Asymmetric movements - gadget or gimmick?

[Emmanuel Bigler]

I hope the moderators will forgive me for a digression but I take the liberty to argue on this point :

you still need to get the g/g parallel to the subject as with any other camera.

My comment is a technical comment about : as with any other camera. And in brief I Would say : as with any other camera, but fitted with a non-retrofocus wide-angle lens.

It has been found by a French photographer (a friend of the French galerie-photo forum) that the good old rule of parallélisme (as we say at home) between the building and the film (or silicon) plane is not true for a camera fitted with a retrofocus lens like many photographers now use for architecture in 35mm format with a tilt+shit lens mounted in front of a SLR.

To be precise, imagine the following situation.
The SRL is mounted with its body (and back) perfectly vertical
In front of the camera, there is a tall building with perfectly vertical and rectangular walls. Yes I know, in the UK you have other kinds of buildings with non-vertical and non-rectangular walls...
http://www.davidmacd.com/images/uk_duba ... bullet.jpg
....but let us keep the problem as academic and simple as possible

You tilt the lens upward, keeping the camera body and film or sensor perfectly parallel to the building. In this situation the image cannot be sharp since the Scheimpflug rule is not fulfilled, but don't small format camera + wide-angle lenses yield a huge depth of field ? So you go for it without any hesitation.

If the lens is a retrofocus, like all SLR wide-angle lenses, in this situation you get a trapezoidal distorsion that you would not get with a view camera lens.
In optical engineering what differs between retrofocus lenses for SLRs and view camera lenses is that retrofocus have a pupillar magnification ratio significantly different from unity. Can be 2 or 3.

This has been simulated with a lens design software by another member of the galerie-photo gang (a professional lens designer) and is due to the fact that retroocus lenses do not have their pupils located in the principal planes, or, which is equivalent, that their pupillar magnification ratio is not unity like in quasi-symmetrical lens designs.

So in fact the good old rule :
For conserving horizontals, if you want things without converging / diverging lines then you still need to get the g/g parallel to the subject as with any other camera.
at a time when architecture photographers use retrofocus tils & shift lenses, should be refined with the additional condition:
provided that you use a lens for which the pupillar magnification ratio is close to one, like good old non-telephoto view camera lenses.

Now, if the lens is a good quasi symmetric view camera lens, agreed with Apple, whether the tilts are base-tilts, axis tilts of asymmetric 1/3 2/3, this does not make any difference as far as converging verticals are concerned. Keep the GG aligned according to the classical rule.

[Apple]

I'm glad you clarified that - I was only referring to LF systems and lenses of course...

Andrew

[Charles Twist]

Hello Emmanuel,
Do I understand that SLR tilt & shift lenses still produce a distorted image of the building? That seems a bit rum. Would you mind giving us a link to the original article, so that we can read a bit more about it?
Thank you.
Charles

[joolsb]

Well, I have to admit to buying my 45SU as a direct result of Joe Cornish demonstrating the benefits on a workshop and after Baxter Bradford (another 45SU owner) pretty much backed up what Joe said from his own experience.

To be honest, I don't very often shoot in the classic 'near-far' style and I don't slavishly conform to the 'rule of thirds' yet I find myself using asymmetric tilts far more than perhaps I should. Why? Because it's so easy! Just focus on the BG and tilt until the FG becomes sharp. No faffing around with tilting and refocussing until it all comes together - just focus once and tilt (or swing, if appropriate). It turns a chore into a joy. Simple as that.

However, that wasn't the only reason I went for the 45SU. My first LF camera was the RSW45 - a very capable beast but not without limitations. I wanted to use longer lenses so an upgrade was necessary. I agonised over whether I really needed the asymmetric movements, given the price differential, but now I'm really glad I went with the SU.

[PAUL O]

Having read through some of the postings on this thread it would appear that there is a misconception/confusion about how these asymmetric movements are employed and what they can and cannot do?
The confusion appears to be this: people think that the "important" elements of the composition need to be on the axis of movement (along the griided lines of asymmetric tilt or swing on the GG) - they don't. All that asymmetric movements do is make applying required swing or tilt a more straightforward! Asymmetric movements do not determine or dictate the style or type of photography you do.
Tricky to put into simple terms without a demo but ... Once the scene is composed it is simply a case of WHATEVER lies on the gridded lines needs to be in focus and a movement applied to bring the scene into focus on the opposite gridded line. You don't have to compose to the rule of thirds etc!
Maybe a bit naive but whether you admire certain "names" within photography (Joe C, etc) or not, if asymmetric movements were simply a gimmick then why would the "names" endorse them? I'm sure that photographers like Joe Cornish and Richard Sexton would be offered a myriad of brands to "promote" by marketing people from ALL camera manufacturers?? They chose the 45SU for a reason? I would suggest that it is only when you ACTUALLY try the asymmetric option for yourself can you comment on whether it is a gimmick or in fact a useful alternative to other types of movement?

[Emmanuel Bigler]

Since Charles is, to the best of my knowledge, one of the moderators here, I assume that I can safely continue the digression.

Do I understand that SLR tilt & shift lenses still produce a distorted image of the building? That seems a bit rum.

Yes, Charles, definitely, provided that the following context applies:

- the lens is a retrofocus design with a pupillar magnification ratio significantly different from 1 ; in this case it was a Canon TSE 24mm, a "solidly" retrofocus lens with a pupillar magnification ratio well above 2 (estimated between 2.5 and 2.8 ; Zeiss distagons for SLR MF cameras have a pupillar magnification ratio around 2, the old 50C distagon model has a pupillr mag. factor of 2.3)

- the building plane and the sensor plane are parallel but the optical axis is not perpendicular to both object and image planes (hence the image can be sharp only on one line, the rest is gradually blurred)

The photographer who discovered the unusual effect is Jean-Paul Planchon. Fabrice subsequently explained the effect and gave an OSLO file for anybody to run it and see what happens.
The discussion is here, in French (sorry)
http://www.galerie-photo.org/n-f-69259.html

The key point is that in this case the image is blurred and composed of the projection of out-of-focus conical projections of the exit pupil. Hence the difference between a lens where the pupils are located in the principal or nodal planes (I mean : The True Offical Nodal Planes as defined in optical engineering and not as defined by Internet chat rooms !), like any symmetrical lenses, and a retrofocus for which the pupils are located quite far from the nodal points.

Reminder : the proper rotation point for panoramic stitching without parallax effects is the entrance pupil and nowhere else ; view camera lens users can safely ignore the distinction with the front nodal point, except if they insist on stiching with LF telephotos

About OSLO(TM) : OSLO is a professional optical design software for which a free version exists, OSLO-EDU, it differs from the professional version mainly by the number of surfaces limited to 10 ; this means one for the object, one for the image and 8 dioptres in the lens. This allows you to silmulate a tessar or an apo ronar but not a 5- or 6-element lens.

CAVEAT : it is always better for a photographer to spend more time in the field taking pictures, even distorted pictures, that spending hours & hours simulating things in front of a computer (except if you are paid for)

--------------------------------------------------

That seems a bit rum.

Speaking about rum : here is a joke in France about the difference between rum from Martinique Island and rum from Guadeloupe Island (a few Islands in the Caribbean sea, remnants of our former glorious empire overseas). It is nothing but a tropical variant of the serious controversy in cheese technology between Brie-de-Meaux and Brie-de-Melun
Aficionados of Martinique claim that Martinique rum is far better than rum from Guadeloupe and vice-versa !

[Charles Twist]

Hello Emmanuel,
Thank you for the extra info. I'll follow that up. It's interesting stuff (to me anyway) so no worries about the subject matter. You'll find the moderators to be quite moderate over here. We even like pictures on this side of the Channel!
Here's one of St Pierre de Montmartre (in Paris) to show that non-rectilinear architecture does not have an AOC like Melun's or Meaux' cheeses (I have never heard of Melun Brie - must try that). From a technical point of view, the film plane was perpendicular to the nave's long axis, on average; but it has to be said that I felt the whole place was held together by divine providence.

Yes I know the highlights are blown - I gave up fighting with them when I found there were more stops difference than the film could handle, just in the windows.
Thanks too to Paul for the extra. I must say that it would be tempting to put the important bits of the picture on a line where focussing will be easier. So I could see how one could become a slave to a non-rule. I also admit that the convenience of non-iterative focussing is a great draw.
Best regards,
Charles
PS

a few Islands in the Caribbean sea, remnants of our former glorious empire overseas

Ah, you had one too...

[Charles Twist]

Julian and Paul,
Do you reckon the 45SU would have been better if the asym movements had been granted to the front standard rather than the rear (in view of the demands of architectural photography)?
Thank you.
Charles

[Emmanuel Bigler]

Thanks Charles for sharing this image inside the Montmartre church ; the image looks strange ; due to the shape of the vaults, or whatever to be understood, the impression is bizarre, and I'm wondering if we are in a case where keeping slightly converging verticals could have made sense ?

I have in mind the "ice-cream-cone" effect of tall rectangulaire buildings taken with an ultra wide angle view camera lens, for which the verticals are perfectly parallel because the photographer has _strictly_ respected the above mentioned rules of projection, but the visual impression, an illusion, is that the top is broader than the bottom.

[Dave Tolcher]

I have had a number of Ebony cameras incl a 45S and needed to change because I needed longer bellows. This meant that either the SU or one of the SV45s were the only real options. Except for some very technical subjects where the asymmetric movements are the only way to achieve a certain effect there is no benefit to it except speed of focus. What the SU offers in addition to the asymmetric tilts is bellows draw. The 45S is a tad short to make it a do anything camera. Asymmetric movements have almost no bearing on the size of JCB or its placement in the composition. Personally I chose not to spend the money and got an SV45Te but if you want a non folder and longer bellows then you have no choice but a 45SU and it will do everything that one could reasonably ever want. But so would a 45S with the same length bellows !

[Apple]

I have in mind the "ice-cream-cone" effect of tall rectangulaire buildings taken with an ultra wide angle view camera lens, for which the verticals are perfectly parallel because the photographer has _strictly_ respected the above mentioned rules of projection, but the visual impression, an illusion, is that the top is broader than the bottom.

I have taken a picture like that with my 90mm SA XL and I get no end of grief at my local club for it being "wrong." It is an external picture of a ruined church steeple with a crenelated top (Heptonstall for those who were there with me.) It was set up "square" by using the spirit levels on the camera and tripod head but viewing the print looks like the building is growing as you go higher up the steeple. It's not physically possible to get this effect of diverging verticals if you're stood on the ground below the tower so it must be the 'illusion' you mention...

[timparkin]

What is the point of asymmetric movements on the rear standard of a camera?

Fist, I've never seen asymmetric swings (but I'm too young in the world of large format cameras).

Hi Emmanuel.. I have assymettric swing on the rear of my 45SU, although Joe has a modified 45SU without rear swing or shift..

As far as asymettric tilt goes, perhaps I can explain a little better why it is useful..

When I want to focus with any view camera, I will view the line that is the axis of tilt and focus on something I want to be in focus.

I will then choose another line parralel to this first and tilt until that line is in focus..

With base tilts, you can't see the line of tilt on the gg at all so you have to do a little bit of guess work..

With centre tilt, you focus on the centre and then choose somewhere near the edge of the gg that you want in focus. The problem is that the object you want is probably (usually) not right next to the edge of the gg. Hence the distance between the axis and you line is maybe 25-30% of the height of the gg.. Remember that number for a second..

With assymetric tilt, you focus on your axis and then choose your other line of focus, Typically this won't be near the edge of the screen but will be close... The distance between these two lines is probably about 50-60% of the height of the gg..

Now when you are tilting, each mm of movement of the top of the camera tilts the back standard by a set angle based on some trigonometry..

For the centre tilt, lets say each mm of tilt makes the angle increase by 1 degree... Because the line of axis on asymettric tilt is further away from the top of the gg.. each mm moves the angle by about 0.5-6 of a degree.. This means you get an extra 50%ish of accuracy of tilt

Also, the distance between the two lines is related to how accurate you can focus on the screen.. To prove this, imagine having a centre tilt and wanting to get a plane of focus with something very close to the centre.. The chances of getting the plane right for the rest of the picture would be slim (for a metaphor, when the lines are close it's like aiming a pistol. When the lines are far away from each other it's like aiming a rifle..).

Ok .. that might not have been simpler..

Anyway - I find that 70+% of pictures I take have seen advantageous use of assymetric tilts as something I've wanted to focus on has been on the axis of tilt..

As for front vs back tilt, for the amount of tilt I usually use (about 1-3 degrees) the effect on the landscape is almost impossible to see, hence I think I've only used front tilt in about 5% of pictures - can anybody point out which of my pictures have distortion in them?

Finally - myself and Charles did have a discussion about the 45SU vs the Toyo and the only things that the 45SU could do that the Toyo couldn't were bellows draw and shift. However, I have my suspicions that some of the things the 45SU could do would be quicker, more enjoyable and possibly more accurate than the Toyo? Would you see that in the pictures taken? no. Would I prefer to use the Toyo and save a grand or so? No..

Tim

[Apple]

As for front vs back tilt, for the amount of tilt I usually use (about 1-3 degrees) the effect on the landscape is almost impossible to see, hence I think I've only used front tilt in about 5% of pictures - can anybody point out which of my pictures have distortion in them?

Hi Tim,

You'd be hard pushed to spot things in a landscape picture with distortion unless you've got known shapes / reference points in, e.g. buildings.

It's when you start taking pictures like Charles' and this one that it becomes obvious if it is "out"



If you're using wide angle lenses then the amount of movement will be small anyway - longer lenses generally need more tilt / swing to achieve the same effect as per the Merklinger equations.

[Paul.]

Am struggling to get my head around this concept and will have to consult the book of knowlage, but just had to congratulate Apple for the stunning picture he has posted here, if ever I achive one as good I will of fulfilled an ambition.
Regards Paul.

[Charles Twist]

Hello Paul,
I won't take that personally...

Hello Tim,
Nice to see you taking part on this one as you have much to contribute. I'll grant you the ease of use thing, but the money you save is a lot more. Prices new are almost at a ratio of 1:3 and if you consider the ready availability of second-hand Toyos vs 45SU, the ratio could be a lot steeper, say 1:8. So not inconsiderable for he or she on a budget.
Now, in your estimation, would it better to have the asym movements on the front standard for architectural work (we are agreed that landscape does not present a challenge)?

To Everyone else,
Has anyone noticed any of the trapezoidal distortion that Emmanuel has reported, when using tele lenses? If I understand rightly the effect was first noticed on a wide-angle lens (with a 35mm SLR), I am not sure how much effect one would get at longer focal lengths, say 360-500mm, with a tele on a LF camera.

Thank you.
Charles