DSLRs, big prints, and digital workflow

[dzeanah]

I've been shooting weddings with Fuji DSLR gear, which works beautifully with the wide variety of lighting I might encounter.  Great in full sun for instance -- about as good as film.  The trade-off is that half of its 12 megapixels are digicam sized and only used for highlights (and prevention of blowouts), and the orientation of the sensor is 45 degrees off from what all other Bayer sensors do, so I doubt the raw converters are optimized for the Fuji cameras even if they support them.

With that said, I'm now taking on a new market that requires biggish prints (up to 20x30 seems to be considered "huge" by non-photographers, and that's likely as large as I'll go) so I'm looking at options to print at that size.  The only time I've printed big from the Fuji was 3-4 years ago, and it was a stretch to get what I thought was reasonable sharpness out of a 16x20.

But that was 3-4 years ago, from what's essentially a 6 megapixel camera, and I really had no idea about digital processing at the time (I don't have much of a clue now either, but I'm working on correcting that.)  I've found photographers that I recognize from online groups a decade ago selling prints online at sizes up to 40-45 inches wide from 12 megapixel cameras.  I've seen comments from others that suggest that the difference between an A700 and an A900 at 30x40 is slight if you set up a double-blind test.  And now I'm really confused.

If true, then I can use my current kit + a D300s and call it good -- I've already got the lenses and support gear, and I already know the interface.  The only problem will be developing a reasonable workflow to maximize the image potential.  If false, then I'm setting myself up for failure.

I come here knowing that a significant percentage of the participants will claim that a 25 megapixel DSLR is marginal for anything larger than 11x14 (slight exaggeration here, but the standards of readers on this board are quite high.)  My target audience isn't photographers, though -- they won't approach a 20x30 and view it from 6 inches away before expressing an opinion on it.  They'll simply see it from a normal viewing distance and react to it emotionally (hopefully -- the "emotion" part depends on me.)

-----

Anyway, with that long preamble out of the way, what I'm looking to do is download some images (either processed if someone wants to donate one they think holds up well to enlargement, or a sample RAW file I find online and do my best efforts on), do whatever work is required to get them to print well large, and have some prints made so I can judge for myself to determine what level of gear is appropriate.  My wife and my wallet are hoping on Nikon DX sized sensors, but I have my doubts.

The problem is that I know my post-processing skills suck.  Lightroom I'm comfortable with, and I own Photoshop CS3 and can work with it if I need to, though my level of comfort there is weak.  If I were to download an image (or borrow a camera, lock it down on a tripod, focus well, lock up the mirror, and take some shots) I wouldn't know how to process the thing to maximize image quality on large prints.

Help?  Are there any online guides I can look at to guide me?  I'm not looking for anyone's "secret sauce" here -- I'm just trying to determine what a reasonable enlargement from XXXX camera is given talented post-processing and judged by my own eyes.  This gives me something to strive toward (better post- skills), but it's going to be tough to make the initial evaluation at my current skill level...

[Alan Goldhammer]

The problem is that I know my post-processing skills suck.  Lightroom I'm comfortable with, and I own Photoshop CS3 and can work with it if I need to, though my level of comfort there is weak.  If I were to download an image (or borrow a camera, lock it down on a tripod, focus well, lock up the mirror, and take some shots) I wouldn't know how to process the thing to maximize image quality on large prints.

Help?  Are there any online guides I can look at to guide me?  I'm not looking for anyone's "secret sauce" here -- I'm just trying to determine what a reasonable enlargement from XXXX camera is given talented post-processing and judged by my own eyes.  This gives me something to strive toward (better post- skills), but it's going to be tough to make the initial evaluation at my current skill level...

This link may prove useful in determining how big you can print.

Alan

[JeffKohn]

It really varies depending on image content, standards for print quality, and a host of other things like lens quality and shooting technique. My general rule of thumb is to have at least 175ppi before any interpolation. More is better, but with careful processing I can get good prints from 175ppi for most images (but maybe not all).

So for 12mp, 16x24 was my limit without stitching; for 24mp, up to 22x33. You can go higher for images that don't have lots of fine detail in them, but I'm a landscape guy so detail is often important.  I wouldn't go to 20x30 from 12mp for most images without stitching.

Another factor is viewing distance. Some argue that people view larger prints from further back,  so you don't need as many PPI. I think that's only partly true, as people will step up for a closer look if an image interests them. If you can control the viewing distance (eg it's not possible to step closer), then you might have a little more leeway here.

[dzeanah]

This link may prove useful in determining how big you can print.

Alan

I've seen that link, and that pegs the quality the A900 can produce at 24x36 as "poor."  Compare that with this comment from Mike Johnston, from this link:

The Sony A850 should be thought of as a medium-format digital camera in a conventional SLR form-factor body. It's perfectly adequate as an all-around SLR, but there's one area in which it excels. Above, I'm holding a 20 x 30" test print made from one of my A900 test shots by Ctein on his wide-format printer (bigger than I can make at home). I think the quality is about what I might expect from a 35mm color negative enlarged to 6x9". Ctein says he thinks it looks better than a 20x30 print from 6x7 cm film. I just can't see a 30"-wide print needing to look better, for almost any conceivable application. This is where the A900 and its cheaper brother, the A850,  really come into their own. If you want or need to make really big prints, even occasionally, and you don't want to spend a fortune for the capability, then this is the axe for you.

Then there's this guy:

If you have a 12 megapixel DSLR (Nikon D300 for example), the largest decent quality print size you can expect will be roughly 19 x 28.5 at 150 dpi.

The problem is that they're all probably right, depending on their evaluation of the quality of the prints in question and the intended purpose of those prints.  Now, I've got one of Ctein's books and I respect the hell outta that guy and his dedication to printmaking -- in fact, I'd guess he probably evaluates prints at a higher standard than any of my clients do, which means his comments about 20x30's are pretty conservative.  Thom Hogan seems to know his craft as well, but he flags a print 6" wider than the one Ctein praised as "poor."  

So either there's some bright line between 30 and 36" wide prints, there's a huge discrepancy between the post-processing ability of the two printers in question, or we're looking at different evaluations of comparable prints using different standards of quality.

Hence the reason I'm trying to do this myself, even with my weak skills.  At the very least I can show the print around to people who are my intended audience and say "what do you think?"  For the most part I think my eye is a bit less critical than most of yours -- I've printed 40x60 from 645 negs that look great to me, but many would claim this is LF territory.

This should be an easy decision.  Maybe I'm making it too complex?

[Wolfman]

This page from an article about uprezing by Jeff Schewe might be of help: http://www.digitalphotopro.com/technique/s...es.html?start=3

[dzeanah]

This page from an article about uprezing by Jeff Schewe might be of help: http://www.digitalphotopro.com/technique/s...es.html?start=3

Thanks.

You know, B&W work in the darkroom sure seemed easier...

[JeffKohn]

So either there's some bright line between 30 and 36" wide prints, there's a huge discrepancy between the post-processing ability of the two printers in question, or we're looking at different evaluations of comparable prints using different standards of quality.

I think there's a bigger difference between 20x30 and 24x36 than maybe you realize, it's the difference between 200ppi and 166ppi before interpolation, which straddles that 175-180ppi threshold that you'll see a lot of people mention.  I can tell you from experience that 40ppi difference is noticeable in the final print, unless you're viewing from across the room.

One problem with Thom's chart is the big gap between 13x19 and 24x36. Also, keep in mind his numbers in that first table are pre-interpolation, but he then goes on to say you can upsize by 100%. So to me his advice is a little wish-washy.

[dzeanah]

I think there's a bigger difference between 20x30 and 24x36 than maybe you realize, it's the difference between 200ppi and 166ppi before interpolation, which straddles that 175-180ppi threshold that you'll see a lot of people mention. I can tell you from experience that 40ppi difference is noticeable in the final print, unless you're viewing from across the room.

That may very well be.

The good news is this: I've found a NEF file online of reasonable quality that I ran through Lightroom and sent out to be printed at 20x30.  So Real Soon Now I'll be able to evaluate a photo from a decent 12 megapixel camera at the size I'm worried about and make a determination.  

I still think that whatever quality I eeked out of the file is less than it's capable of, but this gives me a solid (if low) starting point.

I'll post again once I know more.

[ErikKaffehr]

Hi,

My experience was that A700 and A900 images can be hard to tell apart when printed in A2 (16.5 × 23.4").

http://www.pbase.com/ekr/image/107619976/original (upscaled to same size)

This printed and scanned at 300 PPI

http://www.pbase.com/ekr/image/107823207/original

Viewers could not tell the prints apart.

That said, I repeated the test with different images and sometimes the A900 had an edge in print, but always much less than what would be expected from 1:1 view of the files.

BR
Erik

I've been shooting weddings with Fuji DSLR gear, which works beautifully with the wide variety of lighting I might encounter.  Great in full sun for instance -- about as good as film.  The trade-off is that half of its 12 megapixels are digicam sized and only used for highlights (and prevention of blowouts), and the orientation of the sensor is 45 degrees off from what all other Bayer sensors do, so I doubt the raw converters are optimized for the Fuji cameras even if they support them.

With that said, I'm now taking on a new market that requires biggish prints (up to 20x30 seems to be considered "huge" by non-photographers, and that's likely as large as I'll go) so I'm looking at options to print at that size.  The only time I've printed big from the Fuji was 3-4 years ago, and it was a stretch to get what I thought was reasonable sharpness out of a 16x20.

But that was 3-4 years ago, from what's essentially a 6 megapixel camera, and I really had no idea about digital processing at the time (I don't have much of a clue now either, but I'm working on correcting that.)  I've found photographers that I recognize from online groups a decade ago selling prints online at sizes up to 40-45 inches wide from 12 megapixel cameras.  I've seen comments from others that suggest that the difference between an A700 and an A900 at 30x40 is slight if you set up a double-blind test.  And now I'm really confused.

If true, then I can use my current kit + a D300s and call it good -- I've already got the lenses and support gear, and I already know the interface.  The only problem will be developing a reasonable workflow to maximize the image potential.  If false, then I'm setting myself up for failure.

I come here knowing that a significant percentage of the participants will claim that a 25 megapixel DSLR is marginal for anything larger than 11x14 (slight exaggeration here, but the standards of readers on this board are quite high.)  My target audience isn't photographers, though -- they won't approach a 20x30 and view it from 6 inches away before expressing an opinion on it.  They'll simply see it from a normal viewing distance and react to it emotionally (hopefully -- the "emotion" part depends on me.)

-----

Anyway, with that long preamble out of the way, what I'm looking to do is download some images (either processed if someone wants to donate one they think holds up well to enlargement, or a sample RAW file I find online and do my best efforts on), do whatever work is required to get them to print well large, and have some prints made so I can judge for myself to determine what level of gear is appropriate.  My wife and my wallet are hoping on Nikon DX sized sensors, but I have my doubts.

The problem is that I know my post-processing skills suck.  Lightroom I'm comfortable with, and I own Photoshop CS3 and can work with it if I need to, though my level of comfort there is weak.  If I were to download an image (or borrow a camera, lock it down on a tripod, focus well, lock up the mirror, and take some shots) I wouldn't know how to process the thing to maximize image quality on large prints.

Help?  Are there any online guides I can look at to guide me?  I'm not looking for anyone's "secret sauce" here -- I'm just trying to determine what a reasonable enlargement from XXXX camera is given talented post-processing and judged by my own eyes.  This gives me something to strive toward (better post- skills), but it's going to be tough to make the initial evaluation at my current skill level...

[dzeanah]

OK, time for a follow-up.  If nothing else someone may search to find this in the future...

I ended up printing two images last week and having them delivered.  The goal was to see some output that I could judge myself using my current skill level with postprocessing.  My expectation is that this represents the lowest quality level I'll see; as I learn more about digital post-processing I expect my results to get better.

So, here's what I got:

Print One: 40 inches from a Fuji S2 Pro
My wife has been wanting to print this for our wall for a while, so I figured now was the time.  This is essentially a 6 megapixel camera so I didn't expect much.

Processing was all done in Lightroom, and the output was to use Lightroom to print to JPEG, printing at 250 ppi (the resolution of my lab's printers) using the built-in Lightroom interpolation, with output sharpening at its lowest setting.  The print was created by Miller's Lab which is a portrait outfit I've been using for a dozen years, but what the hell.  I don't have any recent experience with anyone else, and everyone else I've used (notably Ivey Seright) has been for film-based printing so there's no direct comparison anyway.

The image was this one:



Output looks just fine at normal viewing distance (defined as "far enough away that you can see the entire print without using your peripheral vision.)  I'd like to be able to move closer and see more of the print, like the waves that were cresting, but that's just not possible.  Moving really close does show some haloing on the left-hand edge of the clump of rocks on the mid-ground, but from a normal distance it looks crisp.  The wife is thrilled.

That shocked the hell out of me, to be honest.  Especially since it was shot hand-held on the side of the road using a consumer-grade 100-300mm zoom lens.  Defining your viewing distance impacts this a whole lot more than I expected it would, I suppose.

Print Two: 20x30 from a D300s NEF File

I downloaded a NEF file that DPReview included in their review of the camera from here and spent about 4 minutes in lightroom with it.  I kicked it out of lightroom the same way, with a 250 ppi print to JPEG which I uploaded to MPix.  Mpix ~= Miller's so I figured it was a reasonable way to go, plus it was reasonably cheap.

The image is here:



Output looks good from a "reasonable viewing distance" again.  I can't really say much about the image -- I know it was shot at ISO 200, f8 @ 22mm and 1/50s, but that's it.  No idea as to lens, focal point, tripod or VR, whether we would see more detail in the grass if the wind hadn't been blowing, etc.  It's just an image, and I am trying to evaluate it as such.

This makes it hard to evaluate sharpness, though.  Hyperfocal distance was (probably) used, but how much does this impact sharpness at 20x30?  How much sharper would it be if we had a focal point in the image, rather than "most everything?"  

Overall I think the 20x30 is sufficient for my purposes.  I'd like more detail in the image, and I'd like more resolution, but I don't know that moving up to 24 megapixels will provide the sort of resolution I'd like to see.  You all can talk about the allure of MFDBs, but I'd like to have sex again sometime this decade, so that's out for now.

In the end I need to recognize that it's not just about me and my taste.  I've got a target audience (which also means no B&W, unfortunately) with their own set of standards, and from what I can tell 12 megapixel Nikon DX sensors surpass that level of quality.  It certainly will if I stitch a few images together, but odds are the right image will do quite well even as a single capture (and likely at a higher ISO than this -- I'm surprised how little "grain" came through on the final image.  More proof that I need to print more from digital to better hone my film-educated instincts.)

Thanks for the comments.  For now I can pay less attention to gear, and start following the post-processing discussions more closely.

[NikoJorj]

For now I can pay less attention to gear, and start following the post-processing discussions more closely.

My 2c's to this : for post-processing, if you got Lightroom, you've already made a fair part of the trip...
Advanced post-processing skills may help you to minimize artefacts, but won't bring you any more actual detail.
In this light, I'd consider Lightroom's (very good) interpolation and output sharpening, combined with a good use of its capture sharpening, as already sufficient.

[Geoff Wittig]

OK, time for a follow-up.  If nothing else someone may search to find this in the future...
Thanks for the comments.  For now I can pay less attention to gear, and start following the post-processing discussions more closely.

Congratulations!
Seeing how large you can print from a given capture resolution while still being proud of the result is always a fascinating exercise. My personal experience has been that my standards keep rising as the tools and my skills improve. I made some 24x36" prints about 6 years ago from scanned 35 mm Velvia film that delighted me. Now I look at the mushy transitions and pebbly grain with embarrassment. On the other hand, Brooks Jenson has pointed out that the ever-finer levels of image quality that we sweat blood over are often completely invisible to viewers who are not photoraphers.

I find that if I do everything exactly right (rock-solid tripod, mirror lock-up, low ISO, careful focus, great optics, perfect exposure, optimal processing and sharpening), I can print 20x30" from a single frame capture with a Canon 5D II or 1Ds III without any visible flaws. At least most of the time. Some images can stand going up to 24x36", but fine detail such as grass or leaves on trees will start to show digital artifacts or resolution limitations at that size. A very simple, plain, graphic image without lots of detail can go a lot bigger and still look good. If I'm willing to stitch frames together, I never really hit a resolution limit from my camera. The 24" wide paper path of my inkjet is the limiting factor, which is probably just as well. No doubt photographers who work with 8x10" film cameras would be dissatisfied, but I can't imagine how lots more megapixels would make a huge difference. At least to my eye the various limiting factors (lens quality, sensor resolution, inkjet dither) seem to converge at about the same place right now.

Oddly enough, there are still a few ghosts in the machine. Occasionally despite apparently doing everything right, I'll get an image that refuses to look good printed large, most often due (I think) to lens field curvature problems that reduce the optical resolution below the acceptable threshhold. And just as often I'll get a frame with just crazy good detail and resolution beyond what the numbers would predict. My belief is that the lens's curvature of field coincidentally matches the 'shape' of the subject matter and 'punches above its weight class'. It happens often enough to make me keep trying.

[Dick Roadnight]

I've been shooting weddings with Fuji DSLR gear, which works beautifully with the wide variety of lighting I might encounter.  Great in full sun for instance -- about as good as film.  The trade-off is that half of its 12 megapixels are digicam sized and only used for highlights

The "established pro" who took our wedding pictures used a similar Fuji, and the pictures are crap at any size... but much of the problem was operator error, and he is not in business any more.

There are many established pros who had quality MF film gear... and then downgraded to a first generation digital, and have never upgraded. (One locally uses an amateur DSLR but has Hasselblad pictures in his portfolio!).

...but when I have upgraded my first pro digital camera to an H4D-60/P3, I will not expect to upgrade again (I am 61)

[BernardLanguillier]

It really varies depending on image content, standards for print quality, and a host of other things like lens quality and shooting technique. My general rule of thumb is to have at least 175ppi before any interpolation. More is better, but with careful processing I can get good prints from 175ppi for most images (but maybe not all).

So for 12mp, 16x24 was my limit without stitching; for 24mp, up to 22x33. You can go higher for images that don't have lots of fine detail in them, but I'm a landscape guy so detail is often important.  I wouldn't go to 20x30 from 12mp for most images without stitching.

Another factor is viewing distance. Some argue that people view larger prints from further back,  so you don't need as many PPI. I think that's only partly true, as people will step up for a closer look if an image interests them. If you can control the viewing distance (eg it's not possible to step closer), then you might have a little more leeway here.

Indeed, it really depends on the subject.

During a recent exhibit I held it was totally stricking to see that most people would try to get closest to the largest one print of the whole expo (a 80x40 inch pano printed from a 400 megapixels file) and that nobody was trying to get close to the smallest image (a 14x18 inch print shot from a single D3x image). The main reason was the difference of scene.

Cheers,
Bernard

[Bill Koenig]

As others have already mentioned, if you want to print big, stitching multiple images together to make one large image is the way to go.
Down load the free trial of AutoPano Pro, and give it a go.

[feppe]

As others have already mentioned, if you want to print big, stitching multiple images together to make one large image is the way to go.
Down load the free trial of AutoPano Pro, and give it a go.

The OP shoots weddings, so stitching is not really an option 99% of the time.

[dzeanah]

Used to shoot weddings, hence the wrong background for the new efforts.

I'll be playing with stitching, but that looks like a new tripod head, plus some funky mounting bracket, and other stuff will be necessary to do it well.

Lots of stuff to learn.

[Dick Roadnight]

As others have already mentioned, if you want to print big, stitching multiple images together to make one large image is the way to go.

The OP shoots weddings, so stitching is not really an option 99% of the time.

Is is nice that there are wedding photographers who want to be able to print large, high quality images... do you remember when most top pro wedding photographers used Hasselblads?

...well Hasselblads are still the tools of choice for weddings.

Any digital Hasselblad works with Phocus, which gives you the ability to cope with the typical very high contrast wedding picture... you do not need to spend £30,000... but the H4D-60 is the ideal tool,, with good auto-focus and the ability to print 18 * 24" @ 360 original camera pixels per print inch, or 24 * 36" @ 240ppi.