state that if you don't move the camera, all lenses will produce the same image. Just nearer and farther versions. They explain why compression does not exist and that its about lens-to-subject distance... It's all pretty convincing and there are many online illustrations to cite.
So firstly, below is the uncropped version of the image. It has all of the pixels that the camera shot. This will allow you to see how the final shot above is cropped.
London from Richmond Park - Full Image
And, below image is the part of the image that shows image compression.
Look carefully at the buildings.
Some of you will be familiar with some of them, and will have an idea where they are in relation to one another. This is where my observation differs from just about every example online...
We are now looking at very distant objects. We are not comparing them to objects in the foreground.
Only the objects in the distance.
London from Richmond Park - Tightly Cropped Image
So lets look at it... On the extreme right hand side of the clearing is the House of Parliament's Victoria Tower SW1 (that's South West London) and dwarfing it, just three buildings back, is 20 Fenchurch Street EC3 (That's East Central), or, as it is rather disparagingly known as "The Walkie Talkie Building". If this isn't a demonstration of compression... I'll eat my shorts.
Bear in mind that both of these buildings are on the north side of the River Thames. Depth Compression can also cause confusion when considering objects and their lateral proximity to one another.
Note that the London Eye is to the left of the Houses of Parliament and in the centre you can see the twin towers of Westminster Abbey and the Campanile bell tower of Westminster Cathedral with its Byzantine detailing, also both on the north bank. Londoners will note that, the London Eye is, famously, on the Southbank, not the north bank. So, why does it appear to be north of Parliament, the Abbey, and the Cathedral, when viewed from the south west? The lens has "compressed" the river so much that it has become virtually impossible to be able to perceive depth and lateral vicinity becomes hard to interpret as its almost completely 2D. For those of you who are not familiar with London: the London Eye is effectively on the other side of the river from Parliament. They are neighbours, as can be seen on the image below...
Of course, the image is correct as the map below shows...
Lines to the various landmarks
In addition to this, the London Eye looks like it is next to the Nova Victoria (The bronze glazed triangular structure with white diamond cross braces). This building is in Buckingham Palace Rd. There is probably a 15 min walk between the Nova and the London Eye.
It's at this point that the compression effect becomes rather confusing...
We can all see that, at infinity, the image compression is very real indeed, contrary to what is the received wisdom of every article I have read online.
This effect is clearly seen when we are viewing objects that are all very far away, and not when we are comparing close objects and distant objects.
All of the buildings look as though they are on the same subject plane. It would be impossible to tell which building is nearer or farther if they were not overlapped.
As all of the buildings are effectively on the subject plane (I am defining this by noting that they are all equally sharp and therefore depth of field plays no part in the illusion of depth), and the fact is that they are all at the effective infinity setting on the lens.
Additional non-related incidental effects also add to the perception of compression...
I have graded the image. This has dehazed the image therefore buildings that are further back have the similar contrast levels, thus minimising the effect of depth. There's no haze or UV affecting the more distant buildings.
But, what is most confusing to the eye is that the effect of "depth compression" is not replicated through the horizontal plane. This means that buildings seem to stretch across the image, making them impossibly wide. If you look carefully you can see all of the three towers of the Houses of Parliament. I have marked this on the image below so that you can see it...
That's one building, stretching across a third of the city, with Fenchurch street just behind it... Impossible, but there it is.
Can anyone explain this effect with greater clarity?
I hope you find this interesting. If you can explain further please add your comments or let me know. I'm happy to add your information and credit on the post to share the knowledge.
I hope you enjoy viewing the image - Colin
Pixel Peepers' Notes & Newbie Hints & Tips:
SONY A7R MkIII - Lens 70-200mm @ 200mm - f5 - 1/500th Sec - ISO 100
I thought it would be prudent to declare that there is no retouching on image, merely image grading. No tricks. What you see is the original, shot RAW, using:
Grading: I have used the a mixture of the clarity and dehaze sliders in the Adobe RAW Development software to improve the visual quality the cityscape.