Perspective change

[duncdives]

<p>Anyone possibly offer any help for me on this question i wonder?<br>

I have a collection of old postcard images of my local area and would like to recreate the same image to show a comparison between 'then and now'<br>

The original images are around 100 years old so i know they will have been taken on a plate film camera, but i'm not sure what the typical plate size may have been in the early 1900's.<br>

There are a number of fixed items of infrastructure on these early images that i can used as an anchor point on the 'today' images, so that if i display the two as a slideshow with fade transition the old picture should hopefully fade to the new picture with as much detail as possible in registration with where it is between the two images.<br>

My question is how do i know what focal length on a 35mm format will match as closely as possible to that of the larger format plate camera, or is it just a case of trial and error?</p>

[frode]

<p>It is not the focal length that is the problem. It is the standpoint when you take the photograph. The focal length must be small enough for the new picture to cover at least as much as the old picture. Then you scale them in post processing til the buildings fit. But you will have to stand at the same point when taking the new photograph as the photographer stood when he took the old photograph. If not, you will have changed the perspective (i.e. the relative size and position of objects in the image). Perspective cannot be changed in post (you can however change the 2D perspective of the image itself, but that will not accomplish what you want).</p>

<p>You will have to use relative sizes and positions of objects in the original image (and some math, or just walk around) to find the original standpoint.</p>

<p>Hope this was of any help in an interesting project.<br>

Best wishes,<br>

Frode Langset</p>

[q.g._de_bakker]

And besided standing in the exact same spot (which also includes having the lens at the same height) as the camera taking the original photo was, you must also point the camera in the same direction. That does not just mean the obvious, but also that you must match the amount of tilt.

[duncdives]

<p>Thanks guys for your input. I know that some of the comments you have added may sound obvious to some, but they are accepted as definite considerations.<br>

One thing i was mulling over though was how would a different focal length and film size behave with respect to the original image in regard of the perspective compression?<br>

We all know that longer focal lengths give more compression, but what effect does the negative/sensor size have on this?<br>

With some of the original images i am replicating there are some very obvious results of this compression effect. I doubt very much that a lens of any real length was used (or even available in 1914) so is this likely to be caused by the large plate size? </p>

[frode]

<p>No, the focal length or/and sensor size has nothing to do with the "perspective compression". It is your standing point that decides this "effect". You get "perspective compression" by moving backwards (and hence people put on a longer lens to get the photographed object to fill the frame again, and that is probably the reason for people mistaking the effect of a changed perspective to be an effect of the longer focal length)</p>

<p>If you stand at the same spot with the camera pointed in the same direction and take two pictures, one with a focal length of say 20 mm, and another with 200 mm, then with respect to perspective the overlapping area of these two images will look exactly the same. The overlapped part of the 20 mm image is an exact, but smaller copy of the 200 mm image.</p>

<blockquote>

<p>I know that some of the comments you have added may sound obvious to some...</p>

</blockquote>

<p>That others know the answer of a question is not a good reason for not asking the question ;-)</p>

<p>Cheers,</p>

[frode]

<p>I made a demonstration for you. I have taken 4 images. Two of them is taken at a "near" distance, and two at a "distant" distance (large enough to clearly demonstrate perspective compression).<br>

The two images at each distance is taken with two different focal lengths: 21 mm and 100 mm.<br>

Finally each image is cropped and resized to match the "100 mm NEAR" image (which is un-cropped).</p>

<p>As you can see, both images taken at the same distance, but with two different focal lengths have exactely the same "compression" of the depth. If you compare the "100 mm NEAR" image with the "21 mm DISTANT" image you see that the 21 mm has compressed the depth in the image much more that the 100 mm. The 100 mm and the 21 mm at the same distance however have the same "compression" of the depth.<br>

This is to demonstrate that it is only the standpoint where you take the picture from that changes "depth compression".</p><div></div>

[blumesan]

<p>Frode,</p>

<p>A splendid demonstration. Totally convincing and something from which many can learn. Thank you!</p>

[frode]

<p>By the way: I don't agree with de Bakker that you have to point the camera in exactely the same direction (if you used an ordinary rectilinear lens). But if you don't do that, you will have to adjust the image with what is called "Perspective correction" in Photoshop or its like. That in fact is an image tilt correction, not an image perspective correction. Image perspective cannot be corrected in post. Tilt can. Tilt do not affect the perspective, only the plane it is projected on (you tilt the flat sensor plane, tilting it back again in post is only a linear scaling along 2 orthogonal dimensions, usually vertical and horizontal).<br>

So position (including height above ground) is what is mandatory.</p>

<p>Cheers</p>

[JDMvW]

<p>Carry copy of original image with you.<br>

Go to the spot with a long zoom lens</p>

<p>If you're not an engineer, use trial and error to set the viewfinder view to match as closely as possible to the older image.</p>

[frode]

<p>Agree with von Weinberg. Trial an error will probably be easier and faster, even if you are an engineer. And if you, before you start, take a picture of the old postcard with your camera, it will be easy to compare new and old by switching between the images. From that you will get an idea of where to move next for a better result.</p>

<p>Interesting project. I wish I had some old postcards from where I live.</p>

[JDMvW]

<p>Of course, just to make it perfectly (nothing is perfect, though) clear, by "long zoom" I meant long-range zoom.<br /> Most flexible, if you don't already have a hernia, is one of these new 18-300mm lenses.</p>

<p>The older, larger format cameras had nothing like that sort of range.</p>

[q.g._de_bakker]

If you tilt the camera, image geometry will change. As you mentioned, Frode, you can rearrange the geometry later (at the cost of image quality). But why would you if you can avoid having to do so quite easily?<br><br>Perspective is the spatial relation between different parts of a scene. When rendered in 2D, it becomes image geometry. Perspective depends on the spatial relation, the relative position, of the observer to any of those parts (as you so perfectly demonstrated). The well known effect of tilting a camera up or down is that it changes the course of lines in the scene. The change you see when you change from looking straight/level at a building to looking up at the same building (and you don't need a camera and lens to see this), the change from seeing parallel lines to converging lines is perspective. We don't really notice when we do, because we have other senses that tell us (should we not know) that we are looking up. We do when we see the same convergence in an image when not looking up. You can't tilt cameras with longish and long lenses without changing the image contect itself. But you can tilt a camera with a short lens without immediately losing your subject out of the frame. So then you can see the perspective effect change how your subject looks.<br><br>Too many words perhaps, because you are right: you can 'fix' this. But again, if the aim is to get the two images, old and new, in register as much as possible, why not avoid the need to correct this by getting it right at the beginning?

[frode]

<blockquote>

<p>If you tilt the camera, image geometry will change. As you mentioned, Frode, you can rearrange the geometry later (at the cost of image quality). But why would you if you can avoid having to do so quite easily?</p>

</blockquote>

<p>I fully agree with you. The best is to have it as correct as practically possible from the start. There is nothing to gain in making the work larger than necessary (unless you are paid by hour)</p>

<p>;-)<br>

Cheers,</p>

[J.W. Wall]

<p>Is there perhaps a copyright issue with reproducing the postcards?</p>

[James G. Dainis]

Copying the photos or postcards would bring up copyright issues; photographing the exact same scene would not. One can take as many photos of roses, bugs, kittens in a basket as one wants and not fear that someone else has already done so and owns the copyright to that idea.

[James G. Dainis]

"We all know that longer focal lengths give more compression..."<P>

 

That is wrong. But, the full frame results look more compressed with a longer lens than with a shorter focal length lens. Cropping the same area of view out of the shorter focal length full frame image to match the longer lens full frame image will show images with the same perspective. Frode has done a good job of showing that perspective remains the same despite focal length. I wish he had also included a full frame shot taken with the 21mm lens showing what had to be cropped. <P>

 

 

 

 

<center><B>

 

<img src="http://jdainis.com/compression_a.jpg"><BR>

200mm lens at 60 feet:<P>.<P>

 

<img src="http://jdainis.com/compression_b.jpg"><br>

50mm lens at 15 feet<P>.<P>

 

 

<img src="http://jdainis.com/compression_c.jpg"><br>

50mm lens at 60 feet<P>

</center></B>

 

You can see in the latter photo that if you were to crop out the chairs and blow the crop up to match the photo size of the 200mm lens photo that the perspective of the chairs would be the same.

[Ed_Ingold]

If you stand in the same spot, you will get the same perspective. If your field of view is as large or larger than the original, the perspective

will be unchanged if you crop to match.

[duncdives]

<p>Brilliant responses from you forum members. Thank you very much for taking the time to explain, and even produce visual examples.<br>

I think the trial and error approach will be my chosen method. I have already been out with copies of the postcard pictures to try and attain the original viewpoints. Some of which are proving very difficult to find due to property development and tree growth.<br>

At least I do have some faith now that I should be able to replicate the field of view for those that are not too dramatically different.<br>

Out of interest, the project is based on Photographs of WW1 recruits and their training camps which were set up in my village during August and September 1914. The pictures have been lent to me by village members and I have been restoring some of them as a thank you. I figured the comparison of 'then and now' would be a nice added extra for them. As regarding any potential copyright issues, the project is a not for profit venture and the end results will only be visible to village residents at a presentation in November this year. Given that the photographs are 100 years old, I was under the impression that they would be out of copyright anyway.... Please correct me though if you believe this may not be the case.</p><div></div>

[JDMvW]

<p>The odds are overwhelming that your original images are in the public domain, although the corporations (particularly Disney) have had their way on the copyright issue for the last couple of decades, so it's sometimes hard to tell.</p>

<p>Trees and new construction are the problem, aren't they?</p>

<p>I naively thought I could add a new component to an old triptych I had shot, but the <br>

trees "jes' growed"</p>

<p> </p><div></div>