Large Format Digital Photography

Hi, I wrote an essay about my 2-year effort into my "DigiTiler".  Or, why anyone would want to do super high resolution photography through a large format lens.

https://medium.com/@maxrottersman/large-format-digital-photography-through-my-digitiler-8a94a1099379

Your justification there seems to be based on the idea that Bayer filters imply interpolation that disappears when you use 2x2 pixel blocks as single pixels... but that's wrong.

I believe I read your research a while ago. Thanks! Indeed, after I read some about these constraints (it might have even been your paper!) I did some tests and, though only casually scientific, proved to me that there was more resolution than I could get using state-of-the-art digital cameras. I've posted some of those comparisons. Like here

Digitiler: https://easyzoom.com/image/108502

36mpx Sony with 55mm Zeiss: https://easyzoom.com/image/108503

Perhaps one does need 16 pixels for every 1 good pixel.  I have no idea.  But I certainly believe there are limitations.  Anyway, at that point I decided whatever the limitations, two facts remained 1) there was increased resolution from the Digitiler and 2) I was getting that large format look I love.

My goal is to create camera for me, and any other large format photographers who are interested, to have a digital option that provides near the same feel as large format film.

In other words, my explanation is only meant to explain why someone might see that large format look.  I don't mean it to stand on its own.   In my experience, it's difficult to get into this stuff without being trolled.

You might get a kick out of my Large Format Digitalpage, which has a not-too-different discussion. I last updated it in 2013, so maybe it's getting to be time for me to say some more there...?

I'd be thrilled if you get into this stuff again!  I'm always open to talking about this stuff!

Incidentally, there is always LargeSense -- who market 4x5 and even 8x10 single-shot digital backs. However, they go the other direction, using very few, but very large, pixels. Then again, most of what they're doing is still B&W....

Did I forget to mention i don't have $100,000 lying around? HA HA.  I do mention LargeSense when I can.  I'd say we're both trying to do the same thing.  Obviously, if money is no object a LargeSense camera can do so much more than my thing because it doesn't take them 30 seconds to take multiple images.

NancyP wrote:

Large format has its own "look". Nice work, making an x-y rig that attaches to standard backs.

Hi Nancy.  Curious, do you have a large format setup?  A few days ago I sent back to creating a back for a Graflex (which will lead to another Cambo version).

maxotics wrote:

My goal is to create camera for me, and any other large format photographers who are interested, to have a digital option that provides near the same feel as large format film.

This was my goal as well, although I ended up going in a different direction than you did, as I am more focused on having in-camera movements available -- rise/fall/shift/swing/tilt -- than I am with creating huge files.

To that end, I modified a Toyo VX23D to accept an A7R on the back and various medium and 35mm-format lenses on the front. To make composing and focusing easier (as well as to further recreate the "large-format experience," because the monitor allows me to flip the image I see upside down and reverse it left-to-right and it's about the same size as the one I see on the ground glass of my 8x10), I also added a 12.5" external, 4K HDMI monitor:

Like you, I also take multiple files to create a single, final file, but instead of using a long lens and stitching the files together to increase effective resolution and create a wider FoV, I generally use wide-angle lenses and median-blend together between 7-17 files, depending on the circumstances and conditions, to reduce noise and increase the effective signal-to-noise ratio.

For my purposes, the A7R's 36MP files are usually large enough, but because I typically photograph late at night, using long exposures at base ISO, and often during the hotter times of year -- I live (and photograph) in the Phoenix metropolitan area, so during the summer months, the ambient temp is often 100 degrees or higher, even well past midnight -- excessive noise is a huge problem for me.

(So much so, in fact, that I'm even considering adapting a dedicated astrophotography camera with an actively cooled sensor to my terrestrial photography, but I'm reluctant to do so until I can try one first and determine its suitability for my purposes due to the added cost, weight, bulk and additional complexity this approach will entail...)

Surprisingly, perhaps, there are many 35mm format lenses that project oversize image circles sufficient to allow for rise/fall/shift movements of as much as 15mm when used with a 24x36 sensor. More typically, movements in the range of 4-8 are possible, but in most situations, these are sufficient for my purposes, because I'm using them solely to correct for geometric distortions, such as keystoning, and not to stitch together large files from lots of smaller ones.

Of course, I always have that option available simply by changing lenses, although with my camera, the process of capturing the original files will be quite a bit slower than with yours, as its rise/fall/shift movements are all finely geared. Horses for courses and all that...

You mention you're starting a new project using a Graflex camera ... I take this to mean that in addition to creating large files, you'll be adding at least front movements to your repertoire?

In any event, good luck with your project and I look forward to seeing your results from it!

P.S.: For those occasions when I need only rise/fall/shift movements, not tilt or swing, I also have a modified Cambo WDS that I use with the same A7R and lenses, as well as a Cambo Actus, which I use on those occasions when having a small and portable camera is the most important criterion...

I envy you your year-round warmth in Arizona.  Do you have comparisons of what you shot using your method and shooting natively through the A7R?  Sounds interesting!  I've thought about the same thing using pixel-shift on each frame (though I understand that's more of a trick to reduce bayer distortions).

I started with a Graflex back.  Then created a Cambo back (for full tilt-shift), then a robotic version, then a complete camera.  I've built 20-30 full systems by now.  Now I'm going back to the Graflex, then Cambo.  I'm trying to make them good enough to offer to other photographers.  It's a knuckle-scraping learning process in design and small-scale manufacture.  I now understand why so many large-format camera start-ups come and go   I've been on this project on and off for 2 1/2 years.  Maybe one day it will be fun again

Anyway, if all goes well, I might have a new back ready that would work on your camera if you're interested.  I need other photographers to help me think through some issues.  As you know, this is a complex process from start to finish, so the more one has experience in the overall issues the better chance of their success with my gadget.

maxotics wrote:

I envy you your year-round warmth in Arizona.

Yeah? Wait until August and let's see how you feel about our heat then ... lol.

Do you have comparisons of what you shot using your method and shooting natively through the A7R? Sounds interesting! I've thought about the same thing using pixel-shift on each frame (though I understand that's more of a trick to reduce bayer distortions).

Yes, pixel-shifting is something completely different.

Below are 100% excerpts taken from this photo, which involved a 30-second exposure at f8 and ISO 100:

Here's a single file, without median-blending:

And here's the same excerpt, except created using seven identical files blended together in CS6 using the median mode:

Note that the longer the exposure used, the greater the benefit from median-blending files; also, blending more files will reduce the noise even further, but I have found that seven files is a good compromise. If the scene is really dark, though, I will sometimes blend as many as 17 files (and I always use an odd number so the software doesn't have to average the values from two files to create a median value) but most of the time, I'm happy with the results from blending just seven files.

A good primer about this procedure can be read here. (Kasson used the mean mode, whereas I prefer the median mode because it eliminates most transient objects in the frame, including star trails, airplane and helicopter lights, etc. It also blurs any moving clouds, too!)

Oh and it also works for daylight photography, too. In fact, I'm presently working on a project that has me doing infrared photography and it's a huge help when it comes to reducing the noise inherent in those files!

Anyway, if all goes well, I might have a new back ready that would work on your camera if you're interested. I need other photographers to help me think through some issues. As you know, this is a complex process from start to finish, so the more one has experience in the overall issues the better chance of their success with my gadget.

I'm an inveterate tinkerer with all things electrical and mechanical, including cameras, so I'm happy to compare notes with you ... just drop me a PM and we can exchange contact info.

P.S.: In my experience, the A7R outperforms overall both the A7RII and A7RIII for this type of photography, as even with all the shortcomings of its files, it's possible for me to consistently recover more details in the shadow areas than I can using the 2nd and 3rd generation A7R camera bodies ... who knew?!

Interesting!  If only there were more time in the day!  Wait, you found time in the night   Will reach out at some point.  Thanks!

ProfHankD wrote:

AudiiDudii wrote:

To that end, I modified a Toyo VX23D to accept an A7R on the back and various medium and 35mm-format lenses on the front.

Very impressive rig!

Thanks!  It works very well for what I need it to do, but I do wish it was a bit smaller and lighter at times.

Apologies to Duke Ellington on my post title, but the problem I run into is that really big movements end up partially vignetting on the mouth of the FE mount. Any issues for you? Of course, it's much less of a problem on APS-C E-mount sensors.

Because I'm typically using shorter lenses (i.e., 16 mm to 50 mm, max), the amounts of tilt and swing I use are fairly small, hence I've never noticed any issues with vignetting.

Incidentally, the Nikon Z models should work much better for extreme tilt/shift.

True, but in my experience, BSI sensors don't seem to perform as well in the deep, dark shadow areas, all other things being equal.  I have also learned there's no substitute for real-world experience when it comes to evaluating the suitability of camera bodies and lenses for the type of photography I do, as I'm not aware of any reviewer who uses their cameras quite the same way I use mine.

Also, even if the Nikon Z7 does perform better overall than my A7R, the depth of its grip is problematic, as it's nearly 14 mm deeper than the A7R's grip, which will require me to significantly reconfigure my collection of FrankenKameras to accommodate it.  

maxotics wrote:

Interesting! If only there were more time in the day! Wait, you found time in the night

Indeed, I did. 

And while I did later come to enjoy photographing at night, at first, I did it because (thanks to my then very busy day job) it was pretty much the only time I consistently had available to do photography.

I'm interested in doing the same thing, mostly to gain movements but having indexed tiling could be something to experiment with.

I was thinking of a Horseman 985 or VH as it has front and back movements but in a 6x9 field camera format. I'd be using a Sony A7s. I feel very much like I'm jumping in the deep end so I'd appreciate any catches you foresee with that setup.

AudiiDudii,

There's no doubt that cooling the sensor in an astro camera makes a lot of difference but you'd then have to control the camera with a laptop which starts to make it a bit more awkward. Perhaps a better option for you would be a cooled standard camera from a place like CentralDS if you can afford the loss in back focus.

Steve.

I was thinking of a Horseman 985 or VH as it has front and back movements but in a 6x9 field camera format. I'd be using a Sony A7s. I feel very much like I'm jumping in the deep end so I'd appreciate any catches you foresee with that setup.

I believe that camera has a Graflok back so you might be able to use a Rhinocam, or something similar if you only one to take one, or a few images.

The problem with tilt/shift on all DSLRs in my experience is the flare you get as you move the image.  If you tilt up, you're essentially projecting most of the image to the side of whatever is between the lens and sensor.  If you have a shiny tube connecting to the camera, even if black, it can reflect that light onto the sensor.  For example, I used a Canon TS on my Sony A7 and if I changed the movement too much I'd get flare.  So I'd say that's the biggest issue you'll come up against.

BTW, you should check out Black 2.0 -- it's pretty amazing stuff at around 2.5% reflectivity with a very matte finish. The new Black 3.0 is supposed to be about half that reflectance! It's not Vanta black, but this stuff is about as optically dead as you can apply to most surfaces. I've painted insides of adapters, etc., with Black 2.0 very successfully.

Thanks!  I've looked at that but couldn't find any relevant reviews.  Now that you say it's good I'll buy it!  Right now I'm using black paint meant for leather.  No idea if it's any good.

I'm still making progress on my new Graflok back Digitiler.  Hopefully at some point I can provide some of these parts.

That's an aspect I hadn't considered. I'm glad I asked. Thanks.

Steve.

SteveInNZ wrote:

I'm interested in doing the same thing, mostly to gain movements but having indexed tiling could be something to experiment with.

I was thinking of a Horseman 985 or VH as it has front and back movements but in a 6x9 field camera format. I'd be using a Sony A7s. I feel very much like I'm jumping in the deep end so I'd appreciate any catches you foresee with that setup.

For a number of reasons, neither of those would be my first choice, but whether they'll work for you depends upon a number of things.

For example, what focal length range of lenses do you prefer to use?  Without a bag bellows, you'll be lucky to be able to use wide-angle lenses of any kind, due to the longish flange-to-focal distance view cameras typically have.

Also, when capturing images with a small-format sensor, the range of movements used will be quite a bit smaller, so greater precision is required than it was back in the film days and geared movements are helpful, too.

Better, I think, to buy one of the five used Toyo VX23DS present available on eBay and start with that or buy a Cambo Actus and skip the DIY bits altogether.  <shrugs>

There's no doubt that cooling the sensor in an astro camera makes a lot of difference but you'd then have to control the camera with a laptop which starts to make it a bit more awkward. Perhaps a better option for you would be a cooled standard camera from a place like CentralDS if you can afford the loss in back focus.

Alas, those cameras aren't an option for me, as they would require me to use at least medium-format lenses and even the shortest one of those is too long for much of the subject matter I photograph.  And I'm not sure I need a laptop, either, as a tablet should be able to do the job, but we'll see...