4x5 resolution question

[don mills]

i saw an article online (http://www.clarkvision.com/articles/...l.1/index.html) providing a mathmatical formula for determining that the resolution of 4x5 velvia equals or exceeds that of digital capture capable of 150-200mp. the wikipedia entry backs this up (http://en.wikipedia.org/wiki/Digital...m_photography).

"A medium format film image can record an equivalent of approximately 50 megapixels, while large format films can record around 200 megapixels (4 × 5 inch) which equates to around 800 megapixels on the largest common film format, 8 × 10 inches, without accounting for lens sharpness. A medium format DSLR provides from 42 to 50 megapixels, but cannot be enlarged with the same level of detail as medium format film"

but i also have read a luminous landscape review where 4x5 was supposedly "bettered" by phase one's p45+ (39mp). (http://www.luminous-landscape.com/essays/Cramer.shtml)

if someone could explain this disparity i would appreciate it. i use a phase one system but stick to LF for my personal work because i do feel it holds more information (yet i cannot explain why at the technical level).

thanks!

dm

[NicolasArg]

Well, I don't have the EXACT answer to the question, but I would like to mention just 3 things:
1). LL is totally antifilm and the "tests" where a 6mpx DSLR "beats" a MF film capture are enough proof for that. After I read the mentioned test I knew LL is totally biased towards the latest high end digital technologies. Their "tests" are written just like the sprayette infomortials- a brief introduction to mention just how experienced the author is, then a few paragraphs to speak about his long time film affair and then the grand finale about how he saw the light and now is a happy owner of some bazzillion dollars MF digiback which would be outdated faster than my 45 bucks cell phone.
2). I lack credentials to argue with the authors but the scans have clearly more details everywhere.
3). No marketing chatting and weird formulas will change the fact that a bigger "sensor" captures more info.

[Matus Kalisky]

I think it is a bit of beating a dead horse, but let me give it a try. Any digital to film comparison can easily be flawed by not handling the film properly.

But just to estimate how much information can be extracted from a 4x5 film let's say that you have a sharp lens that gives you say 60 lines per millimeter (this means shooting at around f/11 - f/16). 60 lines per millimeter is no stretch for a slow slide film so it should be possible to to get film exposed that really have this kind of information density (if you do everything right!). Now - not to loose any information during the scanning you need to scan at at least 120 samples per millimeter what turns to about 3000 spi.

4x5 film has about 3.75 x 4.75 area exposed what is about 18 square inches.

Now - we are scanning at 3000 spi what gives us 9 Mega pixels per square inch what is about 150 Megapixels per 4x5 film.

Now to turn this into data volume - with 8 bits ( = 1 byt) per color you have 450 MB of data. If you were to scan with 16 bits per color channel you have 900 MB.

With this kind of scan you are able to to print 32x40 inch print at 360 dpi without interpolation.

BTW - any direct digital capture suffers from Bayer interpolation (to recover color information) what yields in some detail loss, so "true" image data in a digital capture are about 70% - 80% from the total pixel count.

Does this help?

[williamtheis]

no Bayer interpolation in large format scan backs, though. so colors are very pure at the expense of long scan times and image artifact problems when the subject moves. there is also a Black&White non-Bayer capture DSLR...

in general I can get excellent resolution and color out of 4x5 chromes but equally good out of my digital BetterLight scan back... I worry about the day when I can no longer get reliable e6 chemistry kits to process sheet film... the BetterLight is a decent answer to this problem for me.

digital, film... they are both tools. you can make good images with either (or make bad ones)

[David Luttmann]

All I can say is what I've seen on print. The latest digital backs lose nothing to 4x5 up to 24x30. Beyond that, the 4x5 is better in some ways, and the digital in others. You have to decide what is important to you. Keep in mind that if you're using a flatbed scanner like an Epson V700, the 4x5 will lose a further quality step.

[Ben Syverson]

You have to define the term "Megapixel." All megapixels are not created equal, which is readily apparent if you compare the real-world performance of various sensors with the same nominal MP count.

I find it useful to abstract the concept entirely, to "real" megapixels -- defined as the minimum megapixel count necessary to resolve the number of lp/mm a system is capable of. Nyquist tells us we need 2X the number of samples as our highest frequency, so the equation becomes:
horizontal lp/mm * sensor width in mm * 2 *
vertical lp/mm * sensor height in mm * 2 = REAL megapixels

So for example, PopPhoto says my 35mm scanner tops out at 67 lp/mm. Many lenses can do better than that, but color film such as Portra and Velvia usually max out around the 60-70 lp/mm mark anyway. So for my particular 35mm setup, I would do:
(67 * 36mm * 2) * (67 * 24mm * 2) =
(4824) * (3216) =
15.5 Real Megapixels.

In other words, a "perfect" 15.5 MP sensor could match my 35mm scans in resolution. Of course, DSLR sensors are far from perfect. They have Bayer filters and usually optical low pass filters. Let's take the Canon 5D Mark II for example. According to DPReview's findings, it resolves about 2100 x 1350 line pairs (confusingly they use lines, not pairs), or about 58 lp/mm. If we plug it in, we get:
(2100 * 2) * (1350 * 2) = 4200 * 2700 =
11.3 Real Megapixels.

That's in line with my expectations. My 35mm scans are still sharper than my DSLR, but they're so noisy (grainy) that the 5D usually "looks better." But SNR is another discussion entirely. And of course if you shoot a slow B&W 35mm film with Leica/Zeiss and print traditionally, you will absolutely annihilate any of the Medium Format Digital solutions out there. Just destroy them.

To finally answer your question... LF lenses (well, actually diffraction) and scanners are the limiting factor in most LF workflows. Even the best LF lenses at working apertures (~f/22) may drop to as low as 30-40 lp/mm, and flatbed scanners are in the same range, if not lower. So let's take 30 lp/mm as a number for 4x5 scanned at home.

(30 * 120mm * 2) *
(30 * 95mm * 2) =
41.0 Real Megapixels

That doesn't seem high, but it compares very favorably with MFD. For example the Phase One P65+, with a nominal rating of 65 MP, could only hope to reach 75% of that, due to Bayer alone. I haven't seen a methodical test of line pairs on that camera, but my guess is that it would clock in between 40 - 45 Real Megapixels. So if all you're looking at is resolution, the P65+ is a solid 4x5 stand-in. But if you shoot 4x5 at wider apertures (f/5.6 - f/11) and drum-scan, it would be possible to hit the same ~67 lp/mm as my 35mm example, which would net you an easy 204.7 Real Megapixels from 4x5.

Of course, it's silly to use resolution as the only criterion for choosing a format. Film uses grain, which serves as spatial and tonal error diffusion, creating a more subjectively "natural" feeling image, versus the fixed square pixels on a sensor. Negative film retains many stops of highlight information that would be impossible to capture with one shot on a digital sensor. Digital has a lower SNR in the "midtones" (but dramatically higher SNR in the shadows), which creates the appearance of a "cleaner" noise-free image. Each system has its advantages.

And you can't discount the intangibles of actually shooting with each system, such as personal enjoyment. I could have a point and shoot digital that takes perfect 400 MP images, but I'd still prefer the view under the darkcloth...

[bobwysiwyg]

I hope I'm taking this thread off-topic, but I'm fairly new to LF, but have seen references to these 'digital backs' and long scan times. How long are we talking about and how does it affect subject selection?

[SCHWARZZEIT]

The resolution of film based images depends to a very large degree on the contrast of the detail at hand. If you shoot a printed test target of black and white lines (about 3 stops contrast ratio) many films are able to resolve more than 100 lp/mm from a super sharp lens which is still a little better than a 5 micron sensor of the same format could do theoretically. In real world scenes, however, this kind of contrast for micro detail is rare and another point is that detail is not necessarily coming in form of straight lines that can be continued by our brain even if the structure is heavily broken by grain noise. So if someone calculates megapixel equivalents based on resolution measurements off high contrast test charts you get huge numbers. But that would be only half the truth.
The resolving power of film really suffers from low contrast. The manufacturer's MTF describe how these films handle that contrast fall-off. It's basically the effect we see when subtle detail gets lost in the grain noise. Most of the detail in a typical scene is rather low contrast which leads to the real world conclusion that you need a larger film area to capture similar effective detail as modern sensors can do. These properties of film make it really impossible to pinpoint a megapixel equivalent for each film format.

The sampling of detail from sensors is totally different because a sensor's resolving power depends only to a very small degree on contrast. You can easily get very sharp micro contrast of low contrast detail on pixel level from a sensor as long as the detail is not buried too deep in the shadows. Though some detail gets lost through the Bayer filtering but the effect is not as severe as what the grain is doing on film.

Here's how I see it from a practical point of view based on my experience as a drum scanner operator:
When I downsize my 2 GB drum scans of 4x5" color negative films to 1900 ppi which results into a 63 MP file it's not as clean and sharp as the technically best images from a P65+ back. But there are quite a few areas where the larger scan holds detail that needs more than 60 MP to be kept. A super fine grained transparency film like Fuji Astia 100F or Kodak E100G can capture definitely more than 70 MP of very clean information on 4x5" format. Because of the larger density range used for the image slide films are better at capturing low contrast detail than print film.

In the end I think it's more important if you prefer the analog or digital character for an image than any technical advantage of one technology or the other.

-Dominique