Luminous Landscape Forum

Equipment & Techniques => Cameras, Lenses and Shooting gear => Topic started by: ErikKaffehr on December 22, 2016, 01:05:26 am

Title: Shift pano, software correction, rotational pano or ultrawide with shift?
Post by: ErikKaffehr on December 22, 2016, 01:05:26 am

Hi,

Here is a comparison of four ways to shoot an image:
http://echophoto.dnsalias.net/ekr/Articles/Shoots/FourWays/

The samples below are from that comparison.

(http://echophoto.dnsalias.net/ekr/Articles/Shoots/FourWays/20160312-_DSC4462_vsmall.jpg)	(http://echophoto.dnsalias.net/ekr/Articles/Shoots/FourWays/20160312-_DSC4465_vsmall.jpg)
(http://echophoto.dnsalias.net/ekr/Articles/Shoots/FourWays/20160312-_DSC4469-Pano_vsmall.jpg)	(http://echophoto.dnsalias.net/ekr/Articles/Shoots/FourWays/20160312-_DSC4473-Pano_vsmall.jpg)

Best regards
Erik

Title: Re: Shift pano, software correction, rotational pano or ultrawide with shift?
Post by: kers on December 22, 2016, 04:14:50 am

The first - tilted 16-36 image fixed in light room (A) is the most difficult to get the proportions right.
But if you can do it it will be the same as the others.

The difference is in the detail- the 100% crops will show differences in quality, due to stretching and less pixels A will lowest in quality.

Title: Re: Shift pano, software correction, rotational pano or ultrawide with shift?
Post by: Bart_van_der_Wolf on December 22, 2016, 05:57:57 am

Quote from: ErikKaffehr on December 22, 2016, 01:05:26 am

Hi,

Here is a comparison of four ways to shoot an image:
http://echophoto.dnsalias.net/ekr/Articles/Shoots/FourWays/

Hi Erik,

Thanks for sharing. The single shot 16mm with software correction has the most natural looking perspective, in my eyes. The stitched results all have somewhat strange looking non-linear perspective distortions, with the 2 image (_DSC4473) TSE24 stitch the most acceptable, but still a bit off.

As a quick try to see what was going on, I reoriented the JPEG version of the TSE 24mm stitch with PTGUI to a strictly 100% keystone corrected frontal look effect (by using only the horizontals and verticals in the image), and there was still a significant (Yaw=0.7598, Pitch=-4.5756, Roll=-0.3147) adjustment necessary (see attached smaller size sample). PTGUI then offers additional control, e.g. adjusting the pitch angle to allow a slight amount of key-stoning (which usually looks more natural to non-architects), or a different vertical compression, or a different projection method. None of these additional tweaks were used in the attached example.

To me, that attached result (before additional tweaks), has a more natural look. The subject is challenging because it is on a hill, and we need a lot of shift to keep things undistorted, which tends to cause a perceptual vertical stretching due to the wide angle of field that is required when using a leveled camera. But the challenging nature of the scene is also a good opportunity to test different approaches.

So as useful as the stitching capability of Lightroom may be, it is (not surprisingly) still no competition for a dedicated Panostitcher, IMHO, due to the lack of control. Of course, the dedicated applications require a different (roundtrip) workflow, which may be acceptable for some, but may not be for others.

Cheers,
Bart

Title: Re: Shift pano, software correction, rotational pano or ultrawide with shift?
Post by: kers on December 22, 2016, 07:03:31 am

I agree with Bart...
the TS lens (right-under)with one shot photo level should not have diverging verticals... LR makes a mistake??
and the rotational panorama has changed the direction of looking (- no control there? in LR)
I like ptGUI for these things are taken care of.
-
principally the hcam and the ts lens would be able to give the best result, if used well.
Leave you the distortion of the ( shifted)lens... in this case you will not see it, but its there.

- apart from that.. the camera is a bit too close to the church to make a relaxt photo

Title: Re: Shift pano, software correction, rotational pano or ultrawide with shift?
Post by: E.J. Peiker on December 22, 2016, 07:43:31 am

To me the first one looks the most natural.

Title: Re: Shift pano, software correction, rotational pano or ultrawide with shift?
Post by: TonyVentourisPhotography on December 22, 2016, 08:09:53 am

I am having trouble with all of these.  I am not discrediting the way you shot them...but something is just amis.  I earn my living shooting buildings and panoramas specifically.  All those examples to me look like something went wrong in the stitch, or the camera was not square with the subject to begin with. 

I use a tech cam for shifting, I have used shift lenses for shifting and shifted panos, I have a dedicated spherical rig from Rrs that I can make 360 panos with every piece of gear that I have, and I also shoot single images and adjust in post. 

Each method is a little different, but you can get solid results it's each as long as each one is treated correctly.  The subject matter really guides what I use.

The shots two and three are not straight...was everything level?  At your distance from the building, each shot should have looked almost identical except for slight variation overall, but not nearly such extreme differences in the building.

Title: Re: Shift pano, software correction, rotational pano or ultrawide with shift?
Post by: kers on December 22, 2016, 08:11:20 am

Quote from: E.J. Peiker on December 22, 2016, 07:43:31 am

To me the first one looks the most natural.

maybe but i do not think it is correct...

the roof on the right side is higher than near the tower..
This could be made like this but if not it should be as high or lower

Title: Re: Shift pano, software correction, rotational pano or ultrawide with shift?
Post by: ErikKaffehr on December 22, 2016, 08:36:41 am

Hi,

These images are not intended to be "beauty contest", mere an illustration of the results from four different techniques. I am a bit of a newcomer to T&S photography.

Sometimes discussions arise if T&S is needed or software correction works as well. There are also some discussions on shift panos vs. rotational panos.

So the idea here was to publish some "food for thought" and possibly start a discussion.

For me it is a learning experience.

Best regards
Erik

Quote from: TonyVentourisPhotography on December 22, 2016, 08:09:53 am

I am having trouble with all of these.  I am not discrediting the way you shot them...but something is just amis.  I earn my living shooting buildings and panoramas specifically.  All those examples to me look like something went wrong in the stitch, or the camera was not square with the subject to begin with. 

I use a tech cam for shifting, I have used shift lenses for shifting and shifted panos, I have a dedicated spherical rig from Rrs that I can make 360 panos with every piece of gear that I have, and I also shoot single images and adjust in post. 

Each method is a little different, but you can get solid results it's each as long as each one is treated correctly.  The subject matter really guides what I use.

The shots two and three are not straight...was everything level?  At your distance from the building, each shot should have looked almost identical except for slight variation overall, but not nearly such extreme differences in the building.

Title: Re: Shift pano, software correction, rotational pano or ultrawide with shift?
Post by: ErikKaffehr on December 22, 2016, 08:48:12 am

Hi Bart,

The top row images were both single images, with SW correction on the left image and lens shift on the right one.

I am with you on using external stitchers, the one I use is Autopano Pro. But I sort of try the internal merge in Lightroom first and only try external stitchers if that doesn't work.

Thanks for chiming in.

Best regards
Erik

Quote from: BartvanderWolf on December 22, 2016, 05:57:57 am

Hi Erik,

Thanks for sharing. The single shot 16mm with software correction has the most natural looking perspective, in my eyes. The stitched results all have somewhat strange looking non-linear perspective distortions, with the 2 image (_DSC4473) TSE24 stitch the most acceptable, but still a bit off.

As a quick try to see what was going on, I reoriented the JPEG version of the TSE 24mm stitch with PTGUI to a strictly 100% keystone corrected frontal look effect (by using only the horizontals and verticals in the image), and there was still a significant (Yaw=0.7598, Pitch=-4.5756, Roll=-0.3147) adjustment necessary (see attached smaller size sample). PTGUI then offers additional control, e.g. adjusting the pitch angle to allow a slight amount of key-stoning (which usually looks more natural to non-architects), or a different vertical compression, or a different projection method. None of these additional tweaks were used in the attached example.

To me, that attached result (before additional tweaks), has a more natural look. The subject is challenging because it is on a hill, and we need a lot of shift to keep things undistorted, which tends to cause a perceptual vertical stretching due to the wide angle of field that is required when using a leveled camera. But the challenging nature of the scene is also a good opportunity to test different approaches.

So as useful as the stitching capability of Lightroom may be, it is (not surprisingly) still no competition for a dedicated Panostitcher, IMHO, due to the lack of control. Of course, the dedicated applications require a different (roundtrip) workflow, which may be acceptable for some, but may not be for others.

Cheers,
Bart

Title: Re: Shift pano, software correction, rotational pano or ultrawide with shift?
Post by: Bart_van_der_Wolf on December 22, 2016, 10:54:08 am

Quote from: ErikKaffehr on December 22, 2016, 08:36:41 am

Hi,

These images are not intended to be "beauty contest", mere an illustration of the results from four different techniques. I am a bit of a newcomer to T&S photography.

Sometimes discussions arise if T&S is needed or software correction works as well. There are also some discussions on shift panos vs. rotational panos.

So the idea here was to publish some "food for thought" and possibly start a discussion.

For me it is a learning experience.

Hi Erik,

That's how i interpreted your post.

Shooting with a camera system that allows optical shift, has benefits and drawbacks. Nothing new, isn't it?
A benefit of shifting is that one gets more pixels that require little/less resampling (keystone correction) to achieve the final image. A drawback is that one makes more use of the perimeter of the image circle, which always has a bit lower quality than the optical image center. When the image is shifted, it's a bit easier to previsualize the final crop if a single image is being used, and subject motion is not an issue (if we can use the right shutter speed). It's also relatively simple to stitch multiple shots because only simple translations are required to align images in post-processing (as long as the shifts used the same perspective point, entrance pupil).

However, rotational stitching offers more flexibility and potentially higher quality (provided good technique was used and the subject was stationary enough). A good quality software then helps to get good quality stitches, with a relatively modest effort. It's a bit harder to previsualize the composite shot, but a little practice/experience goes a long way. In fact, it's liberating to no longer be restricted to fixed aspect ratios and instead let the scene dictate what is needed.

Cheers,
Bart

Title: Re: Shift pano, software correction, rotational pano or ultrawide with shift?
Post by: TonyVentourisPhotography on December 22, 2016, 11:11:13 am

Understood.  What I find really helps is using a software like PTGUI for stitching.  It allows some pretty good ways to correct issues while still controlling projections and how much of a projection.  (you can straighten cylindricals for example to be closer to rectilinear)  If you are using Lightroom, even after stitching, the automatic transformation tool sometimes does a pretty good job.  I would try that and then do some manual adjustments.   

Quote from: ErikKaffehr on December 22, 2016, 08:36:41 am

Hi,

These images are not intended to be "beauty contest", mere an illustration of the results from four different techniques. I am a bit of a newcomer to T&S photography.

Sometimes discussions arise if T&S is needed or software correction works as well. There are also some discussions on shift panos vs. rotational panos.

So the idea here was to publish some "food for thought" and possibly start a discussion.

For me it is a learning experience.

Best regards
Erik