Shoot to Print

[Twilight]

Shoot to Print

by Twilight » Mon Sep 24, 2012 9:26 pm

Hi,

I apologize, this is going to be a long post. But I'm not making a connection in the math somewhere, and I'd like some help.

Short story:

These days, I only print panoramas in three formats. I'd like to figure the best way to convert from FOV to inch/cm to go from shooting to printing. It's got to be easy, I'm just drawing a blank. I got an A+ in calculus, so math isn't the problem, I'm just not seeing a step.

Long story:

I first started shooting panoramics in 1999 on film. Processing film, scanning film, and then starting all over if there was a mistake was a slow process. I had one of the early heads with detents. You picked your lens, set your detent number, and fired away. Shooting was easy. Then I moved to NYC and worked for a studio that did mostly VR photography--we had every pano head made, a $60,000 PanoScan camera, a ten foot diameter computer controlled shooting table for doing object rotations, etc. But I still used pano heads with detents for my work. That company went bankrupt and then I started freelancing shooting 360 degree panos for restaurant.com.

After that I focused more on shooting street photography with my Leica and still life with my Linhof 4x5" view camera. When I would carry around that huge camera and all the assorted gear in my backpack, it was very heavy. So I made a viewing mask. A simple piece of mat board with a 4x5 inch rectangle cut out of it. You could get that out and frame a scene without dragging out the 40lbs of gear to see if the view had potential.

I'm back shooting panos and now I have the same tools: I have three viewing masks for my favorite formats: 2:1, 2.4:1, 3:1. I carry them around and can get an idea first if it is worth setting up my pano gear. Maybe I should have taken up a job with lighter gear!

So, I've never used FOV in any type of calculations. I would just find a scene through a viewing mask, figure which lens I wanted to use, and line up things visually. This usually works fine, but I invariably shoot too many frames, and as I'm now trying HDR panos, I'd like to get the correct number each time so I'm not filling up my hard drives unnecessarily. I only want to crop minimally if necessary. This visual approach works, but while I can do this without understanding the calculations, I'm the type of person that likes to know the underlying concept and math.

However, all of the pano calculators online are done in FOV. I don't know what FOV I want; I know for a given scene I want a 3:1 pano. And, when I go into my print driver on my Canon 6300, there is no setting for HFOV x VFOV! Likewise if I go into a framing shop and ask them to cut me a matboard 110 degrees HFOV by 45 degrees VFOV, I would get a strange look…

Question (finally!): How to go from viewing a scene through a viewing mask and then calculate the number of frames and correct overlap to print a pano of that scene out in the field.

Gear I'm using: Really Right Stuff Pano Elements Package and PCL-1 Panning Clamp (no detents, 2.5° graduations). 24mm, 45mm, and 90mm Canon Tilt Shift lenses. 5D MkII.

Phew! I hope some of you knowledgable people have the endurance to read all of that…

Jon

[klausesser]

by klausesser » Mon Sep 24, 2012 9:44 pm

So, I've never used FOV in any type of calculations. I would just find a scene through a viewing mask, figure which lens I wanted to use, and line up things visually. This usually works fine, but I invariably shoot too many frames, and as I'm now trying HDR panos, I'd like to get the correct number each time so I'm not filling up my hard drives unnecessarily. I only want to crop minimally if necessary. This visual approach works, but while I can do this without understanding the calculations, I'm the type of person that likes to know the underlying concept and math.

Hi!

The most sà­mple way would be to use a touch-controller like the TC or a PapyWizard Tablet, look at the scene through your mask, define a startpoint down left and an endpoint up right and hit start.
The head calculates and shoots all the steps between the points. No need to calculate the FOV first.

Using the TC controller you also can pre-set a FOV by using the angle-mode: say you want to cover 90x120°. Set it on the controller and point the camera to the center of your desired FOV. A good viewfinder - i use a Linhof "Universalsucher" helps here.

That works completely independent from the lens you use. The longer the lens the more resolution you´ll achieve.

best, Klaus

[Twilight]

by Twilight » Mon Sep 24, 2012 10:37 pm

Hey Klaus!

Thank you, but I'm not really wanting to add anymore gear to the setup I have above. I'm going trekking in SE Asia and my kit is too heavy and too dependent on recharging batteries already!

Yes, as I mentioned I have three viewing masks which work well.

Any idea on doing the math?

Jon

[klausesser]

by klausesser » Mon Sep 24, 2012 11:07 pm

Any idea on doing the math?

Why using math? You can do the shooting with a manual head also. Aim a point up left ane a point down right and shoot all steps in between manually.
Depending on the lens it´s moer or less shots.

Or do you mean the angles according to the focal length saying "math"? Are you asking for the setting in degrees for the nodal-adapter´s steps?

There´s a tab for that: http://wiki.panotools.org/Entrance_Pupil_Database

You can find the NPP settings and the amount of shots according to the focal length of your lens.

Maybe i simply don´t understand what you mean saying "the math" . .

best, Klaus

P.S.: don´t use the T/S options when shooting panoramas for being stitched.

Last edited by klausesser on Mon Sep 24, 2012 11:10 pm, edited 1 time in total.

[gkaefer]

by gkaefer » Mon Sep 24, 2012 11:28 pm

hmm..

only found one math for FOV:
http://www.cambridgeincolour.com/forums/thread7555.htm

Georg

[Twilight]

by Twilight » Tue Sep 25, 2012 6:38 am

Hi Klaus,

Sorry, I just like math... I know I can do it manually--I figured that out in 1999 with my first manual head!

I'm interested in the field of view of each shot and percentage of overlap as an equation. So, for a 2.40:1 pano and a 24mm lens in portrait orientation (53.1 FOV), how many frames does it take with 30% overlap? Quick! Calculate it on a napkin!

I already know this from practical testing with the RRS head and all my lenses:

2:1

24mm lens, portrait, 4 frames, 20 deg. rot., start at 330 deg.
45mm lens, portrait, 5 frames, 12.5 deg. rot., start at 335 deg.
90mm lens, portrait, 4 frames, 10 deg. rot., start at at 345 deg.

2.40:1

24mm lens, portrait, 5 frames, 17.5 deg. rot., starts at 325 deg.
45mm lens, portrait, 4 frames, 20 deg. rot., start at 330 deg.
90mm lens, portrait, 5 frames, 10 deg. rot., start at 340 deg.

3:1

45mm lens, portrait, 6 frames, 15 deg. rot., start at 322.5 deg.
90mm lens, portrait, 6 frames, 10 deg. rot., start at 335 deg.

I want to know that mathematics behind it, so I could explain it to someone who doesn't know... for example.

Don't use T/S?? What are you talking about? Who would want their horizon in the center always? Rise and fall are incredibly useful in panoramics! And Scheimpflug is your friend if you understand him. Using tilt to increase DOF (so you don't have to close down the aperture) is very useful--especially when you have low light and have wind moving objects in your scene and want to keep the shutter speed up.

Jon

[Twilight]

by Twilight » Tue Sep 25, 2012 6:39 am

Hi Georg!

Well, I think you nailed it. I'm going to have to take a long look at that.

Thanks!

Jon

[Twilight]

by Twilight » Tue Sep 25, 2012 7:35 am

I guess how I got thinking about this was the fact that all the online pano calculators work on FOV. But when I'm out looking at scenes, I don't think about FOV--I think about format. When I would get in the subway train in NYC with my Leica, I didn't think "Oh, I want to shoot a frame with a HFOV of 43.6°." I compose with the format of the viewfinder.

I do the same thing with panoramics. I don't see a scene in the islands near where I live and think about FOV--I think, "Hmm, maybe a 2:1 or a 3:1--which would fit the scene better?" Then I get out my viewing frames. I need to know if the scene works with a format so that I can print it. (I'm leaving out the times when I shoot an irregular shaped pano because that is what the scene calls for). If you are printing, then you can purchase standard panoramic paper, mats, frames, etc., instead of getting everything made custom. Most of the much older pano shooters I've met used two formats: 6x12 and 6x17. Framing supplies, enlarger neg holders, all types of gear fit those two sizes.

So, why not have a calculator that has a drop down menu with a list: 2:1, 3:1, etc., (and maybe custom sizes), then you select sensor size, focal length, desired overlap, etc., and out pops the number of frames and degrees rotation required?

If you have an iPhone, you can use Panavision's "Panascout" app. It lets you view a scene with the format frames most used in the cinema business. Why? Because cinematographers don't look at a scene and think about FOV, they are thinking about what format would be most appropriate for the project.

Oh well...

[Twilight]

by Twilight » Tue Sep 25, 2012 8:54 am

Very interesting... stitching tables:

http://people.csail.mit.edu/ericchan/stitch/

Jon

[klausesser]

by klausesser » Tue Sep 25, 2012 11:46 am

So, for a 2.40:1 pano and a 24mm lens in portrait orientation (53.1 FOV), how many frames does it take with 30% overlap? Quick! Calculate it on a napkin!

Hi Jon!
How many frames you need per row you can see on the click-stops of your head - horizontal and vertical. Usually the clicks are thought for 25-30% overlap ( at least that is with my modified Manfrotto)

I already know this from practical testing with the RRS head and all my lenses:

2:1

24mm lens, portrait, 4 frames, 20 deg. rot., start at 330 deg.
45mm lens, portrait, 5 frames, 12.5 deg. rot., start at 335 deg.
90mm lens, portrait, 4 frames, 10 deg. rot., start at at 345 deg.

2.40:1

24mm lens, portrait, 5 frames, 17.5 deg. rot., starts at 325 deg.
45mm lens, portrait, 4 frames, 20 deg. rot., start at 330 deg.
90mm lens, portrait, 5 frames, 10 deg. rot., start at 340 deg.

3:1

45mm lens, portrait, 6 frames, 15 deg. rot., start at 322.5 deg.
90mm lens, portrait, 6 frames, 10 deg. rot., start at 335 deg.

I want to know that mathematics behind it, so I could explain it to someone who doesn't know... for example.

I understand now . . - but as a photographer i don´t think this way. I usually don´t have a need to explain FOV to somebody . .

Don't use T/S?? What are you talking about? Who would want their horizon in the center always? Rise and fall are incredibly useful in panoramics! And Scheimpflug is your friend if you understand him. Using tilt to increase DOF (so you don't have to close down the aperture) is very useful--especially when you have low light and have wind moving objects in your scene and want to keep the shutter speed up.
Jon

Being a photographer i´m used to view- and field-cameras and also to T/S lenses - from 8x10" to MF-digital and fullframe 35mm - of course.

Shooting panoramas you definitely A) don´t need T/S and B) shifting and/or tilting lenses when you shoot picture to be stitched often cause issues.
Why? Because you shift the optical axis of your lens. That´s ok with a single shot of course - but when you stitch (it´s always spherical as you may know) the stitcher can run into problems - because it assumes the optical axis to be centered in each image to be stitched.

Depending on the stitcher´s lens-modeling it will get corrected easily - or not so easily.

Regarding you always shoot spherical using a NPP head it´s definitively preferable to do any geometrical corrections in the editor.

See:
Using, let´s say, a 4x5" camera (any camera of any size) a wide-angle lens (each lens) works exactly like a sequential spherical shooting using a NPP-adapter: it photographs a part of a sphere in front of the camera and projects it onto a plane inside the camera.

No problem to shift the optical center of the lens or tilt the focal-plane around the center or doing both at the same time for one shot and correct a perspective or do a Scheimpflug this way. But shooting sequential it MAY (most likely) cause issues for the stitcher.

But: do it the way you´re used to.

best, Klaus

PS.: maybe i should add that what i wrote doesn´t relate to one-row horizontal panos.

Last edited by klausesser on Tue Sep 25, 2012 12:44 pm, edited 1 time in total.

[lumelix]

by lumelix » Tue Sep 25, 2012 11:58 am

Hi Jon
These are two seperate problems, this is the missing link;)
First you should calculate the pix/deg of your shift lenses: sensor size/FOV=pix/deg.
This is fix for every fixed lens, even when shifting/tilt them.
This information helps you to select the best lens for a planned pano and the target resolution for the print.

Second is the shooting pattern. You should use 15-25% overlapping. With every lens, you have
a different rotation angle. When your RRS don't have detents, you should use the graduation marks.
Take an integer value that gives you something between 15-25% overlapping.

For your favorite formats: 2:1, 2.4:1, 3:1 there are different shooting patterns, in landscape and portrait mode.
Best way is to create a table with the possible variations, for every lens and orientation mode, from 2 images up to the max FOV that you will shoot.
In such a table you can also calculate the resulting resolution and the max. print size for your preferred print resolution.

So with such a table in your pocket, you can easily and fast select the best lens and pattern when you know the format and print size.
Is this what you are looking for?

[Twilight]

by Twilight » Wed Sep 26, 2012 11:47 pm

Hi Klaus!

How is the weather in Dusseldorf?

Yep, I'm aware of the potential problems--I was a lens technician for many years in the cine business (that's how I got to visit Germany, I trained at Zeiss in Oberkochen and Canon in Utsunomiya. Enjoyed Berlin, Munich, and...Hutthurm! Will go back some day).

I guess it was more of a mental exercise than anything. Better to get back to shooting...

>PS.: maybe i should add that what i wrote doesn´t relate to one-row horizontal panos.

Exactly... that is the only thing I shoot!

Good discussion.

Jon

[Twilight]

by Twilight » Thu Sep 27, 2012 12:05 am

Hello Lumelix,

I'm sorry I made you write all that. I actually have the table already for my RRS head. Do you see the numbers listed in my post above:

2.40:1

24mm lens, portrait, 5 frames, 17.5 deg. rot., starts at 325 deg.

So, with the 5DMKII in portrait orientation, Canon 24mm T/S lens mounted, on the RRS head, I center the camera on the center of the scene, set the rotating head to 0°, and rotate the index mark to 325°. I take the first image there. I swing the camera clockwise (looking down at the camera from above), 17.5° and count frames. On the fifth frame, I am at 35°, the correct number of frame and overlap for a 2.40:1 pano (with a bit of cropping). My table includes 1:1 (I miss my Hasselblad!), 2:1, 2.40:1, and 3:1. I'm going to calculate one for 16:9 (1.78:1) tomorrow. I can roughly estimate which lens by looking through my viewing masks.

>sensor size/FOV=pix/deg

Yes! And when I looked back at my note from when I originally made the table, that is how I did it! Sheeesh. My memory is slipping...

I also posted this elsewhere, and I received a very nice answer:

http://www.tawbaware.com/forum2/viewtopic.php?f=1&t=6536&p=36000&hilit=+tilt#p36000

"Why people don't simply use frame sizes is that for scenes wider than some angle, the physical size (in "frame" units) can be very different than the number of frames taken once the panorama is stitched together, simply because you are mapping a curved space (the world) to a flat surface (the film). As a result, objects get distorted in scale, and this changes with the projection used to flatten the image. Thus the final size of your pano (in "frame" units) will change depending on the projection (http://en.wikipedia.org/wiki/Map_projection), while the FOVs are invariant."

Ok, I'm done. Time to stop talking an take some pictures!

Thank you Lumelix,

Jon

Hi Jon
These are two seperate problems, this is the missing link;)
First you should calculate the pix/deg of your shift lenses: sensor size/FOV=pix/deg.
This is fix for every fixed lens, even when shifting/tilt them.
This information helps you to select the best lens for a planned pano and the target resolution for the print.

Second is the shooting pattern. You should use 15-25% overlapping. With every lens, you have
a different rotation angle. When your RRS don't have detents, you should use the graduation marks.
Take an integer value that gives you something between 15-25% overlapping.

For your favorite formats: 2:1, 2.4:1, 3:1 there are different shooting patterns, in landscape and portrait mode.
Best way is to create a table with the possible variations, for every lens and orientation mode, from 2 images up to the max FOV that you will shoot.
In such a table you can also calculate the resulting resolution and the max. print size for your preferred print resolution.

So with such a table in your pocket, you can easily and fast select the best lens and pattern when you know the format and print size.
Is this what you are looking for?

[klausesser]

by klausesser » Thu Sep 27, 2012 1:47 am

Exactly... that is the only thing I shoot!

Why didn´t you mention it earlier?

best, Klaus

[Twilight]

by Twilight » Thu Sep 27, 2012 7:00 am

Klaus,

Third sentence of my first post:

>These days, I only print panoramas in three formats.

Fifth paragraph, first post:

>I'm back shooting panos and now I have the same tools: I have three viewing masks for my favorite formats: 2:1, 2.4:1, 3:1.



Jon

[klausesser]

by klausesser » Thu Sep 27, 2012 10:37 am

Klaus,

Third sentence of my first post:

>These days, I only print panoramas in three formats.

Fifth paragraph, first post:

>I'm back shooting panos and now I have the same tools: I have three viewing masks for my favorite formats: 2:1, 2.4:1, 3:1.



Jon

Jon, shooting a mosaic for printing also means shooting sequential and usually for a big print this means shooting spherical rather than cylindrical for being planar-projected then - sorry for being edifying(?).

But: simply don´t listen to me if you don´t want - no problem

Klaus

Last edited by klausesser on Thu Sep 27, 2012 10:38 am, edited 1 time in total.

[lumelix]

by lumelix » Thu Sep 27, 2012 11:03 am

Sheeesh. My memory is slipping...

If it's only your memory and not yourself with the camera, then it's ok

[Twilight]

by Twilight » Thu Sep 27, 2012 5:54 pm

Hi Klaus!

I understand the multi row process. I wasn't sure if you were reading all of my posts thoroughly. I've shot multi-row panoramas, object rotations, etc. Multi-pass panoscan shots also. I did the camera placement math and layout for an object rotation of a 40'/12m NYC Firetruck in the American Airlines Hanger at JFK. That was, of course, with much cruder equipment than we have now.

But let's cut to the chase! That way you can stop feeling like you are 'edifying' me and that I'm purposefully ignoring your edification--that's a mistake. We'll just get to what you really want:

I apologize for not originally posting that I only shoot single row panos at this point. I thought that was clear in my post, but it was only clear in my own mind. I see that would have eliminated much of the back and forth and given you more time for other edification projects. Bitte entschuldigen Sie.

Jon

[Twilight]

by Twilight » Thu Sep 27, 2012 6:17 pm

Hi Lumelix,

No, what it is (besides a bit of mental slipping!!) is that I have a successful gallery show coming down this Sunday (100+ attendees; sold 8 pieces!!), and another being hung on Monday (two pieces with committed buyers before the show is even installed!). Two construction projects at home. And, planning for a 4 month photographic trip through SE Asia (which is why I'm testing gear over and over for best practices). Normally, I'd try to include all the pertinent information for a technical question, which I'm generally quite good at. Alas, my original post was lacking a key statement, as Klaus clearly edified me about (That mistake won't happen again).

So, since priorities are what they are, I'm a little distracted when it comes to posting about pano techniqe, which I'm both dredging up from memories of 1999-2000, and learning about the much more advanced tools we have today.

Really... not a bad place to be in! Selling work is very satisfying on a level that isn't just financial.

Again, I really do appreciate your help.

Jon

P.S. Have been reconsidering the free-rotating RSS rotator. Gets hard to see such small degree scales in low light, and I have to constantly peer at it for each shot--instead of looking for other issues. And, I have to wear reading glasses to read the degree markings. Might want to go to an indexing head. Doing a bit of research on these as the prints dry. I think I'll make a new post, but I'd love to have all of you gentlemen's opinions. Especially Klaus'--and he won't let me get away with leaving any pertinent info out...

Sheeesh. My memory is slipping...

If it's only your memory and not yourself with the camera, then it's ok

Last edited by Twilight on Thu Sep 27, 2012 6:17 pm, edited 1 time in total.

[lumelix]

by lumelix » Thu Sep 27, 2012 7:43 pm

Yes, he don't will!