[pschlute]

you are trying to analyze just 1 aspect that is irrelevant for photography in real life

To be fair to Mike , he started this new thread as a separate discussion about this one subject "subject compression". You are right that there are many other elements to consider, but he chose to consider this specific one in order to isolate it and inform because there was confusion about it in another thread.

When we change distance to the object or use different FLs and try to "fill the frame", we get perspective distorsions / differencies

If we change only the FL (not the distance to subject) we certainly get differences. We get a differently framed image and distortions. But we do not get "compression". This is the point of this thread.The clue is in the thread title

---------- Post added 12-04-20 at 09:03 AM ----------

Yep, DOF is a product of three factors: Distance to subject, focal length, and aperture value. That's it!

At the risk of going off at a tangent, the FL is an interesting parameter when talking about DOF.

For many years I believed that a longer FL had an inherent property of being able to deliver a narrower DOF. So if I wanted a head and shoulders shot and wanted the narrowest DOF I would zoom out to maximum FL right ? Wrong. It was quite a revelation when I learned years later that this was not the case. For instances where the subject size is kept constant in the viewfinder/film/sensor, the DOF is identical across all FL (except for very close macro)

[Digitalis]

Distance to subject, focal length, and aperture value. That's it!

FYI: these were taken on a full format DSLR, taken at a fixed aperture of f/2.8



I use this image in my teaching quite a bit as it not only shows the resulting images from each focal length, it also displays the necessary working distance for each shot. As an anecdote: a few years back I was shooting a tight headshot of a model with an uncommonly fast* schneider 870mm f/6.7 lens** on my Ebony 8X10 camera, the model said it felt like being in a staring contest with the eye of Sauron.

* f/5.6 being fast for 8X10 lenses, lenses any faster than this are a real headache to focus wide open.

**which works out to be roughly 128mm in 35mm format terms.

[BigMackCam]

Even if you are right about same perspective geometry of different focal lengths (two shots of the same scene from the same distance using different FLs will exhibit identical perspective geometry), you are trying to analyze just 1 aspect that is irrelevant for photography in real life. Maybe someone had misunderstanding about perspective compression or simply used that concept wrong, but I could understand why it occurred.

I'd hoped it was reasonably clear from the thread title and opening paragraph that my intention was to correct a common misunderstanding that focal length affects perspective distortion (and, hence, to dispel the myth of "lens compression"), as it had arisen in another thread

Perspective geometry is indeed just one aspect, but an important one nonetheless - and I'd argue it's far from irrelevant to photography in real life. Once we understand that distance - rather than lens focal length - affects perspective distortion, we more easily understand that to emphasise or exaggerate a subject in relation to its background (a cluster of rocks on the beach, a flower in the meadow, for example), we must get closer to it, whilst widening our field of view to retain context from the rest of the scene. Since we know "lens compression" doesn't exist, we can pick a shorter focal length confident in the knowledge that it will provide a wider field of view without affecting perspective geometry.

Still, this is beyond the limited aim of the thread. I set out to correct a very specific and clearly-stated misunderstanding, and trust this has been achieved. If any readers feel it's irrelevant to them, or they see things differently than me, that's OK... No harm, no foul

The main reason why we use different formats of sensors and different FLs of lenses is our wish to "fill the frame" with subject and have maximum details in the picture. To do that we need to change the distance from camera to the object but not to crop image shot at small FL to get analogical sizes shot with telephoto lens. When we change distance to the object or use different FLs and try to "fill the frame", we get perspective distorsions / differencies.
So you are right purely mathematically, but not "photographically".

We can fill the frame, or fit scene elements into it, by changing distance to subject, yes - but also by narrowing or widening the field of view respectively using longer or shorter focal lengths... yet the results are very different. These are both valid techniques whether used in isolation or (more often) in combination, and it's important that we understand the consequences of each technique in a three dimensional sense, rather than simply what's in frame and what's not. I think you and I are pretty much in agreement here - we simply express the concepts differently

Regarding different formats of sensor and different focal lengths of lens, my choice of one format over another is typically driven by - and balances - three considerations: (1) convenience (due to the size and weight of equipment), (2) final image quality requirements and my tolerance therein, and (3) range of creative control over field of view and depth of field simultaneously, relative to my expected use cases. I prioritise these aspects almost without thinking about them. Often, though, I just pick up whatever camera I fancy using that day, and take a couple of lenses with me that I feel are suitable for the occasion. Importantly, I know I can address most practical use cases quite well with either my APS-C or full frame setups, though each has specific advantages and disadvantages. For instance, if I want maximum subject-to-background isolation, I'll choose full frame and one of my fast primes. I own no lenses that would allow me to achieve such shallow depth of field at similar fields of view and working distances on APS-C.

Again, this is all way beyond the original, limited purpose of the thread

[GUB]

For many years I believed that a longer FL had an inherent property of being able to deliver a narrower DOF. So if I wanted a head and shoulders shot and wanted the narrowest DOF I would zoom out to maximum FL right ? Wrong. It was quite a revelation when I learned years later that this was not the case. For instances where the subject size is kept constant in the viewfinder/film/sensor, the DOF is identical across all FL (except for very close macro)

True but the weird thing is that despite this fact the longer focal length creates more out of focus blur which tends to give visually more background / subject separation.
Which kind of reinforces what others say here - that the practical application sort of overrides pure physics.
Another example is the situation Clackers describes which correctly states that FF dof is the same as apsc. Of course requiring the same viewing distance in the two formats. But who does that - We buy our FF , grab that lovely old lens we had on the apsc and fill the frame just like we did with apsc - and hey where has our dof gone !!.

[Serkevan]

True but the weird thing is that despite this fact the longer focal length creates more out of focus blur which tends to give visually more background / subject separation.
Which kind of reinforces what others say here - that the practical application sort of overrides pure physics.
Another example is the situation Clackers describes which correctly states that FF dof is the same as apsc. Of course requiring the same viewing distance in the two formats. But who does that - We buy our FF , grab that lovely old lens we had on the apsc and fill the frame just like we did with apsc - and hey where has our dof gone !!.

This was a headscratcher for me for a while - like you say, longer FLs always provide more subject separation and get a smoother background, despite the "more or less in focus" area being similar which is somewhat counterintuitive. I think it's easiest to imagine it like this: if you keep the subject constant, then with a longer FL the background is magnified (refer to Digitalis' excellent image comparison a couple posts up - the girl's face is the same size but at 200mm you only have a minimal part of the grass/wall in view), therefore the circle of confusion looks larger, therefore... it's smoothed out.

[GUB]

This was a headscratcher for me for a while - like you say, longer FLs always provide more subject separation and get a smoother background, despite the "more or less in focus" area being similar which is somewhat counterintuitive. I think it's easiest to imagine it like this: if you keep the subject constant, then with a longer FL the background is magnified (refer to Digitalis' excellent image comparison a couple posts up - the girl's face is the same size but at 200mm you only have a minimal part of the grass/wall in view), therefore the circle of confusion looks larger, therefore... it's smoothed out.

The out of focus discs are actually bigger according to "How much blur" How much blur? - A visual background blur calculator
But yes it is reinforced by what tends to be a simpler background.

[Papa_Joe]

Exactly! cropping for instance influences DoF, as long as viewing distance is assumed to be the same, since you're actually magnifying the circles of confusion.

This is one of the most overseen issues of pixelpeeping.

[Wheatfield]

Yep, DOF is a product of three factors: Distance to subject, focal length, and aperture value. That's it!

Reproduction ratio rather than distance to subject.

[BigMackCam]

I use this image in my teaching quite a bit as it not only shows the resulting images from each focal length, it also displays the necessary working distance for each shot.

That's a really helpful image. In addition to compression of subject features and necessary working distance, it also demonstrates how the different fields of view result in more or less background elements being visible in the scene, in relation to the subject. Good stuff...

[Hidrieus]

You have been very helpful. I appreciate your effort and time you have invested to educate me and the community. I am grateful.

[Papa_Joe]

This was a headscratcher for me for a while - like you say, longer FLs always provide more subject separation and get a smoother background, despite the "more or less in focus" area being similar which is somewhat counterintuitive. I think it's easiest to imagine it like this: if you keep the subject constant, then with a longer FL the background is magnified (refer to Digitalis' excellent image comparison a couple posts up - the girl's face is the same size but at 200mm you only have a minimal part of the grass/wall in view), therefore the circle of confusion looks larger, therefore... it's smoothed out.

I would say that does more depend on the narrower viewing angle. If you have double FL and same reproduction ratio (double distance from the Subject) the longer FL will give you half the width of field of view in the same distance behind your subject. That can be seen in the first row of digitalis examples. Best example I have ever see, by the way. Thx!

A long time I had promoted that you have the same DOF with the same reproduction ration and the same aperture value (as BigMack has allready stated above) and therefore you should choose the faster lens above the longer to do portrait photography. Until a forum member pointed out the smaller viewing angle to me, which I now believe ist the more important parameter for background seperation.

[stevebrot]

So you are right purely mathematically, but not "photographically".

He is correct photographically as well, hence the easy to perform photographic proof.

You are correct, however, that in practical terms we generally choose our lens based on the frame composition and generally don't think about "compression/foreshortening". At least I don't unless the composition suffers from the geometry of simply being too close.


Steve

[stevebrot]

For many years I believed that a longer FL had an inherent property of being able to deliver a narrower DOF. So if I wanted a head and shoulders shot and wanted the narrowest DOF I would zoom out to maximum FL right ? Wrong. It was quite a revelation when I learned years later that this was not the case. For instances where the subject size is kept constant in the viewfinder/film/sensor, the DOF is identical across all FL (except for very close macro)

Yep...another question entirely, but also one where a ready proof is available. DOF (even the math) can ultimately be boiled down to physical aperture diameter and final magnification for viewing (hence the matter of viewing distance too). In the past, this is where I would throw in a larger version of my profile image showing that depth of field is actually quite narrow on the original capture.


Steve

[stevebrot]

Once we understand that distance - rather than lens focal length - affects perspective distortion

For some reason I tend to do a fair amount of gear photos and have found of late that I have not been taking sufficient care to not get too close. In the photo below, I inadvertently "bent" my new Samyang 14mm a little. Changing from a 35mm to a 50mm lens probably would have lessened the apparent "bend" quite a bit, simply because I would have had to back off.


Pentax K-3, XR Rikenon 35/2.8


Steve

(...am pretty impressed with the XR Rikenon 35/2.8...)

[normhead]

FYI: these were taken on a full format DSLR, taken at a fixed aperture of f/2.8



I use this image in my teaching quite a bit as it not only shows the resulting images from each focal length, it also displays the necessary working distance for each shot. As an anecdote: a few years back I was shooting a tight headshot of a model with an uncommonly fast* schneider 870mm f/6.7 lens** on my Ebony 8X10 camera, the model said it felt like being in a staring contest with the eye of Sauron.

* f/5.6 being fast for 8X10 lenses, lenses any faster than this are a real headache to focus wide open.

**which works out to be roughly 128mm in 35mm format terms.

There ya go... the whole topic adequately explained in way not open to bias or mis-interpretation.

Studying these images is worth more than reading a 10,000 word essay. It always amazes me when people use words to describe readily demonstrable visual phenomena. Why do they do that? It's all about the images.

Well is was correct if you crop an image and hold how you are going to view the image then yes DOF changes and BMC testing did show this

Were you stoned when you wrote that? Not that I'd be one to talk.