The Secret of the 16×20 Frame (DSLR Fine Art Reproduction)
May 4, 2012
Here’s the thing.
At 300ppi, the D3x frame, for example, measures out at about 13×20″. Even my little D5000, at 180ppi, figures out to around that. I’m going to just keep it simple, and standardize on a 16×20 frame across cameras and platforms, but here’s the point: Most DSLRs, today, will give you around 16×20″ at 300ppi, so for even the stodgiest (is that a word?) of skeptics, you can say that, at 16×20″ the camera is shooting at 100%.
Keep something in mind. Outside of the world of art reproduction, we routinely size a DSLR file up, often to a remarkable degree. With Bicubic Resampling in every version of Photoshop from CS3 on, it does a great job, and is pretty much a standard operating procedure. If you don’t agree with that, you can just turn the page now… but it’s well demonstrated. So even at 16×20″, the typical DSLR is actually not even pulling a sweat. If you print at 180ppi, then you get even more. (Again, well documented, if you don’t agree, fine, but all I can say is run your own tests and I’ll buy you a beer if I’m wrong.)
This is what I’ve been talking about in previous posts. If you understand how to work with a DSLR file properly, it’s more than a match for a tri-linear CCD device from Days of Yore.
So, given that, let’s say we want to reproduce art at 100%. We only have to shoot a 16×20″ frame. If you do it the way you’ve always done it, then you set up a big painting, light the whole thing, and then move the camera around, carefully keeping it square, and at the same distance, and then merge the frames together. If you do it using an x-y device, then it’s a different story. Using some sort of x-y easel, you can keep the camera stationary and consistent, and simply step and move the art. You only have to light one small area. The 16×20″ frame.
This is a lesson learned from my explorations with the Cruse Synchron system. The real secret to the Cruse is the fact that it takes a sample from a very small slice of the image at any given time. Using an x-y easel applies this to DSLR, and other, capture. See the graphic at the top of the page. That’s what we’re doing, except with a 16×20″ frame, not a narrow slit.
What exactly is an x-y easel? It’s a simple easel that allows you to move a platen up and down, and side to side – on an x and y axis – much like a CNC machine tool moves. It is precise and repeatable, and ultimately, pretty simple. The camera frame remains static. The work moves through the frame.
Stay tuned… more to come!
Ted Dillard 
June 2, 2012 at 11:03 pm
Very interesting concept. I’ve got a crude x-y easel below my copy stand. I use it with artworks that are beyond 4′ X 4′. The biggest problem I encounter is dealing with pictures stretched on warped stretcher bars. My shop is in Florida. Humidity has does a number on a lot of the art that comes into my shop.
May 7, 2013 at 8:58 am
I’ll disagree with some of this. I do a bit of reproduction of pen and ink drawings, with and without color added, and in my experiments of finding the optimum resolution for rendering detail *indistinguishable from the original* you need at least 600ppi and even 900ppi if you want some extra headroom for enlarging the art later. You want a final resolution very close to the rendering (not printing) resolution of your printer…..that means in the range of 600ppi (HP, Canon) to 720ppi (Epson).
For my 15mp APS-C sensor camera, this means tiling the art in roughly 4×6″ sections and then merging them. For most of the art I shoot, tiling the shots in a 4×4 grid is sufficient.
I would consider 400ppi to be the minimum for any good art reproduction. For a 16×20″ this translates to about 50mp! Assuming you agree that using the full corner-to-corner area of the sensor is a bad idea and that using perhaps 75% of the sensor area is safer, then you’d need closer to a 100mp sensor to capture 16×20 in a single shot.
Just sayin’
And where can a guy find a good x/y easel for photographing art? I’ve looked all over. 🙂
May 7, 2013 at 3:20 pm
Well, we all have to exorcize our OCD demons somehow… lol!
As far as the printer resolution goes, you’re making a fairly common mistake, in trying to connect the 720/1440 “dpi” claims of the Epsons, for example, to the ppi resolution of the file. The fact is, there’s no correlation. The pixel dimensions of the file simply do not translate in any meaningful way to the dithering of the printer, in spite of common misinformation.
But fine… if you want more resolution, go for it. I’m using an old-school Nikkor 55mm Micro 3.5 and a 105 Micro 4.0, with virtually no edge resolution loss, so, no, I don’t go with the 75% idea your suggesting. But I do have a client who likes more resolution, so we simply work with an 11×14″ frame (with the D800).
I’ve got the X-Y easel into “almost-production” mode, so stay tuned!
May 7, 2013 at 9:07 pm
If you re-read what I wrote, I said *rendering* resolution, not printing resolution….I fully understand that the printing resolution (720/1440/2880) has little to do with image resolution but it does affect the smoothness of the screening/dithering…so at least indirectly it can have an affect on perceived image detail/sharpness.
As far as optimum image resolution, if you chose the correct driver settings, you can indeed render the image at 720dpi and not 360dpi like most people have been told. In my own (and others) tests, there is indeed an increase in rendered detail up to about 480ppi on an Epson and about 400ppi for HP/Canon. Depending on the art (I’m doing pen & ink with fine detail and stippling), the differences in detail is not just academic. I suppose for normal paintings the differences wouldn’t be all that apparent.
Anyway, get that X-Y easel finished and produce some plans and we can continue to argue about this over a few beers. 🙂
May 8, 2013 at 5:54 am
So, OK, “rendering resolution”? Give us some links so we can understand exactly what you mean by that… because I’m not sure.
mmmmm. beeer.
May 8, 2013 at 9:00 am
By “rendering resolution” I simply mean the resolution the driver uses to process the image…which is separate from the printing resolution. In an Epson, the resolution used to up- or down-sample an image is generally 360ppi or 720ppi depending on driver settings.
May 9, 2013 at 5:47 am
So. Regarding the above resolution discussion. Welcome to the bottomless pit of “printer resolution”. Yes. There’s image resolution, “rendering resolution”, and printer resolution…
Image resolution is the “ppi”, or pixels per inch of your Photoshop image. Call it “dpi” and I’ll hang up on you.
The printer driver takes those little squares of information and turns them into a droplet pattern using math. That’s what this reader is calling “rendering resolution”.
Then you have the resolution of the droplets of ink on the paper. That is the “720/1440 dpi” spec that the printer companies tell you. It’s not entirely accurate.
The bottom line is, you can (in a practical sense) control only one of these. The resolution of the image in Photoshop.*
What resolution should you use for this? There’s only one answer to that question. The resolution that you find to be best. How do you figure that out? RUN YOUR OWN TESTS. No amount of reading, armchair punditing, or speculating about the math involved in the drivers and algorithms is going to answer that question for you. Make two prints. Look at them. If you’re blind, then show them to someone more critical than you.
Print them at 300ppi. How about 240, 360 or some variant of the 720/1440 number that’s the “wisdom” that’s been pushed since the early 2000s? How about going as low as you can… try 180ppp for yucks? Label them, sleep on it and then go back and see if you can see a difference. Try different papers. You’re going to find you can see differences on one paper that you won’t see on another.
Then. Print with the absolute lowest resolution you need to be able to hold the quality you can see. Why? Why not? (And yes, I’ve seen quality actually degrade from sending a file with resolution that’s way too high.)
Why am I a little strident on this point? Because I’ve read almost a decade of “wisdom” about resolution theory, and it’s got very little to do with what I’ve actually seen off the printer. …over a decade of what I’ve seen off the printer, through every driver and RIP combination available.
Why is resolution a favorite geek-out? I have no idea at all. But my final word is, whatever makes you happy.
The 16×20 frame, with a camera like the D800, is a convenient and effective dimension based on the 100% size of the image I produce, for me, and my clients. It’s actually, as I’ve said above, pretty impressive for even my cheap little D5000. I have one client for whom this is not enough, so we shoot that work with the same camera and lens at an 11×14 frame. It gives us a barely manageable (huge) file size, and a final image he has to scale down to print at life-size, and at that, even more final image resolution than I feel is needed. But it makes him happy.
And that, of course, is the beauty of this whole idea. You want more or less resolution? Move the camera closer, or further away, and take more samples.
Ted
*Yes. I’m fully aware that you can control the RIP, (for the “rendering resolution”) if you’re using a RIP and the printer drivers (for the “printer resolution), but you’ll really have to prove to me that you can do a better job than a good RIP at making those decisions… and the simple choice of 1440 or 2880 is something you’ve already tested.
June 19, 2013 at 5:46 pm
Thanks for the illuminating discussion! I’m using similar methods with my Canon 1Ds3, but what I’m missing is a workable x-y easel, do you have yours in production yet?
Ken
June 19, 2013 at 6:06 pm
Almost! We have a portable, working prototype, and a full-sized easel in development. In a few months I should be able to offer something.