[HoBykoYan]

Dear All,

Is there a way to compute the distance from a person (portrait from head to waist) in relation to a focal length of a lens in order not to distort (flatten or stretch) the face and maintain an optimal 3D/natural look of that person?

By chance i happen to make a person thinner using my 55mm lens 5 meters away (head - to right above the knee) and there should be a formula to calculate such distance somewhere.....

Calling all expert in the portrait field, please share us your knowledge (am just hoping this isn't a dumb question, just trying to learn the right technique).

Thanks in advance...

---

Last edited by HoBykoYan; 03-28-2011 at 02:55 AM. Reason: better title

[newarts]

Dear All,

Is there a way to compute the distance from a person (portrait from head to waist) in relation to a focal length of a lens in order not to distort (flatten or stretch) the face and maintain an optimal ....

A person thinks perspective is correct when the camera is the same distance from the scene as the person would be when viewing the real scene. Therefore the camera-subject distance should be whatever is the normal personal spacing appropriate for the scene in your culture.

By chance i happen to make a person thinner using my 55mm lens 5 meters away (head - to right above the knee) and there should be a formula to calculate such distance somewhere.....

5 meters sounds a bit too far for me. I think the scene you describe is about 2m high. For 2m to fill the APS-C frame in portrait orientation the distance would be:

Distance/55mm = SceneHeight/24mm or Distance = 4.5m just as you say. But I think I am typically closer than 5m to people when I visualize them knee to head.

Here's a great series of shots illustrating the perspective effect by Steve Easterwood -he used a Full Frame camera so you should adjust focal lengths accordingly.


I think you should make a similar varying distance head-knee series of a model you know - then you will be able to decide what is the best distance hence best focal length.

I would like to see such a series if you should happen to make one.

EDIT it is also true that people look at a lot of photos and are thereby conditioned regarding what a photo "should look like". I suspect that you'll find your 55mm lens gives non-cropped results close to what people expect to see. I look forward to seeing evidence or experience one way or another on this question.

[Lowell Goudge]

For simple calculations remember

Image size = subject size x focal length / distance

But I know this does not answer the question and this only applies for being back away from the subject by 10 focal length

I think if you look at the magnification formulas for a lens and then consider how magnification changes for near focused items you might be able to come up with a formula but the bigger issue is not just distance but perspective distortion at the edge of the frame

[freewheeler]

http://www.tawbaware.com/maxlyons/calc.htm

[paperbag846]

As a general rule, I think (for ASPC):

Focal lengths 40 - 55 are good for upper body portraits (and further) although you can get away with some tight headshots if you are working with a 55mm lens.

Focal lengths 70+ are best for headshots (and further).

[m42man]

This is an interesting subject. I think that, for a given amount of depth portrayal of the subject's face, the subject-to-camera distance will be a constant - regardless of how tightly or loosely framed you want the "portrait". In other words, you take the shot from a defined distance, and change the focal length according to how much of the subject's body is to be included.

Steve Easterwood's series of portraits seem to back up the rule of thumb from 35mm film camera days that the best FL for a head-and-shoulders shot will be 75-80mm, and that corresponds to 50-55mm, say, for an APS-C camera. Working backwards, I find that this in turn corresponds to a subject-to-camera distance of no more than about 4 feet.

That would mean that you have to get within 4 feet of your subject, and then choose a focal length which gives you the right degree of framing. (I think this implies that your maximum FL will be 55mm, and that maybe the kit zoom would therefore be a very useful lens - especially given that a small degree of softness is usually perfectly in order for a portrait).

HOWEVER, the above assumes that you want a constant depth portrayal for the face, and I'm not sure this is desirable: I think you need to convey a more distant perspective if you're going to include much of the subject's body. This would imply that you stick with your 50-55mm lens, and increase the distance from your subject.

Possibly the best solution would be a compromise between constant distance and constant FL. This would result in FLs up to a maximum of 55mm, and minimum subject distance of about 4 feet (these figures corresponding to a head-and-shoulders portrait).

HOWEVER (again), you may want throw all the above out of the window, and use longer FLs, which won't give you perhaps the optimum perspective, but will at least allow you a more convenient subject distance!

[sterretje]

http://www.tawbaware.com/maxlyons/calc.htm (already mentioned by freewheeler)
Lens focal length calculator

[Lowell Goudge]

I am not sure we have answered well the OPs post.

I guess this would be somewhat of a math challenge.

If i interpret the OP correctly, what he wants is a calculator that would predict the effects newarts displays in his graphic.

in math terms he wants

shooting distance = Function(focal length, limit distortion0

I don't know the function needed, but suspect it is related to lens magnification, or the way to express distortion, but expect it would be the difference in magnification for the focusing distance +/- about 100mm (thickness of a head?)

Unfortunately as someone pointed out, this would need different distortion limits for someone sitting, to avoid exageration of the feet, etc.... but you see where this is going.

Not sure anyone has developed this type of thing other than the graphics such as what newards presented.

It is really something learned with experience not defined mathematically, althouogh it would appear possible if you had enought time.

Me, I would rather go out and tape pictures

[newarts]

I am not sure we have answered well the OPs post.

I guess this would be somewhat of a math challenge.

If i interpret the OP correctly, what he wants is a calculator that would predict the effects newarts displays in his graphic.

in math terms he wants

shooting distance = Function(focal length, limit distortion0

I don't know the function needed, but suspect it is related to lens magnification, or the way to express distortion, but expect it would be the difference in magnification for the focusing distance +/- about 100mm (thickness of a head?)

I am willing to bet no precise standard can be set for "how close is too close" perspective problem for 3/4 portrait photography because of the complexity of human vision and variation in people's individual preferences (which also might vary with culture.)

One possible experiment to get a start is to look at yourself in a full length mirror at various distances & see what "looks right". Try to do it without moving your eyes or head while your mind builds its impression of your image.

You could even take photos of yourself in the mirror at various distances; crop the self images all so they all are framed knee to head. A range of these cropped photos will look pretty good. The distances associated with those good "non-distorted" photos are what you seek. Once you find the distance choose a lens focal length that frames properly for that distance.

One obvious problem is that you are probably the worst possible judge of when you "look right" having never seen yourself as others see you.

I think a better way to do the experiment is to take photos with flat background of a model from different distances and crop them all to have the same framing. Then have a bunch of people who can also see the live model decide which distance best represents reality, which is most flattering etc.

Please let us know what you discover. I bet 4-6 meters will be pretty good.

==============================
I believe the "distortion" mentioned is simply perspective you are not used to seeing. A head portrait (1/2 meter high ~ 18") taken with an 18mm lens on an APS-C sensor requires the camera be 500(/24)*18 = 375mm.~ 15" from the subject. That's the same as looking at yourself in the mirror with your nose 137mm ~ 7" from the mirror. Try it - you won't look so good I think.

[Lowell Goudge]

newarts

I have no intention of doing this. but just as a special point of interest, my two favourite portrait lenses right now are both M42 presets, a 105F2.8 and a 135F2.8.

And yes, I am back a little when using them.

[newarts]

newarts

I have no intention of doing this. but just as a special point of interest, my two favourite portrait lenses right now are both M42 presets, a 105F2.8 and a 135F2.8.

And yes, I am back a little when using them.

Lowell - I assure you the "you" I suggested doing these things was the op! I'm not going to do the tests either

I like long lenses for portraits too. 100mm for a 1.5 meter high portrait is what, about 6.5 meters? Not too far.

[Lowell Goudge]

Lowell - I assure you the "you" I suggested doing these things was the op! I'm not going to do the tests either

good

I like long lenses for portraits too. 100mm for a 1.5 meter high portrait is what, about 6.5 meters? Not too far.

Yeah, the long lenses are really good for outdoor portraits.

Indoor it is a little challenging to stand back 6.5 meters

Actually some of the shots I like the most, I have taken with my old Takumar 200F3.5 preset, but that needs a little more space still.

I have also done a lot of "candid" street shots with an 85, and have used both Super Tak F1.9 and the samyang F1.4. they are long enough that you are not in someone's face, but not so long that you have trouble with people passing in front.

[newarts]

............
I have also done a lot of "candid" street shots with an 85, and have used both Super Tak F1.9 and the samyang F1.4. they are long enough that you are not in someone's face, but not so long that you have trouble with people passing in front.

I like the Samyang 85:1.4 a lot. I have it as the Rokinon version. Here's an out-of-the-camera K100D jpg shot across the dining table (maybe 2 meters ?) wide open incandescent lighting (with a 100% crop to see detail quality). She looks the same in real life too! Notice the smooth rendering of the out-of-focus stuff in the background. This is a good social portrait lens I think.

[Lowell Goudge]

you should post in the samyang club thread

See the attached for similar street candid, full shot and 100% crop for detail at F2.2
https://www.pentaxforums.com/forums/lens-clubs/137335-samyang-lens-club.html#post1433435