Perspective and 2X flowers

[rjlittlefield]

if you start stitching a lot of "views" together, to get a wider angle, the appearance (loosely referred to as perspective) depends how you map them on to a 2D picture. Rik territory!

Sorry, I don't have time right now to think very hard about panoramas of small subjects.

But regarding this whole discussion, several categories of images come to mind.

One of them is dropping a chromed ball bearing into the middle of a small scene, shooting down on it with an ordinary macro lens, then re-warping the reflection to produce a wrap-around panorama. See for example these chess pieces.

Another one consists of the pictures that my dentist sent to my insurance company, to convince them that a cracked molar really did need fixing. That was shot with some tiny camera with a wide-angle lens on the end of a dental probe. I have only a vague idea of the working distance, but I know it fits between a tooth and the adjacent cheek with only a bit of stretching. Sorry I don't have an example handy.

A third category consists of wideangle shots showing a closeup subject and its surrounding environment. Examples include this stonefly skin and these cherry blossoms, and the bumblebee and fly with flowers in "Souvenier photos from a hike". These are shot with a 5.8 mm lens, conveniently pre-assembled into a compact digital camera.

It's not difficult to set up images with strong perspective. Here's one with a subject and horizontal FOV similar to the flat-looking stack that I posted earlier:



The above image is not even particularly wide angle; in fact its angle-of-view is roughly the same as 95 mm focal length on 35 mm full-frame. But it appears to have a lot of depth because the size ratio between recognizable elements is almost 2:1. That's because the closest blooms are about 1 inch in front of the entrance pupil, and the background blooms in the crop are another inch behind them.

The previous image is actually a crop from one with even more apparent depth. In the full image, the ratio is more like 12:1, so it's pretty hard to miss. Here it is:



If you want lots of perspective, the recipe is pretty simple: use a short lens so you can get close to the foreground, then be sure there's recognizable stuff enough farther back to get a good ratio. Piece o' cake.

I have never claimed that perspective is irrelevant to macro. Perspective is a big contributor to some "up close and personal" impressions, including all the photos I've linked and shown in this current posting. And I'm on record (in the stonefly thread) as saying

I really like these wideangle shots that show the environment as well as the main subject. I got turned on to them when member MacroLuv was new to the forum. All he had was a compact camera, and he was great at using it to best advantage. Later on, MacroLuv got a DSLR system with a 100 mm macro lens and started posting mostly traditional (though gorgeous) narrow-angle shots. That left me with an unsatisfied craving. But now that I have a good compact to complement my DSLR, I'm occasionally able to at least take the edge off the craving with my own shots.

However...

We got to the current discussion about perspective starting from Brian's photos of fuchsias. What I observed about the fuchsias was that they were shot with a fairly long lens and extension, did not have much front-to-back ratio, and were missing any features that would make perspective easy to see. So I expressed the opinion that "very similar images could have been shot even with a much longer lens".

When challenged, I posted visual evidence to explain why I thought that. The response was what sounded like an accusation that I had cooked the data: "It IS a flat subject, presumably chosen to be so. I wonder why it wasn't angled more.????"

Well, sorry, but the data was not cooked. I chose that particular subject (Alyssum, I think) because I guessed it was about the same size and depth as Brian's fuchsias, and because unlike fuchsias (which I didn't have anyway), the Alyssum blossoms provided a repeating element that would enhance whatever perspective effect there was. Apparently there wasn't much, considering that the subject was (mis)judged to have been simply flat. I happily agree that the picture looks flat, boring, and would not have been shot except for discussion of the issues. Any picture shot for exhibition would have been selected and lighted to look much more 3D. All I'm saying is that given this size subject and the ratios of its dimensions, the cues producing that 3D appearance would have come from aspects other than perspective.

Bottom line, it seems clear to me that perspective matters a lot in some cases and not much in others. As far as I can tell, Brian's fuchsias happened to be a "not much"; I'm sorry if that rankles.

Rather than continuing to butt heads about the examples I've shown, I'd like to see some photo pairs contributed by people with other opinions and viewpoints. Direct comparisons would be most appreciated -- same subject, same framing, same lighting, different optics.

The better we can nail down the tradeoff between lens length and photo appearance in a variety of situations, the better equipped we will collectively be to make that decision in the future.

--Rik

[ChrisR]

Of course I don't accuse of cooking the data. Some misunderstanding there, I'm sorry that inference was drawn and shall have to be more careful with my phrasing.

I originally commented on what I saw, because I saw it! I'm unlikely to be persuaded that I didn't or can't, by reference to maths or other pictures. I have tried to identify what's in the fuschia pic that isn't in the other (other than focus) but if the result isn't being seen, I suppose it's not surprising not to have made any headway there.

I think we'll have to leave it that we see things differently.

[Harold Gough]

I don't think using FOV in the description of perspective is correct. When images are stitched or cropped the FOV no longer matches the normal FOV for the focal length. The perspective of an image will not change when it is taken with different focal length lens.

I'm "with you" there, ie, I understand and agree.
Conventionally it's only the positon of the camera which matters, That's fine for crops, but if you start stitching a lot of "views" together, to get a wider angle, the appearance (loosely referred to as perspective) depends how you map them on to a 2D picture.

This is a similar consideration to that of stitching landscape panoramics.

The images need to all be shot from the same position. As this requires rotation of the camera there are limitations to the expanse of the scene which can be accomodated. Moving the camera position, as required by subjects such as a stretch of river or a town street, gives problems of perspective and of parallax.

Harold

[LordV]

Probably totally misunderstood the panorama conversation, but i do sometimes do bug panorama shots especially when stuck with the MPE-65. I normally take these by lateral movement of the camera and then use autopanopro with linear ? projection. Simple 2 shot pano of a rove beetle. Just to make it more complicated I sometimes focus stack the individual shots before stitching them together - 2nd shot is a 3 shot wide (overlapped) pano using 9 shots in total.

Brian V.

Large Version - Click LINK



Large Version - Click LINK

[Harold Gough]

Probably totally misunderstood the panorama conversation, but i do sometimes do bug panorama shots especially when stuck with the MPE-65. I normally take these by lateral movement of the camera and then use autopanopro with linear ? projection.

Not something I heard of before. I did a quick search. One link leads to discussion about small FOV, whereas I was considering large FOV.

This may be of interest;

http://photo.net/learn/digital-photogra ... -autopano/

Harold

[rjlittlefield]

For pano stitching, it's all a matter of how large the errors are compared to the size of detail you can see.

When the images to be stitched are single macro shots and short stacks, there's often not enough in-focus depth for parallax errors to show up. In this case, stitches work well as long as you get focus the same in both images along the seam line, or alternatively, put the seam line where focus is the same. Even if focus is wrong at the seam line, often you can just blur the sharp one to match the other. After all, it's macro -- nobody is surprised when something is blurred!

In macro, one place that parallax gets to be a big problem is with deep mechanical stacks. For example this June Beetle was carefully shot to avoid parallax because of its huge depth -- 37 mm in-focus deep, but only 30 mm long! With normal optics, a parallax error of less than 1 mm in placement of the entrance pupil would have been intrusively obvious in the final 6321x4104 pano.

Doing pano stitching at the scale of a human is kind of the same, but quite different in feel. When you're photographing a living room or a kitchen, depth of field is large enough that again even a 1 mm shift of the camera can turn in visible seam errors along architectural features like long straight lines. Intuition says that things must scale, and they do, but in this case they don't scale the way that at least my intuition suggests they ought to.

--Rik

[Charles Krebs]

Rik,

If you want lots of perspective, the recipe is pretty simple: use a short lens so you can get close to the foreground, then be sure there's recognizable stuff enough farther back to get a good ratio. Piece o' cake.

Piece o' cake! (well...it's a a piece that has given me indigestion on more than one occasion! )

If you look at the examples in your initial and second post in this thread I think all we're seeing is a matter of "degree". Your animated GIF clearly shows a perspective difference, but it is also clearly not that great. If you look at the picture that shows the front of the 38mm lens relative to the subject it is also clear that while 38mm is certainly not thought of as a "long" lens, it is still relatively distant from both the nearest and farthest parts of the subject that is imaged. So for the vast majority of people the minor size variance between near and far subject points is not sufficient for a sense of depth, and as Brian said, lighting and OOF areas become the more important "keys" to provide additional "perspective ". They all work together. As the size relationship differential increases between foreground and background, (and as the objects in the picture are more familiar to us) it becomes easier for us to "process" a sense of "perspective" even if the lighting and OOF cues are less.

I think the "OlyMinCan-28" shots I posted a while back...
http://www.photomacrography.net/forum/v ... php?t=4864
... demonstrate this. The lighting was (too!) flat (out of my control), and by design there are no OOF focus cues. Yet our familiarity with the subject matter and the inherent "understanding" of the size relationships give a tremendous sense of "perspective".

[rjlittlefield]

Charlie,

As usual, I agree completely with your take. With the exception of the telecentric stacks, all of the images shown and discussed here have some degree of perspective. The question is only how much.

The best way I know to measure "how much?" is in terms of the depth:distance ratio, since that also tells the ratio of size scales between foreground and background.

For your OlyMinCan-28 shots of the Quinalt Rain Forest, the ratio is something like 10,000:1. Likewise for my mowed lawn in "Macro landscape stack using compact digital camera".

For these shots, the huge ratio gives an opportunity for compelling perspective. Of those three images, I get the strongest immediate perception of depth from the mowed lawn. I think this is because the fine repeating element of the grass removes even the need for familiarity. In the first image of the Quinalt Rain Forest, the oxalis leaves play a similar role, but for me the oxalis leaves do not provide as strong a perspective cue as the grass because a) there is a big size jump between the most foreground leaves and the next tier back, and b) the absence of discernable flowers except in the extreme foreground gives me the impression that the foreground is something entirely different from what I see on the rest of the log behind it. My first impression of oxalis on the log is not so much of "depth" as "difference". This is even more true of the second image of the forest. My immediate impression is of oddness, and I have to consciously dissect the image to figure out that the feathery stuff is probably some form of moss and therefore must be tiny so I must be very close to it, compared to the trees in the background.

From 10,000:1 for the rain forest and lawn, we drop to something like 12:1 for the recognizable elements in the full image of my Alyssum. The total depth:distance ratio is about 60:1, going all the way back to that Rhododendron trunk, but that really doesn't help since there's nothing repeating, nothing parallel, and nothing recognizable way back there.

In the Alyssum crop in my previous post, the ratio drops to something like 2:1.

And finally, in the Alyssum stack that started this thread -- the one that I tried to make similar to Brian's fuchsia -- the ratio is only about 1.1 to 1.

After all this analysis and discussion, I guess it's now "obvious" why the Alyssum stack shows only weak perspective cues, and what other situations might be the same or different.

But I think it was not obvious at the beginning, judging from clear confusion about the Alyssum's physical structure and later expressions of surprise about how the numbers worked out.

Getting from confusion to "it's obvious" is a good thing. Progress has been made.

--Rik

[ChrisR]

Progress? Well some.
I did work out for myself how/why lenses behave so "long" at macro distances, and what the ratios are which lessen perspective. Good.

Basic, extreme, obvious examples of strong perspective have been displayed, and explained mostly in terms of simple ratios which might be new, for some folk. Good.

What I commented on in the fuchsia pic has been dismissed, as "obviously not much", on the basis of a different picture, and mathematics. The different, and I would say not very similar at all, picture certainly shows less of it. Satisfaction may be drawn from my saying it looks flatter than I thought it would, if wished.

Looking back at the fuchsia pic, it still isn't "not much", to me, but if others aren't going to see it then there's little progress likely here!
I've satisfied myself that it isn't the lighting, or the OOF areas, and have made some progress on the how's and why's, which work for my perception.

[Harold Gough]

I did work out for myself how/why lenses behave so "long" at macro distances, and what the ratios are which lessen perspective.

To my simple understanding, it has to be primarily FOV if you consider that a macro lens on a long extension may have less than 5 degrees, much the same as that of an extreme telephoto.

Harold

[LordV]

I did work out for myself how/why lenses behave so "long" at macro distances, and what the ratios are which lessen perspective.

To my simple understanding, it has to be primarily FOV if you consider that a macro lens on a long extension may have less than 5 degrees, much the same as that of an extreme telephoto.

Harold

Just to amplify Harold's comment- never seen perspective like this in an MPE-65 picture before - not my picture BTW.

http://www.flickr.com/photos/43894176@N ... 621236505/

[elf]

Just to amplify Harold's comment- never seen persepctive like this in an MPE-65 picture before - not my picture BTW.

http://www.flickr.com/photos/43894176@N ... 621236505/

The Fujinon FE185C086HA-1 1" C Mount 2.7mm F1.8 5 Megapixel Manual Iris Fisheye Lens is only $460 at B&H. What kind of adaptor would you need between the MP-E65 and this lens in order to produce that image?

[Harold Gough]

The lack of perspective seems close to infinite!

The lens:

http://www.fujinon.com/ProductDocs/FE185C086.pdf

Harold

[ChrisR]

Harold - no it isn't field of view, careful with that.
Consider two trees, one behind the other, 5 metres apart, with a twig from the closer one poking sideways.
You could include the twig and rear tree, with the same long lens, from 100 metres, or 10metres or 0.1 metre from the twig. The perspective would be radically different because of the relative distances of the two subjects, but the FOV of the lens is the same.
So part of the perspective effect is simply where you are when you take, (and then look at), the picture.

Pictures taken with long lenses tend to look flatter because we're usually further away.

To repeat from before more concisely, with 21.5 mm sensor:
38 mm lens, Mag 2.5:1, angle of view similar to a 216mm lens on 35mm
200mm lens,Mag 2.5:1, angle of view similar to a 1,135mm lens on 35mm

It cuts up a lot finer than fov, or relative depth. There's a lot about planes and lines and shapes. As a crude example these two will show very different depths. A difference in cross section in the other plane/dimension would show more too.

[Harold Gough]

Pictures taken with long lenses tend to look flatter because we're usually further away.

But not for the same total final image i.e. the same magnification, with the frame covering the same portion of the subject.

Surely, the extension is a factor in the viewing position? (I have to admit uncertainty here).

Harold