Stitching Question - Digital SLR vs 4x5

[AlaBill]

I have followed the 2-3 threads in the last couple of days on the advantages of digital stitching to replicate using a 4x5 or 8x10 camera with film and have a couple of questions...

If the desire for a stitched digital photograph is increased resolution to compete with 4x5 film resolution when scanned, won't you have to use a different lens for the multiple digital captures compared to what would normally be used for a similar field of view?

For example, if I use a 180mm lens on a 4x5 camera I get 47° angle field of view (using the diagonal of the 4x5 film). For a single digital capture using a Nikon camera (1.5 crop factor) if I wanted a similar angle of view I would use a 35mm lens with gives me a 44° angle of view with the Nikon digital sensor. Pretty close to the 4x5 view. This is straightforward using the diagonals of both formats.

Now is I want to increase the resolution of the digital by stitching several frames together then I will have to change the lens to one giving a smaller angle of view than the normal lens in order to get the same angle of view with the total stitched image.

Is this thinking correct on stitching digital images for higher resolution. If you just pan left and right or even up and down using a normal lens giving a similar angle of view, then you would actually get a much wider field of view. Correct or am I off base here?

I do see that if I use a much smaller field of view lens for the stitched image then I can increase the resolution of that image. For instance, a 135mm lens on a Nikon digital sensor gives a 12° field of view which is about 1/4 the field of view using a 4x5 camera with a 180mm lens. So using an double row of 6 digital shots (6 on top and 6 on bottom) to allow an overlap of the images would be an increase in resolution from about 12MP to 170MP.

Am I off base with these calculations and the way to look at stitched digital photograph resolutions?

Bill Nelson

[Walter Calahan]

With digital stitching you can use whatever lens you want. The key is to have about a 30° overlap in the framing, and have the camera pivot on the nodal point of the lens. Then you can match whatever angle of view that a film camera can produce on a 4x5, 8x10, etc.

I simply do not let the comparison enter my head. I look at the scene I need to capture and use the tools at hand to do it. For me it's all about learning the gear so the choices become natural. This is how I work in 8x10, 4x5, medium or 35-mm format. I want my brain to know how each lens / camera combination sees before I look through the camera.

[Jack Flesher]

Bill, you are correct and it can be performed that way, though like Walter said, I usually don't think in terms of replicating 4x5 or 8x10 but rather in terms of capturing the image I want at higher or "high enough" resolution.

What this means in practice is that I do often stitch with longer than normal lenses -- in fact the 90 Tilt/Shift on my 5D is a common choice. If I stitched to cover the full area of a 4x5 neg, I would need roughly 20 captures (5hx4w) allowing for enough overlap to stitch. In practice, the final result would match the AoV of a 90mm lens on 4x5, but my final resolution quality (about 100 MP) would meet that of having used a 180 on scanned 8x10. (BTW, this is where the disadvantages of a digital scan back start to become advantages -- capturing 20 DSLR frames for a tiered stitch [rows and columns] is a non-trivial exercise and requires adherence to process; the scan back does it for you.)

Instead, I might capture 6 to 8 frames (2x3, 2x4) for about 50MP of final effective stitch and this AoV and resolution would more closely match say a 150 on scanned 4x5. Similarly used, my 45 Tilt/Shift might match a 75 on 4x5; my 24 an uber-wide 35 or 47 on 4x5. By adding my 2x converter to my 90 T/S, I maintain excellent optical performance and have what would amount to a 240 - 300 on 4x5. (IMO 6-8 DSLR captures is relatively easy to manage and is more convenient than using the scan back on a view camera.)

Lens tilt for altering PoF becomes an issue with the multiple stitch -- it can be done with a T/S on a DSLR, but is quite complex in a tiered stitch -- but the shorter lenses generating more DoF mitigate this to some degree, though not enough for all images. Thus, there are still some advantages to using the view camera to begin with ;)

Cheers,

[Kirk Gittings]

Here are some related articles. The first, in fact, is by Jack!
http://http://www.getdpi.com/stitch.html (http://www.getdpi.com/stitch.html)
http://www.outbackphoto.com/workflow/wf_58/essay.html

[Jack Flesher]

Here are some related articles. The first, in fact, is by Jack!
http://http://www.getdpi.com/stitch.html (http://www.getdpi.com/stitch.html)
http://www.outbackphoto.com/workflow/wf_58/essay.html

And I wrote that long before CS3 was available!

;),

[AlaBill]

Great articles!

Jack the flat stitch with the RRS rail looks to be really intriguing with a shift lens and quick as well but this approach changes the format of the photograph from the original to a slightly panoramic view. Correct?

If you wish to retain the original aspect ratio and only increase the resolution how would you do that?

Bill

[Dave Parker]

Okay,

I have worked with a few digital solutions over the years, so I may be completely wrong and I am not trying to start anything, just get some information....

My questions is if you take ten images with a 10 megapixel camera and stitch them together, how are you increasing the resolution? I have seen claims of say 100 mp? But even if you stitch 10 10mp images together, you have a larger 10 mp image, are you saying if you stitch these together your multiplying 10 x 10 to come up with 100mp? I don't understand, total over all resolution doesn't increase because you lay them end on end?

And like I said, I am really not trying to start anything, just trying to gather some information...

Thanks in advance for any help.

Dave

[Jack Flesher]

Great articles!

Jack the flat stitch with the RRS rail looks to be really intriguing with a shift lens and quick as well but this approach changes the format of the photograph from the original to a slightly panoramic view. Correct?

If you wish to retain the original aspect ratio and only increase the resolution how would you do that?

Bill

Yes, this method changes aspect ratio. If you stitch three horizontal frames you get essentially a 2:5 aspect ratio with a 2:3 sensor DSLR. If you stitch three frames horizontally with the camera placed vertically, you get about 3:4.

The only way to maintain aspect ratio would be to stitch multiple rows. The good news here is I don't find 2:3 a particularly rewarding aspect ratio to constrain myself to, so simple three-frame flat stitching as described in the above article adds some creative freedom ;)

Cheers,

[Gordon Moat]

The more common stitching creates a spherical view of a scene. An extreme example would be a straight road rendered as a curve. Imagine the effect that a mild fisheye lens would give you. Of course, with so much information, you could correct that in post processing. The other advantage over using a single shot and single lens is there is less light falloff at the edges of the image.

The other way to do this is with a stitching back mounted on a view camera. These are not very common items, and only a few companies make them. Most do not allow a large number of captures. If you have a view camera with a great deal of shift capability, then you could combine that with a stitching back to get many captures. The disadvantage of this approach is you would have even more equipment to drag on location. The main advantage is avoiding a spherical result.

Kaidan also make some interesting tripod mounts to make spherical panoramas easier. However, with some software packages you could literally just shoot handheld with the motordriven frames overlapping, then put that together in post. I have seen this done a few times, though obviously it involves more effort to get a smooth result. The best spherical solution would be a motorized tripod head that auto-indexes to each position, cutting down the overall time needed for capture.

I feel this is more a different approach than a substitute for a view camera. If you want to print really large images, stitching is a way to accomplish that. I know a local artist that uses a variation of stitching to create lenticular images, which is another approach; the results are 3D images that changes depending upon what angle you view them.

Ciao!

Gordon Moat
A G Studio (http://www.allgstudio.com)

[Jack Flesher]

Okay,

I have worked with a few digital solutions over the years, so I may be completely wrong and I am not trying to start anything, just get some information....

My questions is if you take ten images with a 10 megapixel camera and stitch them together, how are you increasing the resolution? I have seen claims of say 100 mp? But even if you stitch 10 10mp images together, you have a larger 10 mp image, are you saying if you stitch these together your multiplying 10 x 10 to come up with 100mp? I don't understand, total over all resolution doesn't increase because you lay them end on end?

And like I said, I am really not trying to start anything, just trying to gather some information...

Thanks in advance for any help.

Dave

Hi Dave:

I understand your question I think... The simple answer is you do lay the frames end-to-end (or rows x columns) and thus the resolution for each frame is additive to the total finished image. In a strict sense, ten 10MP laid end-to-end would generate a 100MP final image. However, since we do need some overlap to generate a good seam (read invisible), we lose say 30% to 50% of the overall image data on the stitch. So with ten 10MP frames we might end up with a 50 to 70 MP final image.

Hope this helps clarify the concept,

[QT Luong]

My questions is if you take ten images with a 10 megapixel camera and stitch them together, how are you increasing the resolution? I have seen claims of say 100 mp? But even if you stitch 10 10mp images together, you have a larger 10 mp image, are you saying if you stitch these together your multiplying 10 x 10 to come up with 100mp? I don't understand, total over all resolution doesn't increase because you lay them end on end?



The same way that when you shoot a 4x5 compared on shooting a 35mm, you get an image that you often call "higher resolution" when it has just higher information content.

[adrian tyler]

really interesing batch of threads on stitching, thanks to all. i expect that in the not-too-distant-future we may see a dslr with say a zoom lens that will work it all out for us... while we sit back and munch doughnuts!!

[chris jordan]

Dave, let's say you want a 40x50" print of a given scene, and you are going to use a DSLR camera to take the photo. You could make one exposure with a wide angle lens, and make the print from the resulting 16 Megapixel image (which would look terrible). Or you could put on a "normal" lens and shoot the bottom left quarter of the scene as one exposure, the top left quarter as another exposure, and so on until you had four exposures that when stitched together, would be a 64 megapixel file of the exact same scene as the original picture that was one exposure. If you put on a 200mm telephoto instead, and did the same thing, you would end up with a grid of 24-or-so images, which when all stitched together would add up to the same scene you started with, but at far higher resolution. Each of the exposures would only be a small part of your final 40x50" print.

For a cool example of this being done to insane image quality, check out this:

http://www.tawbaware.com/maxlyons/gigapixel.htm

That image could be printed at the full resolution of the Lightjet (305 dpi) at a size of 87x133", and every square inch of the print would be as sharp and detailed as the Lightjet has the ability to print. Unfortunately there is no Lightjet that prints that big (72" is the height limit), but you get the idea anyway.

My new work is all made this way too-- huge digital files assembled from thousands of smaller ones (sometimes with the help of moscaic software, but most of them I do by hand). If you look at my website you can see kind of what the results look like, although there is no substitute for seeing the real 6x8-foot prints.

[adrian tyler]

... but you gotta have a giga-brain to visualise a 24 grid image, i think that i would need some kind of harware support for anything over 9, gordon's comment on the spherical view too is interesting and means that there are really only certain types of images that are going to work...

until the auto stiching dslr is released...

[chris jordan]

Ade, I'm hoping for an automatic motorized stitching tripod head (which already exists), in conjunction with a video-robot moon-rover type of thing, so I can sit in front of my computer in my underwear guiding my camera while it drives around taking all my pictures for me. When it gets home from a long trip, I'll ask it "Did I have a good time?"

[Gordon Moat]

The same way that when you shoot a 150 lens on a 4x5 compared on shooting a 150 lens on 35mm, you get an image that you often call "higher resolution" when it has simpler higher information content.

When you compare to Chris Perez and others tests of 4x5, it seems about a real 60 lp/mm resolution is about it. Going by a few testers results with some 35mm film (i.e. Erwin Puts review of Fuji Astia 100F and Kodak E100G a few years ago), a real 80 lp/mm to 90 lp/mm seems to be a limit of colour films. So using a 150mm lens on a 35mm camera to cover the same FoV that a 150mm on a 4x5 would cover could result in greater resolution.

Using results from DPReview and other sites, the best of top level D-SLRs in terms of resolution is the Nikon D2X, with near 55 lp/mm to 60 lp/mm, depending upon which site and test you review. Remember that resolution is not file size. So if you simply had the same 150mm lens to stitch together a field of view to match a 150mm lens on a 4x5, then outright resolution would probably be quite close. However, some people will point out a different appearance to the digital capture, which could be perceived as a sharper image, making people think it is higher resolution . . . or comments such as cleaner image.

Notes: All these comparisons assume tripod mounted cameras with ideal conditions, and very good lenses; shoot handheld with anything and you can never approach these limits. Also, remember that test sites indicate highest contrast target resolutions. It is easier to compare high contrast results than medium contrast results, only because it can indicate outright resolution limits. While film can do quite well with high contrast real world imaging, some films do not do as well with medium contrast images, while some digital imaging sensors do better with low to medium contrast targets. Film also does well with details not in a perfect vertical or horizontal orientation, though to get that extra information can require scanning at much higher resolution than might be required for printing. Digital image sensors are oriented horizontally and vertically with their cell site orientation, and do worse with diagonally oriented information. Ease of use and convenience can be much better reasons to use digital cameras than outright performance; that can make one more consistant with their results, so it is important to consider those aspects.

To take advantage of stitching with a D-SLR, you would use a longer focal length lens, and even more frames to stitch together the final image. So imagine using a 300mm instead of a 150mm, requiring many more frames to be captured and stitched together. The overall FoV of the result might still match a 150mm on a 4x5, but the image would contain more information.

Ciao!

Gordon Moat
A G Studio (http://www.allgstudio.com)

[AlaBill]

Chris,

When you say you put these together by hand, just what do you mean? Do you take a large workspace in PS and then cut and paste from individual digital images to the much larger final working image size? When you use stitching, which do you use?

An finally one other question...

Have you seen any benefit from printing 16x20 or 20x24 prints at much higher resolutions, say 600dpi without any uprez, using the stitched image?

Bill

[chris jordan]

Hi Bill, yep that's the workflow of my current project. I cut and paste onto a large canvas from hundreds or even thousands of smaller images. I do my own kind of stitching that is my invention (that doesn't involve using any stitching software).

And to answer your second question, the way to get the best possible digital print quality is to feed the printer a file that is already ressed to the printer's native resolution. For the Epsons that is 360 dpi; for the Lightjet it is 305 dpi; I'm not sure what it is for the Chromira and other printers. If you send the printer a higher-resolution file, the printer has to down-res to its native resolution using its internal software, which is not as sophisiticated as the bicubic interpolation that Photoshop has, plus you can't see what it is doing so you don't get a chance to tweak it afterwards (such as doing some additional sharpening if necessary). So I always res my file to the printer's resolution, then do any sharpening and other tweaking that might be needed (noise reduction, etc.), and print from there. That's a pretty standard practice with digital printing so I don't think you'll find much disagreement on that point.

[adrian tyler]

chris, i have a contact in nasa, and he assures me that the ©moon-rover-digitar-stitcher is up there on mars, when it gets back i'm sure he'll let you take it out for a spin!

ade

[chris jordan]

Ade, I have heard that the Mars rover only captures a few hundred megabytes, which won't be sufficient for my next series of prints, which will be 40x50 miles in size, printed in limited editions of zero.

[adrian tyler]

now that's concept! i'll take it! could you gift wrap?

[Ralph Barker]

Hmmmm. 40x50 miles. I'm envisioning an aerial image of one of the Hawaiian islands, with the "print" being laid out over a stretch of New Mexico desert. Viewing would be by commercial airline (or, hot air balloon for the "grain snoops"). :cool:

[jim kitchen]

There is an interesting note about stitching software, located here:http://www.ptgui.com/ where they illustrate the process with a few examples:

jim k

[Struan Gray]

Borges has a story about a country that was mapped at 1:1.

The Blackbirds used to fly with a camera that moved film past a slit in sync with the ground speed. You end up with a negative that is a few inches by a few hundred feet long, and covers the whole flight path of the mission.

There were also the bomb damage assessment cameras flown on more normal planes, which spun a prism in front of a biogon for wide panoramic images on rollfilm.

But digital makes it all so much easier: you don't have to move the sensor, merely read it out. Simplifies things considerably.

At least one photographer makes truly large format images at 1:1. Fabio Sandri does contact prints of the floors of rooms using large rolls of B+W paper, stitching to get the required width. The images are a intriguing mix of light patterns from the lighting, shadows of any objects or furniture in the room, and the texture of the floor modulating the light reflected back through the paper. There's not much online, but a teeny weeeny jpg and a quote from Borges can be seen here:

http://www.kettlesyard.co.uk/exhibitions/archive/one1.html

[David Luttmann]

Okay,

I have worked with a few digital solutions over the years, so I may be completely wrong and I am not trying to start anything, just get some information....

My questions is if you take ten images with a 10 megapixel camera and stitch them together, how are you increasing the resolution? I have seen claims of say 100 mp? But even if you stitch 10 10mp images together, you have a larger 10 mp image, are you saying if you stitch these together your multiplying 10 x 10 to come up with 100mp? I don't understand, total over all resolution doesn't increase because you lay them end on end?

And like I said, I am really not trying to start anything, just trying to gather some information...

Thanks in advance for any help.

Dave


Now I see where you're coming from. It is interesting though David, that you were convinced of films superiority while yet not understanding what stitching really was.

[Dave Parker]

Now I see where you're coming from. It is interesting though David, that you were convinced of films superiority while yet not understanding what stitching really was.

Figured you would take a swipe at my question David, which is why I don't put much credence in most of what you say. 30 years of shooting for a living has taught me a few things, and I still don't feel digital is superior to film, even stitched images. The best digital images look cold and without depth to me. But as has been stated by a few, including me, that is my preference.

Dave

[adrian tyler]

like chris stated, it is impossible to tell the difference between the two. my digital prints indistiguishable from the same analogue ones.

why does everything have to boil down to "us" and "them", i happen to like film but i'm not gonna die for it, digital image capture can now clearly blow film out of the water, so what? how are you going to educate your children with this fundamentalist attitude, really is this such a big issue?

[Dave Parker]

My Children are grown and educated, so I don't have to worry about that, and for the most part, I don't understand why people care what someone else believes about digital or film, as I have previously stated, I don't care what someone else shoots, and I sure hope they don't care what I shoot. What I do find interesting though is how everybody keeps saying digital clearly blows film out of the water, then goes off because someone don't agree, both have their place and I use both, but I can tell the difference between them and choose to use film for my critical work, anyway, I don't feel that is a fundamentalist attitude because I feel film is better, if someone feels digital is better that is great...shoot what ya want...

Dave

[Marko]

My Children are grown and educated, so I don't have to worry about that, and for the most part, I don't understand why people care what someone else believes about digital or film, as I have previously stated, I don't care what someone else shoots, and I sure hope they don't care what I shoot. What I do find interesting though is how everybody keeps saying digital clearly blows film out of the water, then goes off because someone don't agree, both have their place and I use both, but I can tell the difference between them and choose to use film for my critical work, anyway, I don't feel that is a fundamentalist attitude because I feel film is better, if someone feels digital is better that is great...shoot what ya want...

Dave

That is not fundamentalist attitude, but jumping at every mention of digital and making disruptive comments with the only intent to derail or stop the discussion and insulting other people in the process certainly is.

No, it's not you I'm talking about, I pretty much agree with what you are saying although from a different perspective. There are a few on this board and we all know who they are precisely because they are so shrill about it.

[Stephen Willard]

First, I would like to note that SLR and DSLR stuff does not really belong here. This website is about LF cameras and not DLSR cameras. However, because you are trying to demonstrate that stitching is an alternative to LF photography with equivalent clarity of image, then it is appropriate.

In summary here is how I would characterize this method based on the discussions noted here and my own knowledge.

1. Stitching is a light weight alternative to LF photography resulting in comparable image clarity.

2. Stitching cannot mimic near-to-far focusing movements of the LF cameras. These types of movements allow the LF photographer to use more optimal aperture settings. Thus, stitching forces the DLSR camera to use less optimal apertures then the LF camera. This will reduce the clarity of image that can be achieved with stitching compared to LF photography.

3. Stitching is very slow because you must take multiple shots to produce a single image. Thus, stitching under conditions when light is changing very fast is most likely not practical. This also applies to conditions with lots of movement in the scene such as clouds or vegetation under windy conditions.

4. Setup time for stitching is slower then normal DLSR photography. Therefore, in general, setup time for both DSLR stitching and the LF camera is more or less the same.

5. If you do not employ near-to-far LF camera movements, then for a subject that is the same size in the final uncropped image, we can say that the DOF will be equivalent for both methods independent of the focal lengths of the lenses used to construct the image. I am basing this observation on the MACRO photography rule that states images with same size subjects will have the same DOF issues no matter if you use 300mm lens or 50mm to construct the image.

6. Because LF cameras can use color or b&w negative film, then the dynamic range of the resulting image will be significantly better.

Have I missed any thing? Are my conclusions and observations correct?:)

[adrian tyler]

7. the spherical distortion inherent in being in a fixed position and having to move the lens around on slightly different angles makes stiching less useful for some types of shots.

[adrian tyler]

8. one day some smarty-pants will come up with a camera that you frame, program the file size, and the thing will take one or 24 exposures accordingly... priced accordingly...

[Jack Flesher]

1) Yes

2) Well it can, but the software available to do DoF blending is at present in its relative infancy -- so call this feature available, but impractical for this application at present.

3) Not really. I can usually capture 6 frames in less than 30 seconds. However, scenes with lots of motion or where motion is the image are problematic or impossible; there is no way I can capture the power of a massive wave breaking over a pier if stitching.

4) Not really. I can set my DSLR up and have it ready to stitch in about a minute and compose, focus and meter for the stitch in another minute. I can get my view camera on the tripod and zeroed in about the same amount of time, but I need significantly more time to compose, focus and meter. Though respect others may be more accomplished with the view camera than I am ;)

5) Not really. But only because I don't need to do a 16-frame stitch with a 150mm lens to replicate a 150 on 4x5. Rather I can use a 90 and do 6 frames and end up with identical framing and the same or better image resolution than the 150 on 4x5. Hence, my DoF is extended due to using the shorter lens to begin with.

6) This point is always strongly debated, but my testing shows that the 5D generates over 8 full stops of usable DR -- more than color transparency emulsions. The expanded highlight shoulder of color neg lends the appearance of extended DR, but in reality if one uses an appropriate highlight shoulder adjustment curve in post-processing on the digital file, the effect can be duplicated. Digital then allows for added boost in the shadows that film cannot accommodate. However, I do believe B&W neg can handle a bit more total DR than direct digital capture.

Cheers,

[Armin Seeholzer]

"6. Because LF cameras can use color or b&w negative film, then the dynamic range of the resulting image will be significantly better.

Have I missed any thing? Are my conclusions and observations correct?"

Stephen you are almost right!

Also on digital you can impove the dynamic range with HDR and expand it over every film if you like to do so!
But it is also not working with something wich is moving!

It takes some other 5 years till digital is better then Film, in my opinion!
But it is f**** close now!

My 2 cents, Armin J. Seeholzer

[Marko]

First, I would like to note that SLR and DSLR stuff does not really belong here. This website is about LF cameras and not DLSR cameras.

There was a thread recently that introduced a view camera back (http://www.camerafusion.com/) aimed specifically at using a DSRL in combination with a view camera for stitching purposes.

So, in that sense and used in such a fashion, a DSLR should belong here as much as a (panoramic) roll film back.


2. Stitching cannot mimic near-to-far focusing movements of the LF cameras.

This is an inherently incorrect statement as stitching is a technique that is not camera specific. It can be done with any camera, in any format and with any technology. See above for an example.


5. If you do not employ near-to-far LF camera movements, then for a subject that is the same size in the final uncropped image, we can say that the DOF will be equivalent for both methods independent of the focal lengths of the lenses used to construct the image. I am basing this observation on the MACRO photography rule that states images with same size subjects will have the same DOF issues no matter if you use 300mm lens or 50mm to construct the image.

Even if the stitching were done using only a DSLR alone, there is nothing that would constrain it to a single plane. It can, and it is indeed routinely done in some applications, in all three planes of a scene. In fact, it is one of the most useful techniques for increasing DOF in macro photography (http://www.mplonsky.com/photo/), but it is only our imagination or lack thereof that might be preventing us from using the technique elsewhere.

Of course, since this technique can be practically done only using digital processing, it is available only to those who are not adverse to it.


6. Because LF cameras can use color or b&w negative film, then the dynamic range of the resulting image will be significantly better.

This is another incorrect statement. Using RAW digital capture and adequate processing, digital capture can yield dynamic range that no film can, regardless of fomat.

But let's not turn this into yet another digital vs. chemical shouting match, shall we? The type of camera and medium essentially has very little to do with this as a technique.

Speaking strictly about stitching and its possibilities, stitching can be used to achieve a dynamic range that no light-sensitive medium can normally obtain, film or digital.

[QT Luong]

7. the spherical distortion inherent in being in a fixed position and having to move the lens around on slightly different angles makes stiching less useful for some types of shots.

There is no such thing as "spherical distortion" involved. It's a linear perspective effect that is corrected by any stitching software while building the stitched view. The resulting image is no different than a single-view perpective image, should a perpective projection be chosen as the final rendering. Sometimes it looks strange if the resulting angle of view is very large, but so does a single very wide shot.

[Stephen Willard]

"Also on digital you can impove the dynamic range with HDR and expand it over every film if you like to do so!
But it is also not working with something wich is moving!

My 2 cents, Armin J. Seeholzer

Armin, we all give our two cents worth. Sometimes we are correct and sometimes we are not. However, when I add up all the two cents I have acquired on this sight, I find its close to a million dollars if you can keep a open mind, be considerate of others differences, and learn from it. I know I am a much better photographer due to the information shared on this website.

I am a color negative LF photographer, and I have no insight on how you can expand the dynamic range of the digital sensor, nor do I know how to actually do stitching. Could you please characterize this method of expanding the dynamic range in the same way I tried to characterize the pros and con of stitching for me?:confused:

[Gordon Moat]

Dynamic range usage can be a style choice, which is why some of us use transparency film. In fact, when I look through the recent Lürzer's Archive 200 Best Ad Photographers, the vast majority of images have a constrained dynamic range, and are somewhat high contrast. Obviously, this effect can also be done with digital capture and post processing. The tools should match what the photographer wants to accomplish.

Why does one method need to blow away another? Is film so bad a choice that people look down on prints made from film originals? Will people look down on film sourced prints in the future? Will those who switched to digital never again print their old images from film because they are so technically inferior? Will people stop buying books of photos or old prints because they know those photographers of the past did not have the benefits of digital capture?

Ciao!

Gordon Moat
A G Studio (http://www.allgstudio.com)

[adrian tyler]

There is no such thing as "spherical distortion" involved. It's a linear perspective effect that is corrected by any stitching software while building the stitched view. The resulting image is no different than a single-view perpective image, should a perpective projection be chosen as the final rendering. Sometimes it looks strange if the resulting angle of view is very large, but so does a single very wide shot.

so say a frontal view of a building framed on 150mm on my 4x5, will have the same appearence as the corresponding number of frames and focal lenghth of a 35mm dslr when stitched together?

now that is interesting!

[adrian tyler]

The more common stitching creates a spherical view of a scene. An extreme example would be a straight road rendered as a curve. Imagine the effect that a mild fisheye lens would give you. Of course, with so much information, you could correct that in post processing.

i had kind of visualised this effect, although i have to say that i have no experience of stitching but obiously the software seems more sophisticated than i thought. which is really interesting news.

[Gordon Moat]

i had kind of visualised this effect, although i have to say that i have no experience of stitching but obiously the software seems more sophisticated than i thought. which is really interesting news.

Apple's early QuickTime VR Studio was one of the earliest software packages that corrected the distortion. There is still that aspect of capturing shapes at the edges of your images that will seem slightly odd looking, something like what QT mentioned is similar to a very wide rectilinear. Obviously the effect is greater when the scene view is even wider. Current software packages provide some better solutions to this. The end result would still look different than a single shot on a view camera with the same field of view. When you are aware of this, it is possible to consider it in composition, and to correct later in post processing (as much as possible).

It might seem like trading one set of problems for another. There is a slight disconnect between composition, capture, and post processing. I think that becomes more seemless with practice and experience. This approach will always seem different than a single shot on large format, which is why I view this as an alternative and not a substitution.

I have some experience doing QuickTime VR images, which when you think about it is stitching. The difference is that usually a superwide or fisheye lens was used and rotated around the nodal point, though it would have been possible to use longer focal lengths. As for more normal stitching, I have done two, three or four shot film captures stitched together, though the positioning was done by shifting, and not by rotating the camera. A substitute method to simulate shift would be a sliding rail, sort of like a sideways macro position plate; this avoids the need for spherical correction later.

Ciao!

Gordon Moat
A G Studio (http://www.allgstudio.com)

[adrian tyler]

thanks gordon, yes, i can see that shifting would create a different result verticaly than rotating, and rotating will produce something rather like a normal "panorama", however a sliding rail would need to be mighty long to make an effective "travelling panorama" for a landscape say, no? (...sorry to be so slow on the uptake!!)

[David Luttmann]

Figured you would take a swipe at my question David, which is why I don't put much credence in most of what you say. 30 years of shooting for a living has taught me a few things, and I still don't feel digital is superior to film, even stitched images. The best digital images look cold and without depth to me. But as has been stated by a few, including me, that is my preference.

Dave

Not taking a swipe....you said that based on your experience, your opinion is that a stitched image cannot equal 4x5. Yet you then ask about how stitching works because you didn't understand. I'm sorry if you're feeling offended....but I am curious as to how you know stitching can't be better than film, yet you don't know what stitching is.

Simple question really. Or maybe you don't care that the quality can be better regardless....which is fine with me as well. Like you said, shot with what you want. I would add though that maybe in the future, if you don't know what something is, maybe you shouldn't be so quick to claim it inferior.

Hopefull this wasn't too shrill for anyone to digest ;-)

[Dave Parker]

Not taking a swipe....you said that based on your experience, your opinion is that a stitched image cannot equal 4x5. Yet you then ask about how stitching works because you didn't understand. I'm sorry if you're feeling offended....but I am curious as to how you know stitching can't be better than film, yet you don't know what stitching is.

Simple question really. Or maybe you don't care that the quality can be better regardless....which is fine with me as well. Like you said, shot with what you want. I would add though that maybe in the future, if you don't know what something is, maybe you shouldn't be so quick to claim it inferior.

Hopefull this wasn't too shrill for anyone to digest ;-)


David,

Where you are mistaken is, I do understand how stitching works, I didn't realize people were multiplying the total frames to come up with a MP number, I don't care how many frames you stitch together you still have the same bit depth in all the frames, your bit width may increase, but if you stitch 10, 10 megapixel frames together you still have a 10 megapixel bit depth.

But don't be mistaken, I do understand how to stitch frames together and actually have done it...I just happen to feel, differently about it. As far as equal or better, I still feel that 10 frames stitched together don't equal a properly scanned 4x5, but that is me...and I can tell the difference in a picture that I see in person, as far as looking at images on the net, I discount those due to the limitations of monitors and computers. But in the long run, it don't matter what I think at all does it, and about the only thing I really care about, is if my clients are happy with my work and pay me...

Dave

[Lazybones]

Say it, Dave. Say it...

[Sheldon N]

David,

Where you are mistaken is, I do understand how stitching works, I didn't realize people were multiplying the total frames to come up with a MP number, I don't care how many frames you stitch together you still have the same bit depth in all the frames, your bit width may increase, but if you stitch 10, 10 megapixel frames together you still have a 10 megapixel bit depth.

But don't be mistaken, I do understand how to stitch frames together and actually have done it...I just happen to feel, differently about it. As far as equal or better, I still feel that 10 frames stitched together don't equal a properly scanned 4x5, but that is me...and I can tell the difference in a picture that I see in person, as far as looking at images on the net, I discount those due to the limitations of monitors and computers.

Dave

For clarification sake..Bit depth is an entirely different issue than megapixels.

Bit Depth refers to the number of colors or shades of grey that an image can contain, basically how fine of a gradient you can create without showing steps from one shade/color to the next. A high bit depth image can display a perfectly smooth gradient with no perceptable steps, while a low bit depth image will show a gradient more like a step wedge.

Megapixels simply refers to the number of pixels contained in an image, derived from pixel width multiplied by pixel height.

There is no connection between the two. Bit depth is a function of the sensor & digital capture system built into the design of the camera

You are correct that a stiched image does not contain any greater bit depth than a single image. It does however contain more total pixels.

I believe we all understand and agree with your intended message - just because a digital stitched image is bigger and higher resolution than a single digital image, that does not take away any of the inherent disadvantages (or advantages) of the fact that it is a digital capture.

[Dave Parker]

Say it, Dave. Say it...

As I said, It don't matter what I think, I happen to feel the nuances of film are better, so what, I am just expressing my opinion as anyone else here does..

Dave

[Jack Flesher]

you still have a 10 megapixel bit depth.


Dave, I like you, respect you and respect your right to express your opinions here, so please understand this is not intended as a slam but rather added clarification...

The expression "10 megapixel bit depth" is entirely meaningless in the digital world... It is akin to me saying I don't like Tri-X because of its inferior dye transfer layers.

Pixels are the individual photon receptor sites. There are two significant things worthy of referencing when discussing them; the total number of them and their diameter, which in turn correlates to their density on the sensor. The more of them there are, the more information we have to render image detail over a given area. More pixels is the digital equivalent of using larger format film.

Bit depth is simply how many distinct gradations of color an individual pixel can render in each of the three primary color channels. 8-bits per channel = 24-bit color (3 primary x 8-bits for each) = 2^24 distinct colors being able to be rendered. The more bit-depth, the more colors we can replicate. This, when coupled with the definition of the working color space, would be reasonably equivalent to the color response curves or tonal range in film.

Dynamic range is the ability of the pixel to respond to and accurately render different intensities of light. This would be equivalent to exposure latitude in film.

Cheers,

[Dave Parker]

Dave, I like you, respect you and respect your right to express your opinions here.

However, FTR the expression "10 megapixel bit depth" is entirely meaningless in the digital world... It is akin to me saying I don't like Tri-X because of its inferior dye transfer layers.

Cheers,

Jack,

Sheldon, explained quite well what I was trying to express.....This is an argument that will only be settled when a true 4x5 sensor is available to the mainstream photographer. And we can compare results of that sensor to the stitched images...

One thing I do find interesting, the Sh*T sure hits the fan on photography websites, when either someone says "digital can blow away" or the other says "digital is inferior to film"

And as I continue to say, we each have our opinions and at this point in time, when someone tells me, digital is superior, I am going to disagree based on what I do, but and have to agree with an earlier post today..why does it have to be them or us..both sides could take a lesson here and stop with the this is better than that, cause us arguing is never going to change anyones mind.

Dave

[Jack Flesher]

Dave,

I am not "arguing" with anything you said, but rather discussing it... FTR, I have not claimed digital is superior to film, nor have I claimed film is superior to digital -- FTR2 I personally shoot both!

I simply felt it necessary to give some standard definitions so everybody starts speaking the same language -- only then can we all understand what it is we are trying to discuss when making comparisons.

If folks don't like what I wrote, by all means let the mods delete it.

[Dave Parker]

Dave,

I am not "arguing" with anything you said, but rather discussing it... FTR, I have not claimed digital is superior to film, nor have I claimed film is superior to digital -- FTR2 I personally shoot both!

I simply felt it necessary to give some standard definitions so everybody starts speaking the same language -- only then can we all understand what it is we are trying to discuss when making comparisons.

If folks don't like what I wrote, by all means let the mods delete it.

Jack,

No by all means, I was not addressing it to you specifically, I as well shoot both in my business, but there seems to be others that float around here and other sites...

But I do appriciate yours and Sheldon's posts on the subject, and I appreciated your explanation yesterday

I feel have a standard baseline is a good thing.

Dave

[adrian tyler]

my daddy's biger than your daddy!

[Gordon Moat]

thanks gordon, yes, i can see that shifting would create a different result verticaly than rotating, and rotating will produce something rather like a normal "panorama", however a sliding rail would need to be mighty long to make an effective "travelling panorama" for a landscape say, no? (...sorry to be so slow on the uptake!!)

Well, not nearly as long as might be imaged, though it would depend upon overlap and the FoV of the lens in use. Image if only a 20mm by 30mm image was captured in each frame of a stitch, from a 24mm by 36mm area (or sensor), then that would constrain movement to tighter than the FoV for each frame. However, use an even small area of capture, and less FoV, and many more frames could provide for a greater enlargement.

A simple comparison would be to imagine 20 frames from a 35mm (or D-SLR) stitched to match (roughly) the area of a piece of 4x5 film. Use a tigher FoV lens and capture 80 frames, then you could potentially print even larger. So depending upon the amount of overlap desired, the movement might be as small as 95mm horizontally by 125mm vertically. However, we are only considering matching a 4x5 FoV for a given focal length.

Ciao!

Gordon Moat
A G Studio (http://www.allgstudio.com)

[Dave Parker]

my daddy's biger than your daddy!

Well if so, you might want to learn how to spell "Bigger"

:D

[gregstidham]

Maybe I am missing something here. :) There is a relationship between focal length and film size. It is the main reason I shoot on LF.

It does not really matter what focal length I use. I would still need around 42 frames from my D2x to match the coverage of one sheet of 4x5 and maintain the perspective of 4x5.

D2x sensor = 23.3 x 15.5 mm
4x5 = 102 x 126mm

4x5 surface area = 12,852mm
D2x surface area = 361mm

Frames needed = 35, with frame overlap = 40 - 50 frames

I guess take lots of cards into the field when you shoot. :)

[Sheldon N]

I think what you state regarding the number of frames needed would hold true if you matched focal lengths on 4x5 and your D2x. If you went with a wider lens on your D2x like a 50mm, and used a 150mm lens on your 4x5, you could get away with fewer frames to match the field of view.

Matching perspective just takes maintaining the same angular field of view and camera/subject positioning.

I'd also guess that you'd find the 35-40 frame stitch with the D2x to exceed the resolution of a single 4x5 frame by a fair margin. That stitch would be a heck of a large file.

[Marko]

It does not really matter what focal length I use. I would still need around 42 frames from my D2x to match the coverage of one sheet of 4x5 and maintain the perspective of 4x5.

D2x sensor = 23.3 x 15.5 mm
4x5 = 102 x 126mm

4x5 surface area = 12,852mm
D2x surface area = 361mm

Frames needed = 35, with frame overlap = 40 - 50 frames

That would be true if one used the DSLR with one of those stitching backs recently introduced with the LF camera. It would not just maintain the perspective and FOV (and DOF), but would also provide at least as much information as film, if not more, depending on the sensor used.

Sheldon explained the rest.


I guess take lots of cards into the field when you shoot. :)

And leave lots of film holders at home. Sounds like a fair trade to me, for an occassional experiment. :)

[gregstidham]

I'd also guess that you'd find the 35-40 frame stitch with the D2x to exceed the resolution of a single 4x5 frame by a fair margin. That stitch would be a heck of a large file.
A large file indeed. The resolution benefits for some clients can't be denied IMO. It would then just become an artistic issue of what "look" I'm after for my work.

[Rob Landry]

Why does one method need to blow away another? Is film so bad a choice that people look down on prints made from film originals? Will people look down on film sourced prints in the future? Will those who switched to digital never again print their old images from film because they are so technically inferior?

Gordon, I too get the same impression reading these 'this vs. that' threads. Is film really that bad of a choice now?

Whether or not digital can blow away LF film should be irrelevant if you truly enjoy LF photography, which is what this site was supposed to be about, at least I thought it was until stitching fever took hold. For those of you that have traded or will trade all of your LF gear for a DSLR, by all means do so but please refrain from trying to convert the rest of us.

[gregstidham]

I've been working with digital photography since Photoshop 2.0 and I still think film is often the most elegant, simplest, and cost effective answer to a variety of photographic questions.

Stitching is just another tool. It is certainly not the best solution for everyone. New technology has always interested photographers.

IMO, use the tool that feeds your soul and drives your art.

[Doug Dolde]

I saw the funniest looking thing at Samy's Camera in LA Friday. It was a Leaf Aptus 75 mounted on a Mamiya RZ67. We are still in transition phase. A mix of the old and the new. Wait 20 years and see what film is worth.

[Gordon Moat]

. . . . . . Wait 20 years and see what film is worth.

I hope in that amount of time, the images I am producing now, and in the recent past, are seen as valuable by someone. When you consider images a few decades old selling well (or really well for better known photographers), I think there will always be interest in the past. I doubt people will dismiss the images of the past just due to having been made with technology of the past.

Oil paintings are hopelessly low resolution outdated technology, yet people still buy them. Maybe in 20 years the only film will be in an art supply store, and the only option might be self developing. I would imagine in such a situation that film would be expensive, just like oil painting already is today.

Ciao!

Gordon Moat
A G Studio (http://www.allgstudio.com)

[David A. Goldfarb]

Well, now that everyone is so enthusiastic about stitching, maybe the prices on 4x10", 7x17" and 8x20" cameras will drop through the floor. They were so 2005-06, anyway.

[Jack Flesher]

Well, now that everyone is so enthusiastic about stitching, maybe the prices on 4x10", 7x17" and 8x20" cameras will drop through the floor. They were so 2005-06, anyway.

If there is a valid "film versus digital" argument at present, this is possibly it. However the single large negative remains the leader for alternative process prints, though I guess a digital negative could be made from the stitched image...

Cheers,

[Howard Slavitt]

High quality stitches, especially for panoramic format, are very viable on the Canon 5D using Canon tilt shift lenses. I just got back from a two week trip in New Zealand. The vast majority of my "super" landscape images were made by stitching. My panoramic stitches involve just two captures, one with the lens shifted all the way to the left and one all the way to the right (and moving the camera slightly in the arca swiss tripod mount to correct for parallax error). First, when there is not much movement in the scene (clouds are easy, waves in the foreground are not), the stitches are a breeze. I do them manually in Photoshop. It takes me about 5 minutes per stitch. They're perfect. The panoramics will print up to 40" wide at close to 4 x 5 quality. I also did some stitches with the camera mounted vertically using 3 stitches (middle, left, right), to get an aspect ratio close to 4 x 5, and these will print at least to 30" on the large dimension, if not larger, at 4 x 5 quality. Second, I have been more successful than I'd expected doing perfect-looking stitches when there's movement in the frame, including waves in the mid to foreground. By being conscious of the movement in the captures, and trying to match captures as much as possible, and by using advanced selection, feathering, and blending techniques in Photoshop (I'm a fairly advanced PS user), I've been very pleased with the results. Third, the depth of field on these images, especially the panoramics, is amazing. In the first, place, I was using the 24mm and 45mm lenses, which, due to their short focal lengths, permit far reaching depth of field anyways. The Canon tilt shift lenses allow you to pan (shift) all the way in one direction, and then do some tilt to hold foreground focus. It's almost impossible to get the tilt right if you focus and tilt visually because of the small viewfinder image, but I did a fair number of tests in advance involving the more common types of compositions I gravitate towards, and have a "cheat sheet" that permits me to hold focus throughout the scene. For the most part, focusing at infinity or near infinity, and using about 1 degree of tilt on either lens, and shooting at f16 for the 24mm (otherwise the image isn't sharp to the edges when fully shifted) or f11 with the 45mm, permits me to get panoramics images with everything in focus from foreground to background. Sure it's a bit of a hassle, sometimes more of a hassle than others, but I never miss an exposure (using the histograms as verification), and get much better highlight and shadow detail than I obtained shooting slide film on a view camera. Also, no concerns about getting film abroad or having it ruined by xrays, and no film expense, and not needing to carry two camera systems (an SLR for travel shots and a view camera for landscapes). YMMV.

[steve simmons]

One of the questions for me is where do I want to spend my time - in some wonderful landscape location with a view camera or holed up in front of my computer.

Lets see

outdoors in the real world vs inside in front of a monitor

outdoors in the real world vs inside in front of a monitor

outside in the real world


teve simmons

[Marko]

One of the questions for me is where do I want to spend my time - in some wonderful landscape location with a view camera or holed up in front of my computer.

Steve, with all due respect, I think you're misplacing the point here. The time spent with your camera out in the field is the same, regardless of which capture technology you are using.

What really is different is the processing time. So, you should be asking yourself whether you prefer to spend the time in the darkroom or in front of your computer. The answer is individual and sujbjective - whatever works for you is the right answer for you, but it is probably different for me or the next person.

[Gordon Moat]

. . . . . So, you should be asking yourself whether you prefer to spend the time in the darkroom or in front of your computer. The answer is individual and sujbjective - whatever works for you is the right answer for you, but it is probably different for me or the next person.

Hello Marko,

I agree on the subjective aspect. However, it seems that often the comparison is darkroom or computer. That leaves out a third choice that some of us use: dropping off film at a lab. Another consideration amateurs and enthusiasts might miss, but professional consider, is minimizing post processing time. It is not always as simple as darkroom .vs. computer, especially for those of us without a darkroom.

To add a bit more to this, Nikon just did a pre-PMA announcement of a 10MP compact called the P5000 (http://www.dpreview.com/news/0702/07022006_nikonp5000.asp). When you go to the Nikon website, and read the brief information they provide, you discover there is an auto-stitching feature built into the camera. They also claim that 10MP allows nice 20" by 30" prints. Obviously many of us know that a 10MP compact is not like a 10MP D-SLR, but once you add that auto-stitching feature it seems to me some people are going to find that fairly close, or good enough. Maybe at some point in the future (soon) people who dumped 4x5 for D-SLRs and stitching to reduce weight, might go with compact digital and auto-stitching. At the expected price of a P5000, you could buy two for about what a D80 costs, arrange them on a board for maximum FoV, then get some even larger images to stitch, all in a very small package. Strange times we live in . . . . .

Ciao!

Gordon Moat
A G Studio (http://www.allgstudio.com)

[Marko]

Hey Gordon,

I agree with you about the third choice too, in general. But the talk here is about stitching and I somehow doubt a lab would do it for you, at least at a reasonble price and with the level of quality you need.

As for automated perspective stitching, I prefer to do it manually, but if I have to automate, I'd much rather leave it to Phothoshop on my computer than to OEM's software in camera. Two reasons for that - a) a stand alone computer is way more powerful than an on-board one and b) Ditto for Photoshop,OEMs are not really famous for the quality of their software. But the camera you mention does look interesting - on paper - for travel, hiking and such.

I'd rather call the times "interesting" than "strange", though.

[Gordon Moat]

I only had a lab do a stitch for me once, and that was for an RA-4 print. However, they did get the first attempt wrong, prompting a do over. Luckily the time delay still allowed me to drop off that print for exhibition. Yes, much better to control those stitching aspects than let automation handle it. Though with an as yet unreleased camera coming soon, I think it is too early for me to judge the results . . . I might be surprised.

On a practical level, I think four frames is about the limit of stitching I would want to do from scanned 4x5 transparencies. I have an architectural photographer friend who sometimes uses a 75mm and two overlapping frames to get a wider panoramic view, sometimes useful for across the street set-up shots. I think most mentions of D-SLR stitching I have seen involved much more than four frames, at least if the idea was truly to emulate (replace?) a one shot 4x5 image.

Decisions, decisions . . . okay, interesting it is . . . and sometimes curious.

Ciao!

Gordon Moat
A G Studio (http://www.allgstudio.com)

[Kirk Gittings]

Because I think it is an important point that I don't want to get buried in this thread, I am going to rip off Steve's post and make a new thread entitled "Photography a Desk Job?.

[Bill Eger]

Bravo! Photographs of the past are equally valuable to other forms of history. They also bring nuance that is difficult to capture with words. In my view, there is clearly different technology available every year but the eye of the photographer is the key to great work in any medium.