Guessing Focal Length By Looking At A Photo

[Ricochetrider]

This question arises from my own experience at shooting with different lenses. Sometimes it's pretty obvious, like when shooting super wide angle stuff... other times, less obvious. Maybe it takes a super experienced eye, to discern between, say, 50mm and 135mm (in 35m format for example). Is there any sort of formula or a loose "rule" of thumb which would be helpful in determining even an approximate focal length in a shot?

 

To add another example- I have a 35mm Voigltander rangefinder and 40mm & 50mm lenses... Seems me that these 2 are WAY too close together to even begin to see a difference.

 

But just in general, I wonder if even the most experienced photographers can look at a shot and tell roughly what focal length lens was used?

 

Thanks!

Tom

[Niels - NHSN]

…

But just in general, I wonder if even the most experienced photographers can look at a shot and tell roughly what focal length lens was used?

 

Thanks!

Tom

They’d say they can, but they probably won’t.

On known objects you’ll probably have an idea from the distortion.

[Gary Naka]

Not if the picture is cropped.

At the SAME distance, the perspective of different lenses is the SAME

 

Experience with using different lenses at different subject distances, and in similar situations.

IOW, what lens would I use to duplicate that shot?

• If I am indoors shooting a group pic, it will 98% of the time be a WIDE lens. So I expect the lens used in a similar pic to be a WIDE lens. The smaller the room, the wider the lens.
   
  
• A shot on the football field from the bleachers, a 300+mm lens, simply because of the distance.
  
• A shot from outfield to home plate, a 600+mm lens, again simply because of the distance, and what lens I use to do that shot.
  

[steve_gallimore1]

Wasn't there a piece of computer software that claimed to be able to do exactly that?

[Dieter Schaefer]

What Niels and Gary said. Not when the image is cropped or a pano stitch and likely not if I don't know the subject/location.

 

DOF might give it away that this wasn't shot at a long distance with a long tele but with a 85mm at f/11 on a DX body (might still be cropped, can't recall)

 

If you know the location and have sufficient experience you might guess that this one again wasn't shot from a long ways off with a supertele but with at 140mm on a DX body; the amount of background included is an indicator (but only if you know where this was shot from and what the distance to the mountains is):

 

What do you think this was shot at (I used a DX body)?

Louvre Blue Hour

 

Or this one (again on a DX body)?

Elkhorn Road

[Karim Ghantous]

Wasn't there a piece of computer software that claimed to be able to do exactly that?

Yes. It's called fSpy. And it works, although the longer the lens, the less useful it might be. And you do need straight lines in the image. A bird in the sky is not going to yield any useful data. CGI artists use it, so if anybody has objections, argue with them, not with me.

 

fSpy

[Jochen_S]

I'm not overly experienced. Usually you can guess a lens' ballpark? A used focal length seems easier to spot in pictures where it got abused, for example when people try portraiture with their 35mm as an only lens. In other cases you might know the limited choice of focal lengths somebody is carrying and use those for guessing guidance. (21), 35, 85/90mm look pretty distinguishable, but spotting if somebody, doing a lot with their 90mm, dug a 135mm out seems harder.

Does it matter? I mean somebody takes an inspiring shot with their kit, I like to reproduce it with mine, assuming both are kind of complete, I shall find something sufficiently suitable. Not having 24, 28 & 105, 135mms at hand inside one's lens line shouldn't break anybody's neck, but yes, sure, I've done studio shots with old kit zooms, when I needed a certain focal length, due to space constrictions.

[JDMvW]

Not really - as said it all depends on the standpoint

[BeBu Lamar]

It's possible but the prerequisite that the photo isn't cropped. When you crop all bets are off.

[Bill C]

But just in general, I wonder if even the most experienced photographers can look at a shot and tell roughly what focal length lens was used?

 

Hi, actually it's possible in many cases, and not really that difficult. But it's not a method that most photographers would expect.

 

Some years back I did some trials to see how close I could estimate on portraits. (It was sort of an experiment to see how well I could judge "perspective" in a photo.) As I recall I could often estimate fl within 20 or 30%.

 

The basic idea is that there is a "correct viewing distance" for photos, where the viewer essentially duplicates the angle of view of the original lens. And at this correct viewing distance there can be a strong sense of "realism." This is something that was once well-known in photography, maybe 1960s or earlier, not so much today.

 

Anyway, I wanted to test how good my judgment of finding this viewing distance was on some random portraits. This was more curiosity about how large the "sweet spot" was, etc., than the focal length. So I viewed a number of images, using one eye, moving forward and back until the image just looked "right." Then, whatever my viewing angle was should essentially match the original camera viewing angle. Which it mostly did to a reasonable approximation.

 

If anyone wants some further understanding of the "correct perspective" principles, I recommend Rudolf Kingslake's book, "Optics in Photography." The pertinent parts can be read as a preview in Google books.

[Karim Ghantous]

It's possible but the prerequisite that the photo isn't cropped. When you crop all bets are off.

Cropping is fine. But fSpy asks for the dimensions of the image plane, which you do need.

[q.g._de_bakker]

Cropping is fine. But fSpy asks for the dimensions of the image plane, which you do need.

No, cropping will make guessing impossible. Knowing the size of the image plane of the uncropped image might help (if you also know the distance). But once cropped, all you can guess at is the focal length of the lens that has the angle of view that would produce the cropped image without cropping, at that unknown distance.

You also need to know the taking format size, of course.

Three parameters you have to guess at, and one completely unknown. Whether by software or not, a futile undertaking.

 

But why would you want to guess what focal length was used to begin with?

[Karim Ghantous]

No, cropping will make guessing impossible

Everything is a crop of something else when you think about it. A 35mm image taken with an 80mm lens is a 'crop' of a Hasselblad image taken with an 80mm lens. Etc, etc, etc.

 

The reason why this tool exists is for matching VFX to photos or footage. It's not often needed, because productions always record focal length, T stop etc. for every shot.

[q.g._de_bakker]

Everything is a crop of something else when you think about it. A 35mm image taken with an 80mm lens is a 'crop' of a Hasselblad image taken with an 80mm lens. Etc, etc, etc.

 

The reason why this tool exists is for matching VFX to photos or footage. It's not often needed, because productions always record focal length, T stop etc. for every shot.

Indeed.

So you need to know the things i mentioned: a) taking format. b) distance, c) size of the image plane, and the Great Unknown, d) whether the image was cropped.

Pointless exercise.

 

If you need the info for 'productions' you need to take notes. Simple as that.

[glen_h]

As far as I know, you mostly know when it is wrong.

 

I have noticed watching football (American football, for those in other countries) that it is not unusual to have a shot from behind, low, and straight down the field.

This is with a fairly long lens, and from the perspective, it looks like the team (seen from behind) is almost to the goal line.

But then later, seen from a little higher, you find that they are nowhere close.

[Karim Ghantous]

Indeed.

So you need to know the things i mentioned: a) taking format. b) distance, c) size of the image plane, and the Great Unknown, d) whether the image was cropped.

You don't need to know distance to get an output. Actually I was wrong above: you don't need to know the size of the image plane at all. I haven't used it in so long that I overestimated what inputs you need. So I loaded it up and loaded the example image. You don't need any inputs, but you do need sufficient reference points. What it does is give you a focal length based on your nominated image plane.

[q.g._de_bakker]

You don't need to know distance to get an output. Actually I was wrong above: you don't need to know the size of the image plane at all. I haven't used it in so long that I overestimated what inputs you need. So I loaded it up and loaded the example image. You don't need any inputs, but you do need sufficient reference points. What it does is give you a focal length based on your nominated image plane.

What nominated image plane would that then be?

 

Anyway, as mentioned by many before, guessing based on perspective is impossible, because perspective does not depend on focal length, but on viewpoint, mainly distance to the subject.

A futile exercise, unless you know enough that you would wonder why, if you know so much about an image, you do not know focal length as well.

And what point does it serve?

[glen_h]

What nominated image plane would that then be?

 

Anyway, as mentioned by many before, guessing based on perspective is impossible, because perspective does not depend on focal length, but on viewpoint, mainly distance to the subject.

A futile exercise, unless you know enough that you would wonder why, if you know so much about an image, you do not know focal length as well.

And what point does it serve?

 

First, you only get relative focal length, commonly relative to full frame 35mm.

 

And yes you only get viewpoint, but also that it is possible to get the shot from that viewpoint.

You can crop to get a smaller part of the image, but you can't get more than is in the image.

 

One normally views a print, or video image, at a reasonable distance. If the perspective is

wrong for that distance, it is noticeable. As above, I most notice this on TV football games.

[Karim Ghantous]

What nominated image plane would that then be?

Perhaps a screenshot of the app might help explain:

 

All you need to do is to find a sufficient number of reference points. On the right you can see the output data. On the bottom right, you select the camera that you either know was used, or that you want to use to get the same perspective. The focal length is then determined by that.

 

This tool is not a vanity exercise, such as an overlays or preset pack (get both my custom Adobe preset packs for 50% off, for one week only!). This actually has a practical use, and people do occasionally make use of it. And programmers have devoted their time and effort to make it available at no cost, to anyone who wants it.

[q.g._de_bakker]

No, no. A screenshot, or Glen's explanation, does not change anything. It is a guessing game, at best. And it contains one unknown that can only be guessed at blindly.

A futile exercise.

[glen_h]

No, no. A screenshot, or Glen's explanation, does not change anything. It is a guessing game, at best. And it contains one unknown that can only be guessed at blindly.

A futile exercise.

 

Yes it is a guessing game, but some guessing games are closer than others.

 

Many (many!) years ago, I had thought about buying a 43mm-86mm zoom lens.

Someone explained to me that 86mm was close (enough) to 90mm used

for portraits. (But I didn't buy it.)

 

I don't know that I would be able to tell from a portrait which lens was used,

but it seems that some people can. We are especially sensitive to the

look of faces.

 

Now, as I said above, I suspect that we (or I) can tell when it is wrong,

though maybe not quite why, or by how much.

 

To make it more specific, consider portraits taking with a plain background

(so no background reference), normal portrait studio lighting, with 60mm,

90mm, and 135mm lenses, and the face height 1/4 of the frame height.

(As noted above, it is different camera positions that cause the

perspective difference.)

 

Should someone (or some computer) be able to tell which lens was used?

That is, someone is given an image, told that it is one of the three lenses,

but not given all three of the images. (I suspect it is much easier with

all three of the images.) That would be to within about 22%, to the

nearest one.

[q.g._de_bakker]

Indeed, if there is a very limited number of known focal lengths to guess... And the situation is controlled...

You can achieve 100% accuracy, you know, if there's only one focal length to guess.

A futile exercise.

 

And you are still forgetting the Great Unknown. If that portrait is taken using three focal lengths from the same position, the images cropped to show the specified head size, you'd be guessing wrong. Even identifying the correct longest focal length is impossible. A nice example of a Gettier Case, by the way.

So forget about it.

[glen_h]

This one:

 

Picking The Right Focal Length Lens For Portrait Photography

 

has pictures taken from different distances, filling the frame in the same way.

[q.g._de_bakker]

Yes. There are many examples in 'how to' books and on the internet.

Those do not make it more plausible that guessing the focal length is possible with a reasonable degree of accuracy. So i appreciate the effort, but your link is useless.

[httpwww.photo.netbarry]

Generally, except for fairly extreme cropping, I can tell between a fairly long tele and a wide angle basically because of the compression, but other then that, I find the midranges difficult to differentiate in any exact way. Especially between a 40 and 50.