View Full Version : Print size chart when using digital capture
FYI: Just thought I'd point out this new chart to those of you who (like me) use digital cameras as part of your camera kit, and are monitoring developments in the high-end DSLR/MF digital back market. It's a nice little summary of what can be theoretically achieved when lenses match the capabilities of the digital sensor.
While one can quibble with their DPI assumptions (I know some folks prefer more than 200 DPI where possible), in general it reinforces the notion that 31/39 MP digital backs are roughly equivalent to drum-scanned 4x5 film.
Of course, the current generation of digital wide-angle lenses (particularly with regards to 16+MP DSLRs and 31/39 MP digital backs) generally does not resolve to the level of high-end sensors, so the results on this chart often cannot yet be fully realized. Problems currently occur when off-axis light (camera movements) hits digital sensors. The current generation of "digital view cameras" is still rather heavy and clunky when used with digital backs, particularly in the field. And better methods of determining critical focus with digital view cameras are needed (it's hard to confirm critical focus using such a relatively tiny, coarse ground glass, particularly when using movements). So I certainly don't plan on selling my 4x5 anytime soon.
That being said, it will be interesting to see how the digital technology evolves over the next year or so. Then, of course, if only the cost of this stuff would come down..... :-)
Thanks for the link Eric.
I'd be one that quibbles with their 200 PPI assumption for "excellent" and would cut it off at 240. I'd then move "good" to 180 through 240 and would not go lower unless it were for a billboard or banner.
FWIW I have seen the new 39MP files up close (on the computer and printed to 36x48) and while exceptional, they are still not quite up to par with drum-scanned 4x5, all else equal. But it is a close race. The DS has more detail and better shadows. While the digital file has no grain at all, I feel the detail and shadows from the DS win out. And the work-flow obviously favors digital, so I suspect many will abandon ship because the digital file is close enough and more convenient. More cost-effective in the long-run too if one shoots a lot.
(As an aside and IMO only, the Betterlight scanning back at 48MP (6000x8000) is pretty comparable to DS 4x5 film -- and has the edge with no grain. Some BL models will go even to higher resolutions -- 9000x12000 or 12000x15990 -- though few LF lenses are capable of fully utilizing those resolutions. Also, the BL is not as universal as film or single-capture digital, though it is significantly cheaper than 39MP direct capture. It also has the advantage of using existing 4x5 cameras and lenses so it may make sense for some depending on their application.)
I can very easily see the difference between 360 DPI and 240 DPI from a digital capture. In addition, the bayer artifacts from digital capture are easily seen in foliage, even at 360 DPI, although at that level of enlargement not too objectionable.
Of course unlike most middle-aged photographers I am nearsighted instead of farsighted and I like to look at prints with my nose on the paper. So at 1-2' viewing distance YMMV.
I'm not sure how many saw this but a while ago Charles Cramers tried out this 39MP back and posted results at the Luminous Landscape. He was using Capture One to process the RAW file. Now in the same way that different developers work better with some films than others, Capture One has not yet been optimized for the P45. Charles tried this test against 4x5 film drum scanned at Calypso and the RAW Developer software package for the Mac. His newly posted samples can be found here:
His new comparison to the scanned Velvia concluded that he found more detail in the leaves with less noise. His impression is that at the 60" print size, the digital files would be comparable to the 4x5 film. As many of us print signifcantly smaller than 60", at 30" & 40", there would be no perceptable difference.
I'll echo the comments for the Betterlight. The 48MP sample I tried provided better resolution & less noise than drum scanned Astia 4x5. Their top models exceed 20,000 pixels wide and as such, would provide better quality than scanned 8x10.
Now if only these guys could do something about the price tags!
I'm with you regarding 240 DPI. WCI once told me that with the Epson inkjet printers, reduced resolution starts becoming noticeable once the DPI falls below 240, albeit when viewing the print through a loupe. While the argument is that the naked eye will not notice such differences, I'm not really so sure (most folks who have seen my prints like to get up close and personal with them, marvel at the smallest details, or even look for the elusive film grain). One of the joys for me of LF photography is that my prints can fulfill such lofty expectations.
Which LF lenses do you find do justice to your scanning back? My concern has been that the sharpest 4x5 lenses tend to fall within the 110 to 240mm focal lengths (e.g. SS110XL thru Fuji 240A), and that lenses beyond this focal length range tend to be softer and might noticeably hold back the scanning back. What has been your experience so far?
Also, the street sentiment I recently encountered here in Los Angeles is that the advent of 30+ MP digital backs has drastically cut back the market for scanning backs, limiting these devices to art reproduction and other niche applications. Can Betterlight remain economically viable in such an environment?
Yes, I saw the Charles Cramer P45 article, and the intriguing thing to me is that he shot with a Mamiya AFD, whose lenses may not fully resolve the P45. If he had used a Rodenstock HR lens, he might have gotten a still sharper result. On the other hand, I'm sure you've seen the latest LL article on diffraction limits with the P45 (www.luminous-landscape.com/tutorials/understanding-series/u-diffraction.shtml (http://www.luminous-landscape.com/tutorials/understanding-series/u-diffraction.shtml)), which shows lens performance starting to noticeable deteriorate at apertures smaller than f/8, even when using a Rodenstock HR lens. I had always assumed that one of the benefits of an eventual shift to an MF digital view camera would be increased depth of field due to the smaller format, but if f/8 in MF digital turns out to be roughly equivalent to f/22 on 4x5, I'm not sure that you've gained any depth of field at all. I normally assume a three-stop depth of field difference between 645 and 4x5, which would mean that 645 f/8 and 4x5 f/22 are roughly equivalent. MF digital sensors are not full-frame 645 (at least not currently), so perhaps you'd gain another stop or so. But nothing like the order of magnitude improvement I once envisioned.
It will be interesting to see how all these issues pan out. We live in interesting times!
I would say most any LF lens can do justice at 6000x8000, but that only means they will deliver as they do for film, not to the full potential of the sensor. If we add the full potential of the sensor as the qualification, your hunch is about right. I would add that my Nikkor 90 can just do it, as can my Nikkor 300 and Fuji 450 with room to spare -- and I'm sure there is other long glass that would serve well too. However shorter lenses with full IC's for 4x5 seem to be the Achiles heel. For example, my 75 Super Angulon was more than sharp enough in the center, but ran out of resolution about 2/3 the way out to the edges. Nonetheless, a print from it looks as good as one done with film, so at least nothing is lost in that regard ;)
As for the scanning back's future, it's hard to say. True, the 33+ MP backs are delivering incredible results and certainly offer more convenient capture options, but from what I have seen the BL still carries an edge in absolute quality and I for one believe it still has a lot of potential for the future. Clearly it reigns king in arenas like art-repro yet in addition to its better absolute detail it captures in true color without any Bayer interpolation. To me it is kind of like the difference between shooting 4x5 and 8x10 -- it provides a better but few photographers are willing to put forth the extra effort to glean those improvements.
BTW, for anybody interested I have just posted a few images from a recent shoot in the field with the Betterlight on my blog. Note that the 100% crops are 600x800, or 1/100th of the total image size: www.jack.omniblog.com (http://www.jack.omniblog.com)
Hi guys, am I right in thinking that the digital backs you are talking about (the ones that are equal to 4x5 film) all must be run from a laptop? Top of my wish list is a digital 4x5 back that can be used without a computer...
Real time backs like the Betterlight must be tethered. Backs like those from Imacon and the P45 do not require a tether....they have built in storage or use CF cards.
Not only does the Betterlight require a laptop tether, but it also has its own dedicated drive box (it stripes the files directly to a hard disk). So you have the back tethered directly to the box and the box tethered to your computer. And of course the box requires its own power source, so add in another battery.
The good news is the laptop, box and batery can all remain in the pack with the back as the only tether from the pack to the camera. Nonetheless, it is a lot of peripheral gear for field-useablity.
Judging from the numbers in the Perez/Thalmann test http://www.hevanet.com/cperez/testing.html there are plenty of lenses that provide enough resolution to exploit the Betterlight. Lenses for the P45: I happened to talk to Bill Atkinson shortly after the Atkinson/Cramer/Reichmann P45 shootout, and his observation was that there was no significant difference between Hasselblad lenses, Mamiya lenses, and Rodenstock lenses. By the way, Bill also disagrees with the notion that the Tango is not amongst the sharpest, and says he can observe a difference in resolution between 10,000dpi and 11,000dpi.
But the Perez/Thalmann test results show resolution differences when using film at f/22, and my concern is that these differences would become magnified when using the Betterlight. For example, the two Fuji 240A samples tested consistently recorded resolutions of 60 lpmm or greater even at the edges, whereas the Nikkor 300M and Fuji 450C samples tested recorded resolutions no greater than 54 lpmm (and in one case as low as 34 lpmm). Since the 240A data proved that the test setup is capable of resolving at least 60 lpmm (actually one lens tested resolved as high as 85 lpmm), mustnít the lower resolutions recorded with the Nikkor and Fuji be solely due to limitations with these lenses? And if resolution is capped at 54 lpmm by lens design, how can a digital scanning back possibly record any greater resolution than that recorded with film?
One of the limitations of the Perez/Thalmann tests is that they were performed at a magnification ratio of 1:20, which for 4x5 equates to a focus distance of roughly 8 feet. Much of landscape photography is shot at or near infinity focus, which may be why Jack has experienced good results with his 300M and 450C (my understanding is that the Fuji C series are optimized for infinity, and this has certainly been my experience with my 450C and 600C lenses). But if one wants to shoot close-up with these lenses (as I frequently do), then my concern is that the scanning back would not provide any additional resolution over film. Similarly, I use my SS80XL for roughly 20% of my shots (frequently with lots of front rise), and here again, the concern would be that this lens gets relatively soft at the edges, and (like Jack's Super Angulon) would not provide any additional resolution versus film. To makes things worse, the Betterlight has a 1.25 focal length multiplier, so I'd have use a 65mm lens to get a similar angle of view to my 80mm, which almost certainly will be even softer (and have significantly less image circle). So add these resolution limitations to the challenges of slow scanning speeds and multiple tethered system components, and the Betterlight becomes an extremely specialized tool.
I agree that the digital scanning back could become more universally interesting if subjected to continuing technology development. I worked in the computer hardware industry for 20 years, and I see no reason why a scanning back could not eventually migrate into a far faster and more integrated solution (perhaps not completely unteathered, but laptop computers are getting so light and power efficient that this may not matter much). The question is does Betterlight have enough revenue to finance such continuing R&D? And will the market be big enough? I certainly hope so. But if more and more people find 30MP to be "good enough", then this certainly does not bode well for future scanning backs.
As for lens resolution with the P45, I'm glad to hear that the Hasselblad and Mamiya lenses held up well in the shootout test. Perhaps the resolution difference experienced by Michael Reichmann between his Contax and Rodenstock wide-angle lenses was just specific to that particular focal length (or limitations in the Contax), and should not be extrapolated to other focal lengths. Yet I keep hearing rumblings about Hasselblad and Schneider developing new lenses to better resolve digital sensors. Perhaps these rumblings are false, as with so many rumors. Hopefully we won't have to wait too much longer to find out!
Powered by vBulletin® Version 4.2.0 Copyright © 2013 vBulletin Solutions, Inc. All rights reserved.