8X20 envy...

[Matus Kalisky]

You chose the hard way, Jim. But it not only produces a beautifull work, but also brings you a lot of satisfaction (obviously) what makes the whope proces worthwhile.

Still - I would have a purely technical question towards the stitching. When you take the two photographs - do you just use the front left-right shift (you would need a lot I guess) or you rotate the camera towards the two different views. My experince is that if I do the second I get the images "skewd" so it would not be possible to stich them directly without using some dedicated stitching program. This could work if the overlap is very small but still - at such print size you are doing any imperfection in the stitching would be magnified.

[jim kitchen]

Dear Matus,

Yes, it is a hard way to make an image, but again my curiosity got the better of me...

I use a combination of shifts and tilts to keep the perspective and depth of field within reason, and I also just rotate the 8X10 camera to capture the two images, where the images have a common overlap area that approximates ten to fifteen percent, and where I am attentive to a single item in the overlap area to act as the hinge between the two images as I set the camera in the field. I am never completely successful with this alignment exercise because the one of the images always seems to miss the focal mark in the overlap area, where the captured images are skewed up or down from each other because of leveling issues, focusing issues, and shift issues, which cause the final cropped 8X20 format to be smaller than two combined 8X10 negatives.

I should qualify the term cropped 8X20 format...

I identify the width of the combined images in Photoshop after I align the two digital image layers within the new common canvas I created to accept the two digital layers, and then I force the height of the cropped image to fit the 8X20 format from this identified, and accepted width. The final combined and cropped layers are not physically equal to an 8X20 negative, but the combined cropped file contains the digital dimensions of an 8X20 negative, thanks to Photoshop. The 8X20 cropping process creates discarded information from the original layers, which I cannot avoid, but the discarded information at the top and, or the bottom of a layer is usually minimal within my process.

I tried using Photoshop's CS2 stitching first, but I quickly abandoned that algorithm, compared to manually erasing specific parts of the top layer to match the bottom layer's image through a variety of opacities. The parallax issue is always present within the combined digital images, at least for me this problem persists, but with care, repeated undos in Photoshop, and time, I seem to be able to marry the two images with minimal imperfections.

It does take a long, long time with files this large, but I proved to myself that this process can be done with the tools that I currently have. I also discovered that I still have a small reserve of patience...


jim k

[argos33]

Jim,
Thanks for this informative and helpful post. I was considering the exact same process as you, but was deterred by my slower-than average computer and figured it wouldn't be up for the task. You have shown that with patience it is indeed quite possible, and the results are impressive! Nice work!

I too would like to see a crop of the squirrel

Evan

[Marko]

Beautiful work, Jim! But you could seriously benefit from Photoshop CS3 Extended. It's aligning (and stitching) algorithms are a world apart from those in CS2.

[jim kitchen]

Jim,
Thanks for this informative and helpful post. I was considering the exact same process as you, but was deterred by my slower-than average computer and figured it wouldn't be up for the task. You have shown that with patience it is indeed quite possible, and the results are impressive! Nice work!

I too would like to see a crop of the squirrel

Evan

Dear argos33,

Merci...

I thought that I should try to see whether the limits of my system could work, and although the entire process takes a while to complete, the process allowed me to test my own patience. So, now I know.

jim k


Incidentally, here is the little fellow, after he ate...



And here is his infamous sibling...

[jim kitchen]

Beautiful work, Jim! But you could seriously benefit from Photoshop CS3 Extended. It's aligning (and stitching) algorithms are a world apart from those in CS2.

Dear Marko,

Thank you kind sir for your comments...

I do have CS3, but I have not cracked open the package yet, but I will do that shortly, I guess. I understand that Adobe significantly improved the stitching routine, compared to CS2, but for the moment I am a bit too old school to let an algorithm make my life so completely easy that I forget why I started the exercise.

The future sure looks like it will change.

As a side note, the next item on my agenda will be to take five or six 8X10 negatives, where I will focus on five or six different progressively distant points in the scene, totally ignoring any camera movements, develop them, scan them, and then I will marry the scanned negatives together, while using a digital Scheimpflug routine...

That process should really tie my old Mac's knickers in a knot.

jim k

[D. Bryant]

Dear Matus,

Yes, it is a hard way to make an image, but again my curiosity got the better of me...

I use a combination of shifts and tilts to keep the perspective and depth of field within reason, and I also just rotate the 8X10 camera to capture the two images, where the images have a common overlap area that approximates ten to fifteen percent, and where I am attentive to a single item in the overlap area to act as the hinge between the two images as I set the camera in the field. I am never completely successful with this alignment exercise because the one of the images always seems to miss the focal mark in the overlap area, where the captured images are skewed up or down from each other because of leveling issues, focusing issues, and shift issues, which cause the final cropped 8X20 format to be smaller than two combined 8X10 negatives.

I should qualify the term cropped 8X20 format...

I identify the width of the combined images in Photoshop after I align the two digital image layers within the new common canvas I created to accept the two digital layers, and then I force the height of the cropped image to fit the 8X20 format from this identified, and accepted width. The final combined and cropped layers are not physically equal to an 8X20 negative, but the combined cropped file contains the digital dimensions of an 8X20 negative, thanks to Photoshop. The 8X20 cropping process creates discarded information from the original layers, which I cannot avoid, but the discarded information at the top and, or the bottom of a layer is usually minimal within my process.

I tried using Photoshop's CS2 stitching first, but I quickly abandoned that algorithm, compared to manually erasing specific parts of the top layer to match the bottom layer's image through a variety of opacities. The parallax issue is always present within the combined digital images, at least for me this problem persists, but with care, repeated undos in Photoshop, and time, I seem to be able to marry the two images with minimal imperfections.

It does take a long, long time with files this large, but I proved to myself that this process can be done with the tools that I currently have. I also discovered that I still have a small reserve of patience...


jim k

Jim,

I've essentially done the same thing as you are doing except that

1) I use a 4x5 camera not an 8x10.
2) If I rotate the camera I try to do so with the lens nodal point centered on the axis of rotation.
3) Use left and right shift to make each exposure.
4) Stitch the images in Photoshop CS3.
5) I don't refocus the camera if I can avoid it.

After reading your description it seems you are accomplishing your goal the hard way, a good stitching application can ease your effort a lot. As I mentioned CS3 works very well. Also I would not scan in RGB but scan to grey scale from one of the color channels. I ususally use the green channel when scanning pyro developed negatives.

All images need some amount of capture sharpening as well as some sharpening at the final print size.

Don Bryant

[jim kitchen]

Jim,

I've essentially done the same thing as you are doing except that

1) I use a 4x5 camera not an 8x10.
2) If I rotate the camera I try to do so with the lens nodal point centered on the axis of rotation.
3) Use left and right shift to make each exposure.
4) Stitch the images in Photoshop CS3.
5) I don't refocus the camera if I can avoid it.

After reading your description it seems you are accomplishing your goal the hard way, a good stitching application can ease your effort a lot. As I mentioned CS3 works very well. Also I would not scan in RGB but scan to grey scale from one of the color channels. I ususally use the green channel when scanning pyro developed negatives.

All images need some amount of capture sharpening as well as some sharpening at the final print size.

Don Bryant

Dear Don,

Thank for your comments...

As I mentioned earlier, this was an exploratory avenue to see whether I and my equipment could do this without too much grief, and I discovered that I could, as long as I patiently waited for the process to complete. I never did consider this to be a hard way to process my images, just time-consuming, but obviously a few folks may believe that to be true.

My nodal point rotations were never the best, causing offset images, and my second negative seemed to be erratic at best, causing the experiment to crash like a hard drive. I did try the stitching process very early in my procedure, but I decided to ignore that process like a really bad date, then again Photoshop probably ignored me.

That said, I do appreciate timesaving methods, and I will explore the new CS3 functions later. My experience with RGB scanning is just a choice I made after playing with other capture methods, where I came to my own conclusion that the shadow detail happened to be better within my own equipment, and my own subsequent post process method. Everyone has their favorite scanning procedure, and for the moment my procedure happens to be RGB.

As a side note, I did find that my 8X10 married files did not need sharpening, but my initial experiments with several old practice 4X5 negatives did. Matter of fact, I considered adding a touch of Gaussian Blur to the 8X10 married images, to actually soften the final prepared image. My normal sharpening is limited to 80, 0.3, 0 with a regular 8X10 image, prepped for printing.

Anyway, I do appreciate your insight, and your suggestions...

jim k