Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: Which stitch?  (Read 1486 times)
Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 558


WWW
« on: May 15, 2010, 01:21:28 PM »
ReplyReply

Hi!

I'd like to add stitching to my 35 mm tool box to fake a larger format. As I see it there are 4 options:

1-rotating the camera using the 3-way tripod head
2-rotating around the entrance pupil
3-shifting the lens, using a lens with larger image circle
4-same, shifting the camera

The hardware MUST NOT WEIGH ANYTHING AT ALL - well almost.

This excludes option 2 as well as retrofocus MF wideangles let alone dedicated T/S lenses.

For this reason alone, option 1 would be the most attractive to me.
Apart from the zero additional weight (and cost), it would work with my 35, 50 and 85mm Contax lenses, whereas options 3 and 4 would require hardware that limits the use to (LF or Digitar) lenses from ca. f=100 mm and up. However, a Biometar 80 (260 grams) could be used on the Zörk (250 grams).

I read
http://luminous-landscape.com/forum/index....c=30503&hl=
post #1
>Modern programs like Autopano Pro using the "Smart Blend" rendering option reduce the need to rotate the camera around the nodal point of the lens

Just how good is this? Is it self-contradictory to try to improve IQ by faking a larger format, and at the same time accept interpolated rather than real pixels?

The same question goes for option 3. Can PanoTools handle the parallax? And if yes, how good is the IQ?

Kind regards - Hening.
Logged

milt
Jr. Member
**
Offline Offline

Posts: 70



WWW
« Reply #1 on: May 15, 2010, 01:39:53 PM »
ReplyReply

The IQ difference between #1 and #2 depends strongly on what kind of scene is being shot.  If there are only distant objects in the shot, #1 is fine.  As closer objects are included, switching to #2 becomes more important.  What kind of scenes do you plan to shoot?

You say "no weight - almost". Try taking a look at the Nodal Ninja pano head.  It's only a few ounces. I have gotten that head up to 17K (feet) and I'm in my eighth decade.  The weight difference between a normal tripod and a good CF tripod is more than the weight of the Nodal Ninja.

--Milt--
Logged

Los Gatos, California | http://miltonbarber.com
Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 558


WWW
« Reply #2 on: May 15, 2010, 05:28:13 PM »
ReplyReply

Hi Milt!

Thank you for your interest!

>The IQ difference between #1 and #2 depends strongly on what kind of scene is being shot. If there are only distant objects in the shot, #1 is fine. As closer objects are included, switching to #2 becomes more important.

Yes I understand that

>What kind of scenes do you plan to shoot?

The closer ones of your gorgeous! "white-bristlecone" images would be typical for the distance range (and for the choice of subject as well).

>The weight difference between a normal tripod and a good CF tripod is more than the weight of the Nodal Ninja.

The "problem" here is that I have that CF tripod on beforehand ;-) so I can't save more on that. But if there is anything lighter than this 700 grams Gitzo GT 0530, I'm all ears...

> You say "no weight - almost". Try taking a look at the Nodal Ninja pano head. It's only a few ounces. I have gotten that head up to 17K (feet) and I'm in my eighth decade.
 
I have never been strong. And now, at 68, I'm weakened by Lyme disease on top of old age. Every up-hilling is HARD - even WITHOUT ANY luggage. -
I have on this occasion re-visited the NodalNinja site: "NN3 MKII weighs only 1lb 1oz (475 grams)" That is without QR plates, leveller, ...
Of course, the advantage is that it would work with ALL focal lengths.

A question is also: Have I gotten this right: Even pano stitches obtained by method 2 have to be treated by software, transformed from barrel to rectangular (apart from the blending). A form of software perspective correction including interpolation, right? In principle much like the treatment necessary with method 1, correct? In other words like what we try to avoid with hardware shifting. How good is this compared to method 4? In other words: panning around the entrance pupil avoids parallax, but not converging verticals - correct?

And I still would not have hardware shifting for perspective correction (without stitching).
 
Method 4, "shift stitching", OTOH, implies extolling the image circle. - Uff, it's difficult to look through!

BR - Hening.
« Last Edit: May 15, 2010, 05:36:13 PM by Hening Bettermann » Logged

MarkL
Sr. Member
****
Offline Offline

Posts: 334


« Reply #3 on: May 15, 2010, 06:19:58 PM »
ReplyReply

All my landscapes are stitched (multi row) and the only extra hardware I use is a levelling base which is purely for convenience. Try no gear at all first and only add gear if you have problems with what you shoot. Ok, shooting architecture you are going to need the fully rrs scaffolding setup but the software can deal with a lot - I've shot handheld on aperture priority with different focus points in the pano and stitched with no issues.
Logged
milt
Jr. Member
**
Offline Offline

Posts: 70



WWW
« Reply #4 on: May 15, 2010, 06:41:17 PM »
ReplyReply

>>What kind of scenes do you plan to shoot?

>The closer ones of your gorgeous! "white-bristlecone" images would be typical for the distance range (and for the choice of subject as well).

That's pretty close!  My (reasonably limited) experience would say #1 wouldn't work very well.

>A question is also: Have I gotten this right: Even pano stitches obtained by method 2 have to be treated by software, transformed from barrel to rectangular (apart from the blending). A form of software perspective correction including interpolation, right? In principle much like the treatment necessary with method 1, correct? In other words like what we try to avoid with hardware shifting. How good is this compared to method 4? In other words: panning around the entrance pupil avoids parallax, but not converging verticals - correct?

There is no perspective-related processing difference between #1 and #2.  #2 is the correct way to do it, #1 is just an approximation that saves weight but impacts IQ.

I have no experience at all with #3 or #4.

--Milt--
« Last Edit: May 15, 2010, 06:42:39 PM by milt » Logged

Los Gatos, California | http://miltonbarber.com
wolfnowl
Sr. Member
****
Offline Offline

Posts: 5727



WWW
« Reply #5 on: May 16, 2010, 01:53:11 PM »
ReplyReply

Quote from: Hening Bettermann
I read
http://luminous-landscape.com/forum/index....c=30503&hl=
post #1
>Modern programs like Autopano Pro using the "Smart Blend" rendering option reduce the need to rotate the camera around the nodal point of the lens

Just how good is this? Is it self-contradictory to try to improve IQ by faking a larger format, and at the same time accept interpolated rather than real pixels?

The same question goes for option 3. Can PanoTools handle the parallax? And if yes, how good is the IQ?

Kind regards - Hening.

This might answer your question: How to straighten a panorama (in Autopano Pro)

Mike.
Logged

If your mind is attuned to beauty, you find beauty in everything.
~ Jean Cooke ~


My Flickr site / Random Thoughts and Other Meanderings at M&M's Musings
Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 558


WWW
« Reply #6 on: May 16, 2010, 04:32:10 PM »
ReplyReply

Hi Mike,

thank you for the link. It is impressive what APP can do, but it does not answer my question about IQ. However, I found some of an answer here:

http://luminous-landscape.com/forum/index....=43649&st=0

In the 2 images shown in post #18, I think the roof top  looks smoother in the original image than in the interpolated one.

I felt confirmed in my theoretical considerations and have decided and already ordered a Zörk shift adapter.

Thank you all for your interest! - Hening
Logged

JeffKohn
Sr. Member
****
Offline Offline

Posts: 1671



WWW
« Reply #7 on: May 16, 2010, 05:10:32 PM »
ReplyReply

Quote from: Hening Bettermann
Hi Mike,

thank you for the link. It is impressive what APP can do, but it does not answer my question about IQ. However, I found some of an answer here:

http://luminous-landscape.com/forum/index....=43649&st=0

In the 2 images shown in post #18, I think the roof top  looks smoother in the original image than in the interpolated one.

I felt confirmed in my theoretical considerations and have decided and already ordered a Zörk shift adapter.

Thank you all for your interest! - Hening
He wasn't using stitching software in that example, so I'm not really sure what conclusions you can draw from that about the interpolation quality of APP. APP can use Spline-64 interpolation, I have no idea what method Marc was using in his example (it may have just been Photoshop's Bicubic or Bicubic Smoother, which is not optimal). Also consider that in the case of APP you can stitch at very high resolutions, so even if there is a small quality loss at the pixel level, it won't show in prints.

I'm not saying the shift adapter is a bad way to go. I use tilt/shift lenses for stitching all the time. But with the shift approach you're more limited in the aspect ratios you can capture, as well as the amount of resolution you can capture. 3-4 shots with a stitch adapter can't come close to a multi-row APP stitch from dozens of shots, for instance.

I'm curious which lenses you plan to use with the Zoerk adapter?  There are a limited number of MF lenses that can be used with such an adapter and produce stellar results with high-density DSLR sensors.
Logged

Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 558


WWW
« Reply #8 on: May 16, 2010, 06:39:48 PM »
ReplyReply

Hi again

> He wasn't using stitching software in that example, so I'm not really sure what conclusions you can draw from that about the interpolation quality of APP. APP can use Spline-64 interpolation, I have no idea what method Marc was using in his example (it may have just been Photoshop's Bicubic or Bicubic Smoother, which is not optimal). Also consider that in the case of APP you can stitch at very high resolutions, so even if there is a small quality loss at the pixel level, it won't show in prints.

Right, my conclusions were probably not strictly logical or precise. I think I thought in terms of "interpolation in general vs. real pixels".

> I'm not saying the shift adapter is a bad way to go. I use tilt/shift lenses for stitching all the time. But with the shift approach you're more limited in the aspect ratios you can capture, as well as the amount of resolution you can capture. 3-4 shots with a stitch adapter can't come close to a multi-row APP stitch from dozens of shots, for instance.

I am aware of that. My plan is not multi-row stitching. My images require focus stacking on beforehand, and the scene would never be static enough for multi row panos on top of that.

> I'm curious which lenses you plan to use with the Zoerk adapter?  There are a limited number of MF lenses that can be used with such an adapter and produce stellar results with high-density DSLR sensors.

The shortest lens I think will be the Biometar 80 (260 grams). The Flektogon 50 is too heavy (480 grams). Then I want to adapt LF lenses I have: An Apo Symmar 100, Apo Sironar S 135, Apo Fujinon 180, M Nikkor 200, Repro Claron 210. If the need arises, the Apo Ronar 300 and C-Fujinon 450.

I hope that some manufacturer will eventually come up with a full frame, high end EVIL camera, which will allow for shorter non-retrofocus lenses as well as save weight in itself. By the time that happens - maybe at the Photokina in September?? - I may invest in a short Digitar. These are delightfully light! The 72 is 220 grams in the electronic shutter - of which the shutter accounts for the 180 grams! So in a black Compur (80 grams) I would get away with 120 grams! This is the range I like...

Kind regards - Hening.
Logged

Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad