Stitching for a larger effective capture size.

[Michael Floyd]

Why would one bother?

Is there really any advantage gained from the extra effort and constraints imposed by the extended shooting period? (and the extra time in post.)

Are there tangible advantages, ok let's call them differences, that stand apart from greater resolution. Expressed another way, if these stitched images from a larger virtual sensor are downsized, does some essential quality from the larger capture remain? This may be direct or indirect, as in as a result of the lenses used or somewhat different exposure setting required to achieve a given image.

I don't have the answers - I'm just starting out exploring this path and am open minded about what I may find on the way.

I've no doubt this has been hashed out before, I wasn't there unfortunately. So I request any participant's indulgence, at least as far as offering some kind of photographic support to any expressed opinions.

-

Backstory:

Not too long ago I was primarily shooting adapted 135 film lenses on Fuji aps-c, and finding that stitching up to either 36x24 or 30x30 capture really let the lenses show their true worth. I was procrastinating over which full frame body to move to for adapted lenses as I really enjoy the Fuji shooting experience. As I procrastinated and many months passed I started to notice that older GFX bodies were coming up more and more affordably. The short story is that I purchased a 50R with the intent to simply shoot it with my 135 lenses and save all that stitching.

Solved. Happy. End of Story.

... however with the GFX in hand, I of course acquired a few film 645 lenses. Which put me back in a similar space as before, capturing just a portion of the offered image circle. The difference being that before I felt that I was just not quite reaching a level of image quality that I desired, and now I'm well and truly there, and more.

One does wonder though, if the leap from 24x16 to 44x33 impacted me in the way it did, what would a further leap up again bring? And I'm not just thinking final results in a technical way, certainly not simply looking at 'sharpness' or quantifiable qualities. I'm interested in the feel, the process and how that impacts the image, the larger experience, the journey.

-

I've been stitching a few ways so far - initially panning about the nodal point of the lens, although mostly more recently flat shift stitching, by which I mean leaving the lens fixed and moving the camera parallel to the sensor to effect the larger capture.

The easiest and quickest is with my Pentax 645 lenses on a Kipon shift adaptor. Two vertical frames side by side gives me pretty much exactly the old 645 frame size, maintaining the 4:3 aspect ratio. Here's a couple:

Note: I've downsized all the images to modest sizes because my personal search is not about greater resolution, I'm more than happy for others to focus on that though, it is certainly mind boggling the level of detail that is achievable.


P645 35/3.5 at f/5.6


P67 105/2.4 at f/2.4

-

Moving up in capture size: I'm waiting on the arrival of a Vertex stitching adaptor for Pentax 67 lenses to GFX, until that arrives I have a somewhat cumbersome arrangement that does work and conveniently allows me to shift as far as I want in any direction, so it is good for exploring the outer reaches of the image circle. The following two images are stitches of 5 images, two rows of two landscape frames with the fifth frame laying over the top in the area of movement, so as to capture eye's and heads as one. The resulting capture size is about equivalent to 6x7 film.


P67 105/2.4 at f/2.4


P67 200/4 at f/4

-

I've been leaning heavily on Jim's excellent offering on 'Format size and image quality' to help guide me as I explore.

https://blog.kasson.com/the-last-word/format-size-and-image-quality/

 

[Erik Kaffehr]

Why would one bother?

Is there really any advantage gained from the extra effort and constraints imposed by the extended shooting period? (and the extra time in post.)

Are there tangible advantages, ok let's call them differences, that stand apart from greater resolution. Expressed another way, if these stitched images from a larger virtual sensor are downsized, does some essential quality from the larger capture remain? This may be direct or indirect, as in as a result of the lenses used or somewhat different exposure setting required to achieve a given image.

I don't have the answers - I'm just starting out exploring this path and am open minded about what I may find on the way.

I've no doubt this has been hashed out before, I wasn't there unfortunately. So I request any participant's indulgence, at least as far as offering some kind of photographic support to any expressed opinions.

-

Backstory:

Not too long ago I was primarily shooting adapted 135 film lenses on Fuji aps-c, and finding that stitching up to either 36x24 or 30x30 capture really let the lenses show their true worth. I was procrastinating over which full frame body to move to for adapted lenses as I really enjoy the Fuji shooting experience. As I procrastinated and many months passed I started to notice that older GFX bodies were coming up more and more affordably. The short story is that I purchased a 50R with the intent to simply shoot it with my 135 lenses and save all that stitching.

Solved. Happy. End of Story.

... however with the GFX in hand, I of course acquired a few film 645 lenses. Which put me back in a similar space as before, capturing just a portion of the offered image circle. The difference being that before I felt that I was just not quite reaching a level of image quality that I desired, and now I'm well and truly there, and more.

One does wonder though, if the leap from 24x16 to 44x33 impacted me in the way it did, what would a further leap up again bring? And I'm not just thinking final results in a technical way, certainly not simply looking at 'sharpness' or quantifiable qualities. I'm interested in the feel, the process and how that impacts the image, the larger experience, the journey.

-

I've been stitching a few ways so far - initially panning about the nodal point of the lens, although mostly more recently flat shift stitching, by which I mean leaving the lens fixed and moving the camera parallel to the sensor to effect the larger capture.

The easiest and quickest is with my Pentax 645 lenses on a Kipon shift adaptor. Two vertical frames side by side gives me pretty much exactly the old 645 frame size, maintaining the 4:3 aspect ratio. Here's a couple:

Note: I've downsized all the images to modest sizes because my personal search is not about greater resolution, I'm more than happy for others to focus on that though, it is certainly mind boggling the level of detail that is achievable.


P645 35/3.5 at f/5.6


P67 105/2.4 at f/2.4

-

Moving up in capture size: I'm waiting on the arrival of a Vertex stitching adaptor for Pentax 67 lenses to GFX, until that arrives I have a somewhat cumbersome arrangement that does work and conveniently allows me to shift as far as I want in any direction, so it is good for exploring the outer reaches of the image circle. The following two images are stitches of 5 images, two rows of two landscape frames with the fifth frame laying over the top in the area of movement, so as to capture eye's and heads as one. The resulting capture size is about equivalent to 6x7 film.


P67 105/2.4 at f/2.4


P67 200/4 at f/4

-

I've been leaning heavily on Jim's excellent offering on 'Format size and image quality' to help guide me as I explore.

https://blog.kasson.com/the-last-word/format-size-and-image-quality/

Thanks for sharing your images, interesting approach and great work!

If you shoot medium to long focal length, stitching is like using a larger format.

If you happen to have a lens with a large image circle, you can use make use of that image circle shifting the lens between exposures. An example, having a 24x36 mm camera in vertical position and applying +12/0/-12 mm of horizontal stitch will yield a 48x36 mm image.

Rotating the camera it is a bit different. You may get converging horisontals or verticals and the stitching program would need to correct for that.

View attachment 20881a05882c4ea8b1122af4b9cb811b.jpg

View attachment 7cabccabf75f4403831bf793aeeff0e7.jpg

View attachment 5d7d70c995e54d4bb5c4749c7b8eb92e.jpg

Check the images at original size!

Best regards

Erik

--
Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles

 

[JimKasson]

Once more I am amazed by your ability to stitch non-stationery subjects so well.

Jim

 

[JimKasson]

The ultimate in flat stitching is probably a scanning back like the Betterlights, which effectively stitch 8000 to 12000 exposures.

 

[Rob de Loe]

I like this path you're on Michael. It gets me thinking.

Using the techniques you're refining, you may be able to make photographs that look the way they do because of the way you made them. It may not be obvious to everyone; it could end up being one of those "there's something about this I can't pin down" situations.

There's always the chance that someone will say, "Yes but you could achieve the same result much more easily by doing X...". It might even be true. But I believe in process enough to know that it might not be true. Something about the way you make the piece will be imbued in the piece and shape its final form in ways that might not be easy to define, or replicate with a different technique.

Or maybe it really is a case of making it unnecessarily complicated for no real benefit! There's only one way to find out, which is to stay on the path.

 

[BobP]

Or maybe it's the type of projection that's used for the stitching. I've found cylindrical to be excellent for some forest scenes without producing unnatural stretched edges of rectilinear images. People say "they look like paintings" without being able to say why, perhaps this is how a painter's eye scans and assembles the scene in front of them.

Bob.

 

[Rob de Loe]

Or maybe it's the type of projection that's used for the stitching. I've found cylindrical to be excellent for some forest scenes without producing unnatural stretched edges of rectilinear images. People say "they look like paintings" without being able to say why, perhaps this is how a painter's eye scans and assembles the scene in front of them.

Bob.

That "feeling" in the viewer is what I'm after too with my approach to photography. I don't expect viewers (any) to know or care about how I did it. I just want some part of the enjoyment of the image, and the "success" of the image as a photograph, to trace back to how I did it. Technique is never enough of course, but I'm nerdy enough to enjoy knowing that sometimes it is.

One small nitpick to your point: I believe that with the kind of flat stitching Michael is doing, the stitching software is not using a projection. Others will know whether or not I'm right but I'm pretty sure that's the case. The reason is the lens is not moving in the process of creating the stitch images.

 

[JimKasson]

One small nitpick to your point: I believe that with the kind of flat stitching Michael is doing, the stitching software is not using a projection. Others will know whether or not I'm right but I'm pretty sure that's the case. The reason is the lens is not moving in the process of creating the stitch images.

All lenses perform some projection. A normal lens performs a rectilinear projection. Stitching programs can also do rectilinear projections. When you use slid-stitching with a rectilinear projection and stitch with a rectilinear projection, you get the field of view that the lens would have on a larger sensor.

 

[John Crowe]

I have been shift/stitching with a Canon 17 TS-E lens for ten years. First on a 5DIi and now a 5DSR. This year I have added a Kipon 645 Mamiya to EF Shift adapter, Mamiya 645 50mm Shift and Arsat 30mm fisheye. I have even rotated and stitched a FF circular fisheye on APS-C to get the best possible results at the time.

I am an ultrawide angle fanatic so it is not only about getting the large file size but also the much wider view. The 17 TS-E stitched to get the view of a 12mm lens at about an equivalent 85MP is far sharper than the Laowa 12mm f2.8 Zero-D at 50MP. Of course I have to use the Laowa when I simply cannot shift/stitch.

Similarly the 50mm Shift, shifted/stitched to 54mm x 68mm equivalent at 212MP, with a view of about a 24mm lens, is superior to just a 24mm lens on 24x36 at 50MP. It also happened to be a very affordable option.

I get awesome results with the Arsat as well since it keeps the subject larger in the frame while still getting down to about a 22mm view.

Both these lenses offer me options in the normal wide range.

All in all I am able to achieve similar results to medium format without the expense, and some loss of system flexibility.

Of course I consider a move to medium format, but I know I would be in the same boat as you continuing to shift/stitch, even though I am already really at the limit of my needs for printing 24" x 36".

I even made an attempt at making a 6x6 film body for my 17 TS-E.

I have come from a background of 6x6 and 4x5 film photography and yes, I do find the results of shifting/stitching worthwhile. I even do a few normal stitches with various normal and telephoto lenses, through the course of a year.

You are definitely on a rewarding track!

 

[Michael Floyd]

Thanks for sharing your images, interesting approach and great work!

If you shoot medium to long focal length, stitching is like using a larger format.

If you happen to have a lens with a large image circle, you can use make use of that image circle shifting the lens between exposures. An example, having a 24x36 mm camera in vertical position and applying +12/0/-12 mm of horizontal stitch will yield a 48x36 mm image.

Rotating the camera it is a bit different. You may get converging horisontals or verticals and the stitching program would need to correct for that.

Check the images at original size!

Best regards

Erik

Thanks for the reply Erik, thanks as well for the encouraging words.

Your images look great, I especially like the last one. Do you typically reserve stitching for times when you want a wider view than you are carrying, or do you sometimes stitch up even though you could frame your scene in a single shot?

 

[Michael Floyd]

Once more I am amazed by your ability to stitch non-stationery subjects so well.

Jim

Thanks Jim, I confess to being a little perplexed by your amazement, as I'm not doing anything special with the stitching at all.

For sure I'm taking care in the shooting, and adding an extra frame if I think the movement might cause problems. I'm also expecting each time to have to mask out the areas I do/do not want, although thus far Capture One's inbuilt and virtually optionless stitching feature is handling everything I've thrown at it. It even works out that I'm flat stitching and applies no stretching out of the corners.

 

[Michael Floyd]

I like this path you're on Michael. It gets me thinking.

Using the techniques you're refining, you may be able to make photographs that look the way they do because of the way you made them. It may not be obvious to everyone; it could end up being one of those "there's something about this I can't pin down" situations.

There's always the chance that someone will say, "Yes but you could achieve the same result much more easily by doing X...". It might even be true. But I believe in process enough to know that it might not be true. Something about the way you make the piece will be imbued in the piece and shape its final form in ways that might not be easy to define, or replicate with a different technique.

Or maybe it really is a case of making it unnecessarily complicated for no real benefit! There's only one way to find out, which is to stay on the path.

Thank you Rob, I really like and appreciate the way you've understood and then paraphrased my thinking in a more direct way.

I was expecting, and welcoming to, a bit more push back in the way that you outline, and perhaps that is yet to come, so it is quite emboldening to read things like 'can't pin down' and 'imbued'.

I'm as keen as anyone else to learn from the hard data and various specs and comparisons that are so expertly shared in this forum, it is a privilege to be welcomed into the wealth of experience here. At the same time, I personally need to hold on to the ability to sometimes throw all of that out the window and go with my gut and my heart. What you've written really helps me keep that hold, to not let the insecurity of relative inexperience overwhelm.

 

[Michael Floyd]

One small nitpick to your point: I believe that with the kind of flat stitching Michael is doing, the stitching software is not using a projection. Others will know whether or not I'm right but I'm pretty sure that's the case. The reason is the lens is not moving in the process of creating the stitch images.

This is correct. Capture One works it out automatically, Hugin requires me to jump through a few hoops, although is a vastly more feature rich stitcher. I daresay I'll move on up to one of the better regarded stitching programs at some stage.

 

[Michael Floyd]

I have been shift/stitching with a Canon 17 TS-E lens for ten years. First on a 5DIi and now a 5DSR. This year I have added a Kipon 645 Mamiya to EF Shift adapter, Mamiya 645 50mm Shift and Arsat 30mm fisheye. I have even rotated and stitched a FF circular fisheye on APS-C to get the best possible results at the time.

I am an ultrawide angle fanatic so it is not only about getting the large file size but also the much wider view. The 17 TS-E stitched to get the view of a 12mm lens at about an equivalent 85MP is far sharper than the Laowa 12mm f2.8 Zero-D at 50MP. Of course I have to use the Laowa when I simply cannot shift/stitch.

Similarly the 50mm Shift, shifted/stitched to 54mm x 68mm equivalent at 212MP, with a view of about a 24mm lens, is superior to just a 24mm lens on 24x36 at 50MP. It also happened to be a very affordable option.

I get awesome results with the Arsat as well since it keeps the subject larger in the frame while still getting down to about a 22mm view.

Both these lenses offer me options in the normal wide range.

All in all I am able to achieve similar results to medium format without the expense, and some loss of system flexibility.

Of course I consider a move to medium format, but I know I would be in the same boat as you continuing to shift/stitch, even though I am already really at the limit of my needs for printing 24" x 36".

I even made an attempt at making a 6x6 film body for my 17 TS-E.

I have come from a background of 6x6 and 4x5 film photography and yes, I do find the results of shifting/stitching worthwhile. I even do a few normal stitches with various normal and telephoto lenses, through the course of a year.

You are definitely on a rewarding track!

Hi John, thanks for the comment, much appreciated.

I am surprised that more landscape photographers do not do as you are. Their subjects are often quite static, they typically aim for great resolution and often have a taste for wide field's of view.

I've come across a few others who have come to a similar path after a history shooting 4x5 film, I wonder if there is something in that? It's perhaps easier to notice some subtle absence than to find the same quality for the first time.

I'd be keen to see an image or two if you have any that you're happy to share? Perhaps something from a normal or tele lens?

Thanks for the encouragement.

 

[Michael Floyd]

One small nitpick to your point: I believe that with the kind of flat stitching Michael is doing, the stitching software is not using a projection. Others will know whether or not I'm right but I'm pretty sure that's the case. The reason is the lens is not moving in the process of creating the stitch images.

All lenses perform some projection. A normal lens performs a rectilinear projection. Stitching programs can also do rectilinear projections. When you use slid-stitching with a rectilinear projection and stitch with a rectilinear projection, you get the field of view that the lens would have on a larger sensor.

Would it be correct in this instance to say that the lens is providing the rectilinear projection, and the stitching is merely not adding anything extra (nor taking anything away)?

If I set Hugin to 'rectilinear' when stitching one of these shift stitched images, the program further pulls out the already rectilinear (from the lens projection) corners too far such that straight lines become outwardly curved at the ends.

 

[Jones Longshot]

Why would one bother?

Is there really any advantage gained from the extra effort and constraints imposed by the extended shooting period? (and the extra time in post.)

Are there tangible advantages, ok let's call them differences, that stand apart from greater resolution. Expressed another way, if these stitched images from a larger virtual sensor are downsized, does some essential quality from the larger capture remain? This may be direct or indirect, as in as a result of the lenses used or somewhat different exposure setting required to achieve a given image.

I don't have the answers - I'm just starting out exploring this path and am open minded about what I may find on the way.

I've no doubt this has been hashed out before, I wasn't there unfortunately. So I request any participant's indulgence, at least as far as offering some kind of photographic support to any expressed opinions.

Michael,

My answer to your questions depends on further qualification. It's like asking if warmer is better than colder. There needs to be more context. Every type/genre of photography is going to have different criteria, but I'll offer my opinions from the perspective of someone attempting to make art and not other types of photography.

Ignoring the idea of equivalence for a moment. I find myself attracted to images made with long focal lengths on large sensors. Don't for get that 8x10 film is a sensor. There is a quality about them that just isn't there in images from small sensors. There are lots of arguments and lots of examples online trying to show there is no difference but it seems much like arguing the flavor of fresh vs frozen foods. Only the actual experience, tasting them for yourself, can fully inform you. I am one of those people that feels that photographs need to be experienced as final physical prints. Web presentations are incapable of conveying the full impact. You will never get the full experience of an Avedon print in an 8bit sRGB jpg or png on a web page.

I don't think of stitching as merely increased resolution. I think of stitching as a means to use longer, better corrected lenses, on a bigger sensor to capture more quality information for constructing better images. Having more and higher quality information at the start makes it much easier to create photographs that achieve my vision. Maybe instead of thinking about stitching as a technique, think of it as a mindset around photographing. A deliberate approach to seeing, analyzing, capturing and making photographs.

Starting with the obvious, there will be more pixels. I think of this increase as better spatial information. If each frame captures a smaller angle of view for the same resolution sensor and lens then texture/detail is easier to see. The better resolution becomes a more natural representation of texture in the final image even if we have to down sample for a particular purpose (web presentation). But we all know that.

The longer focal length lenses I use seem to be better corrected than the shorter wide FOV lenses. At least I think so. They tend to be sharper in the corners and don't have noticeable coma or distortion. I'm not sure I could defend this with math or formulas but If I break up a scene into 9 frames and stitch them together then the overall effect reduces the noticeable lens defects. Or maybe I have reduced the appearance/magnitude of these defects in relation to the entire image. This also gives new life to some older less spectacular lenses. I like to think of this as improving a lens by a factor of 3. No doubt imaging scientists could argue the fine points but it's a way of thinking about it that I find useful.

Sometimes stitching might just be needed to increase the field of view of whatever lens you have at the moment. Think about when you are backed up against a wall but need a wider lens to capture the scene. Or anytime you want to preserve the perspective but need a wider FOV. Stitching solves this pretty easily.

Don't forget that focus stacking and exposure bracketing are also stitching. Combining all of these techniques can create truly amazing photographs. I find the long toe and reciprocity characteristics of some films to be very hard to simulate with digital capture. Sometimes exposure bracketing helps with this.

However for me the real reason I stitch is that I'm looking to recreate the experience of shooting with a view camera. Using a piece of large format gear necessitates a different mindset and process. Everything is deliberate and generally more well thought out. You need to be more thoughtful about the actual placement of your camera. You need to more carefully consider aperture and DOF. You become more aware of the subject brightness range. Your brain starts to consider the interactions of all the tiny details and how each change you make affects the final image. I love this process. I love being slow and deliberate.

With a view camera you have to invert the image in your mind. I think this has a certain decoupling effect. It pushes the image through my brain in a way that separates it into it's technical and aesthetic qualities. The process of stitching lights up all the same things for me and allows me to consider the visual impact of the image completely separately from the technical and refine both.

The impractical side of stitching is that we need static subjects. I guess the view camera is similar in that regard too. There is a big advantage in single frame capture that way. For this reason I stayed far away from photographing people until just a few years ago. At some point I decided that I wanted to try to photograph people with the same approach as my other work. The process with people has many of the same side effects. I feel that the images are more deliberate and I'm happy with the results.

So yes, I do feel there are tangible advantages but only in the context of the type of images I make. I would never try to have this discussion with a BIF person.

Jones

 

[Michael Floyd]

Hi Jones!

I'm so pleased that you've commented here as your images are a great part of the inspiration for heading me in this particular direction.

You have written much for me to absorb, additionally my laptop just ran out of battery and will remain that way until the sun comes up to charge it tomorrow.

I will reply properly tomorrow evening from the keyboard, rather than poke away at this silly little phone screen I'm on at the moment.

Cheers, Michael

 

[John Crowe]

Here are a few samples. I will add a couple more when I get to my laptop.


Canon 17 TS-E, Canon 5D II.




Canon 17 TS-E, Canon 5DSR, foreground is focus stacked.


Canon 17 TS-E, Canon 5D II.


Canon 17 TS-E, Canon 5DSR, foreground is focus stacked.


Canon 17 TS-E, Canon 5DSR, foreground is focus stacked.

I use anywhere from three to nine images for the focus stacked foreground and then add the two shifted images. There are times when I have to focus stack each of the portions of the image, and then stitch them.

 

[JimKasson]

Those are great!

 

[Hamiltionian]

The ultimate in flat stitching is probably a scanning back like the Betterlights, which effectively stitch 8000 to 12000 exposures.

I think its a shame that there has never been a new version of the betterlight developed with modern electronics. It would still certainly be cumbersome and slow but presumably could still produce incredible output. I bet they could sell at least a ten of them.