Stitching with LCC: how to avoid differing color cast?

[rupho]

Dear Jack and Alan,
thanks for the replies and sorry for taking so long to get back on this. I was totally swamped lately.

I wanted to share with you this screen shot where both methods were applied ( PS: Automate> Photomerge and Script >"load into stack"... as described by Paul and Photomerge as suggested by Jack.
On the left side of the screen shot you see both individual images that got only basic work done in C1 with their respective LCC applied.
As mentioned before it left me with 2 distinctively different tonal values due to the differing LCC correction algorithm.

The middle image got stitched using Photomerge and the image on the right got the PS Script applied.

Basically my observation is: both methods come to the same result.
I have used Photomerge numerous times before and it works very well, BUT in tis case is turned out to be not sufficient enough.
The stitched photo shows too much differing cast which becomes even more obvious when seen in full size.

In the case of these 2 images: one was shot with no rise the other was shot with 24 mm rise. I think thats a situation where the LCC correction has to do so much work that it affects the color too much for PS Photomerge command come to an acceptable result.

My work-around so far has been to to adjust the individual RGB channels to match each other in C1 by pushing the sliders to the same value as compared to its sibling BEFORE I export the image and apply the stitch.

Grischa

 

[Woody Campbell]

On shift vs. pan:

With a wide lens stitching to achieve really, really wide: panning requires the stitching software to do a lot of work in terms of fitting the images together because of "mercator" wide angle distortion (or linear distortion, depending on your lens) at the edges of the images. There may also be issues with blue sky, for example, where the range of exposure values is much longer than a single frame. The software may or may not deal with these issues well and the results may or may not be pleasing. You don't know unitl you sit down at the computer. Shifting you have only the lens's weirdness to deal with and you can see it in advance. Shifting wins in this context.

With a long lens: perspective is flat - you can pan stitch all day with no problem. I use a 90mm on my Leica and easily pan stitch to achieve the equivalent of a 28. Panning is fine in this context. Shifting is marginally better assuming excellent lens quality across the portion of the image circle that you are using (a really big assumption) because the final file is less "manipulated", but JLM is correct that if the big assumption is not correct you are better off in the middle, good part, of the image circle.

In the tech camera world, wides tend to have small image circles and longs tend to have big images circles. Perverse, isn't it.

 

[rupho]

Woody
This recce shot was taken with a 32 mm lens on an Alpa Max with zero and 24 mm rise .
Even shot on a proper nodal point I think vertical rise stitches taken with TSE lenses or tech cameras om reas rise and fall will be more accurate than panning
IMHO anyway
Grischa

On shift vs. pan:

With a wide lens stitching to achieve really, really wide: panning requires the stitching software to do a lot of work in terms of fitting the images together because of "mercator" wide angle distortion (or linear distortion, depending on your lens) at the edges of the images. There may also be issues with blue sky, for example, where the range of exposure values is much longer than a single frame. The software may or may not deal with these issues well and the results may or may not be pleasing. You don't know unitl you sit down at the computer. Shifting you have only the lens's weirdness to deal with and you can see it in advance. Shifting wins in this context.

With a long lens: perspective is flat - you can pan stitch all day with no problem. I use a 90mm on my Leica and easily pan stitch to achieve the equivalent of a 28. Panning is fine in this context. Shifting is marginally better assuming excellent lens quality across the portion of the image circle that you are using (a really big assumption) because the final file is less "manipulated", but JLM is correct that if the big assumption is not correct you are better off in the middle, good part, of the image circle.

In the tech camera world, wides tend to have small image circles and longs tend to have big images circles. Perverse, isn't it.