529 DC - REPRODUCTION QUALITY
January 9, 2003
The fine detail and sharpness of images created by professional digital cameras on a CCD
chip tend to reproduce much better than those created using film -- even when the
original digital creation file size is much smaller than the drum scan made from a piece
How is this possible? There are several factors including first generation image, lack
of grain, greater dynamic range and the ability to make minor (and sometimes major)
corrections after creation that account for this fact. While hard for people who have
been raised on film to accept, Digital Is Better, and it's better today on the
professional level cameras -- not at some foreseeable point in the future.
Let's take it step by step. Everyone would agree that a first generation image film is
better than any second generation of that image. The drum scan of a first generation film
image is a second generation. Also recall that until a couple years ago most photo buyers
were reproducing from a "reproduction quality" dupe -- and buyers were scanning that dupe
(third generation) to make their separation negatives.
On the other hand, once you have a digital file each copy of each bit of data is an exact
copy without loss (as long as you don't try to manipulate those bits in the process).
Thus, clearly if you start out with digital capture the laws of physics would indicate
that you should have a sharper and more detailed original. We also know that every time
we make a film dupe of a transparency there is some loss in image quality.
Digital originals have no grain. This is a major factor in apparent sharpness. Digitally
produced originals are subject to noise inherent in their imaging chips and circuits.
This noise can degrade the image, but with the professional level cameras shooting at low
ISO's the amount of degradation is extremely low. [It should be noted that with the
professional cameras the ISO can be adjusted through a broad range (200 to 1600 with the
Nikon D1H and top level can be pushed even higher with custom settings). Some
photographers push the ISO to the higher level in order to try to achieve the "grainy
Writing in Photo Electronic Imaging in 1999 Andrew Rodney said, "Film has grain, and
grain seems to obscure a great deal of detail that digital capture completely eliminates.
This combined with the far wider dynamic range of digital cameras seems to put
conventional film to shame."
Rodney shot film with a Hasselblad and drum scanned it to 102MB. He compared that with an
19MB digital capture that was digitally interpolated to 102MB after capture. The results
from the digital were much better. Since the time of this test the quality of digital CCD
chips and interpolation software have improved. (For Rodney's full article check
www.digitaldog.net/tips.html and click on "Film vs. Digital".)
Many photographers, particularly those shooting for magazines and newspapers, are already
shooting entirely digitally. Photographers who have not yet gone digital need to give it
Charles Mann earns his living as a full time Equine photographer. Many would consider
this a very narrow specialization. For the last five years his shooting has been almost
entirely digital, mostly with a Nikon D1H. Most of his work is generated from covering
all the major national and international equestrian events from the Olympics right on
down. At the last Olympics he shot over 20,000 images of the contestants. His digital
images have been used on more than 100 magazine covers and he sells many of his images to
advertisers who provide products and services to the Equine market.
Are these images of horse and rider jumping fences really good enough -- and the D1H file
sizes large enough -- to produce top quality covers and double page spreads for major
magazines? To answer that question Charlie showed me a 20x24 inch poster he did for
Purina of a horse and rider jumping a fence. The image is tact sharp. Purina has also
used the same image 10 feet wide as the background for a trade booth.
Even more impressive for those who still believe in film was a 2002 Calendar that has 9
of Charlie's images and 3 from another equine photographer who still shoots with film.
All of Charlie's images are as equally tact sharp as the image on the poster. The images
from the other photographer, who shoots on film with Canon equipment, are fine images by
every normal magazine and editorial standard. If these images were standing alone no one
would have a second comment about them, but due to the grain effect they appear
significantly softer than Charlie's images.
It is interesting that our eyes have become so accustomed to accepting grain in an image
that we disregard it until we see two images of the same subject matter side by side.
When one is captured digitally and the other shot on film and drum scanned in order to
get a file suitable for reproduction the digitally captured image will usually look much
No one thinks that the difference in these results means that Nikon is better than Canon.
The printer and the paper is obviously same. The only difference is that in one case
separation negatives were created from pieces of film, while in the other separations
were created from digital files provided by the photographer.
But the Nikon D1H only produces a 3.8MB RAW file and everyone in the stock photo industry
is telling us that we need a 40MB or 60MB scan as a standard? How is it possible to get
such a good image from such a small file?
Let's talk about that 3.8MB RAW file. Nikon lets you save an image in either the RAW
format or in one of 3 JPEG formats. The RAW format gives you more data to work with, but
it greatly limits the number of images you can save on one Compact Flash Card when you
are shooting. Consequently, many photographers only save in the RAW format when they know
for sure that they need the highest resolution file, and that the image is going to be
used very large.
If you save in the JPEG format then the 3.8MB file is lost and you have to live with
whatever remaining file size you've saved at. There is some loss of data when the file is
JPEGed, but in most cases very little. This problem is most apparent with images that
have a lot of very fine detail.
Because most of Charlie's work is sequences of horses jumping until recently he saved
everything as Nikon normal JPEG which is about a 600K file. The poster image was saved as
a normal JPEG and it started out as a 598K JPEG file!! As in all cases where he
needs a larger file size Charlie uses Genuine Fractals to interpolate the data and "res
up" the file.
Wouldn't the image have been even better if the 3.8MB RAW file had been saved, or if the
image had somehow been shot as 40 MB? Undoubtedly, but as you look at the reproduction of
this image it is hard to conclude that it could get any crisper. (And remember, it is
definitely crisper than film.) This same question used to be asked about film. Wouldn't
that 35mm image have been better if it has been shot on 4x5. Sure, but is it humanly
possible to catch that magic moment that was caught with a 35mm camera on a 4x5 that you
have to lug around and set up on a tripod before shooting?
And even if you do have a lot of additional resolution is it of much value if the image
is going to be reproduced with ink on paper using a printing press. The ink, paper and
printing press are the limiting factors in how good the image will appear. As we have
moved into the digital environment in the past few years most professional photographers
have become aware that film has a lot more detail and resolution than it is possible to
reproduce in the printing process. Since the purpose of our business is to reproduce the
captured images on paper through a printing process the real question is "What do we need
to get the best reproduction?"
This gets back to the old argument we had a decade or so ago about film size. Everyone
thought you needed to deliver large format (6x7, 4x5 or larger) film because you would
get better reproduction from a larger piece of film than a small 35mm. It wasn't true
then once separators learned how to deal with 35mm, and it isn't true now.
How Large A File Is Really Needed?
I'm not saying that you should rely on small files, and all other things being equal it
is probably better to always capture the largest file possible an save the RAW format.
But the problem is that all things are never equal and you may need to balance issues
like the weight, cost and mobility of equipment, limited storage on data cards, etc.
Also, if you're going digital you don't have to rely on small files. The Canon EOS-1Ds
gives you an 11 million pixel photo (11 MB) that according to Rob Galbraith, "are just
dripping with detail, the sort of detail that appears equally crisp and fine in 5x7, 8x10
and 12x18 inch enlargements. At its best, the pre-production EOS-1Ds body in hand here
generates photos that match or exceed the level of clarity of the best 6x6 Hasselblad and
Mamiya 6x7 prints I've ever made. In my case, that's fibre-based and REC custom prints
from TMax 100 B&W vs both colour and black and white EOS-1Ds photos printed on a Canon
S9000. Arguably, in such a comparison, digital should be at a disadvantage, given the use
of a desktop inkjet for the digital output. As it turns out, this is no handicap. The
smallest landscape detail in EOS-1Ds frames hold up at or beyond the level of the
traditional darkroom prints. I'm talking about fine, smooth, photographic detail, free
from sharpening-induced pixelation or other digital oddities."
If your goal is to produce quality images for reproduction recognize that it is easy to
do with the hardware and software available today.
Recognize that to get equal reproduction you must start with a much larger drum scan from
film in terms of megabytes than is necessary if the image is captured digitally.
Recognize that when more pixels are needed to make a larger reproduction of a relatively
small digitally created RAW or JPEG file interpolation software will usually produce a
more satisfactory result than if the image was shot on film and drum scanned at the
From: James West, Alamy.com
I thought the articles on digital stock were very interesting, however one
point is slightly misleading in the section on Reproduction Quality. You
compare a 3.8 RAW file size with 'a 40MB or 60MB scan'. To make a proper
comparison you need to convert the RAW into RGB format. Similarly, when
quoting RAW file size, you should state 'RAW' where relevant, otherwise
your description of the Canon 1Ds '11 million pixel photo (11 MB)' reads
as though the RGB file size is 11MB when in fact it is approximately 31MB
To convert RAW file sizes into RGB file sizes you take the total no. of
pixels, which in the case of the Canon 1Ds is:
Convert the figure into 'bits' by multiplying by 24 (an RGB file has 8
bits per channel; RxGxB = 3x8 = 24);
Convert to bytes by dividing by 8 (1 byte=8bits);
Divide by 1024 for KB
Divide by 1024 for MB
You probably already know that but once I feel an equation coming on I
just can't stop it.
Editor's Note: I've always been confused by how to do this calculation and
this is the simplest, most straight forward explanation I have seen.
From: Alan Bailey, Rubberball
We've been shooting with canon's new 1Ds for about a month now. What we've
found is the first camera that produces an image better than a comperable
120 or 220 shot out of our hasselblad (taking into account the effects of
film grain after you drum scan the image). Simply amazing! Also, you're
not running up against film grain when you blow the 1Ds image up, so the
images interpolate up much better than film.