Understanding how a 39mp digital back is comparable to 5 x 4 film?

[andrew_keam]

Hi, not sure if this is the right forum to post this question in but was just wondering if someone could

explain how a 39mp digital back is equivalent to 5 x 4 film. I have heard this comparison many time

but dont quite understand it. I understand such comparisons when talking about small prints but not

when it comes to large mural sized prints. This equals a 120mp file which when printed at 250dpi

only equals a 32inch x 24inch ( or thereabouts ). I recently saw an exhibition of Edward Burtynskys

at Photofreo in Fremantle, Australia and his prints were huge. Maybe 50inch x 60inch and still

looked very good. I also read in the paper that he shot them with a digital camera and a gyroscope,

leaning out of the helicopter. How is this achieved? I primarily shoot 5 x 4 but am seriously

considering getting a digital back after seeing those prints. I am also interested in aerial and have

tried shooting 5 x 4 with it which was a bit of a disaster ( possible but not ideal ). I also am looking

to print quite big ( 50in x 40in plus ). Is a large amount of this done in the capture software stage or

is it in programs like Genuine Fractals?

 

In this article in Luminous Landscape, it states that the maximum print size for a 12mp camera is 16

x 24inches @ 180ppi while a 21mp camera cam produce a maximum print size of 21 x 31 inches

@180ppi. ( I realise that print quality is subjective ). This brings me to my next question. What is

the lowest ppi you can print at ( considering that it is a big print ) without losing quality? Also, most

places that I have enquired about printing give a set file size for prints ie 250ppi or 300 ppi. How do

you get around this?

 

Thanks, I realise that there are a lot of variables in there but any help would be greatly appreciated.

Thanks Kind Regards Andrew

[User_503771]

First off, in watching "Manufactured Landscapes," the subject of which is Burtynsky and his work in China, I didn't see him using any digital equipment. (The movie was first shown in 2006.) I did see him using medium and large-format equipment, and it was pretty obvious that he was shooting film.

 

Regarding the pixel count question: I suspect that you already have the right idea, but you've run into a "fact" that runs contrary to what you already understand.

 

I personally don't believe hat 39 megapixels is going to be comparable to a well-exposed and processed 4x5 sheet of film as regards detail, sharpness and "nuances". If you scan a 4x5 sheet, you get a humungous file that is, I'm sure, far more than 39 megapixels. Rather, I'd suspect you're probably well over 100mp. I did try scanning a 4x5 film at 2400 dpi but backed out partway through the scan. I've read reviews saying that you can easily go over a gigabyte in the file size from a 4x5, and I believe that -- but I don't recall what resolution they were describing there, but I think it would be around 3200.

 

I suggest that you take one of your own high-res scans and make a copy of it. Resize it and print a crop onto 4x6 (or larger if you don't mind using the ink). You could do this several times at different resolutions, printing each one and noting on the back its resolution.

 

Compare these crops at a few viewing distances. (For this, maybe a half of an 8x10 would be better, but you decide -- you get the drift.) See for yourself how a print looks printed at 150dpi, 180dpi, 300dpi, 600dpi -- whatever above that your printer can handle -- provided that your original scan is higher resolution than the printed output.

 

I always try to scan at the highest resolution I can stand. One reason is that I don't want to have to scan the same piece of film more than once; I'd like my high-res scans to be the best possible, then just archive them and put them to whatever uses I wish in future. The other reason is that I like to make my final file size large enough so that I can print at the highest resolution my printer will give me -- regardless of all the stuff people say about "acceptable" print resolution!

 

I know that when I go a photo exhibit, sure I like to see the print at "normal" viewing distance. But when I see a print I like, I also get real close, to see how the thing looks that way too. I know a lot of people who do that, especially fellow artists. So I've always printed (darkroom and digital) to the absolute highest standard I can possibly muster. That way the print will have something for everyone, "normal" viewer (who often get close up too) and artist alike.

 

Anyway, find your own acceptable viewing distance for your work, and its acceptable resolution, then do the math. Then you'll know for sure whether the 39mp quote for 4x5 is true for you. (As I said before, it's not true for me!)

 

I hope this helps....

[ellis_vener_photography]

The big difference is that the 4x5 film needs to be scanned. Even the best drum scans of film are a second generation dupe, and what I have found is that to get comparable resolution of detail the scanned file needs to be about 3 times the size of a direct digital capture of the same scene.

 

Since a 16 bit per channel TIFF from a Phase One P45+ (39mp) back yields an approximately 224 mb size file (24x18 inches at 300ppi) to get the same print quality out of a very well done drum scan you'd need to start with a scan that is 3x as large to get the same quality of print.

 

Secondly because the print is in many ways the first tangible generation of the photo a direct digital photo scales up to larger print sizes without also scaling up both the artifacts from the scanning process and the defects in the film emulsion.

 

Finally high resolution high quality digital cameras ( Canon EOS 1Ds Mark 3, EOS 1D Mark 3, and the Nikon D3 ) and medium format digital backs have native dynamic ranges that surpass color films. Difference is detail and with a nikon D3 capable of independently measured dynamic range of around 12 EV you can record more fine differences in highlights and shadows than color film can.

 

But Michael Hendrickson alludes to something else as well. Photos shot with a digital camera admittedly don't have the same visual "bite" that photos made on film do. that "bite" or presence leads to a perceptual phenomena of greater acuity of film. Another way of stating this is that technically well shot and processed digitally shot photos have smoother tonality. For this reason I know a very well known color photographer who leaves his D3 setto ISO 1600 to get that kind of low noise presence back. I've seen his prints from his D3 at ISO 1600 printed up to 20" x 30" and in those prints they render highlight and midtone details comparable to 35mm Kodachrome 64 film printed at that size -- but are a bit "noisier" in the dark shadows.

 

"In this article in Luminous Landscape, it states that the maximum print size for a 12mp camera is 16 x 24inches @ 180ppi "

 

Having printed my own D3 and 1D Mark 3 images at 16x24 @ 300dpi I find that statement out of date. A lot depends on the pritner and paper used.

[dennismcarbo]

Hi Ellis,

 

"Since a 16 bit per channel TIFF from a Phase One P45+ (39mp) back yields an approximately 224 mb size file (24x18 inches at 300ppi) to get the same print quality out of a very well done drum scan you'd need to start with a scan that is 3x as large to get the same quality of print.Since a 16 bit per channel TIFF from a Phase One P45+ (39mp) back yields an approximately 224 mb size file (24x18 inches at 300ppi) to get the same print quality out of a very well done drum scan you'd need to start with a scan that is 3x as large to get the same quality of print."

 

Doesnt this assume both Photos are capturing the same level of detail from the start ?

 

Granted I am new to the 4 x 5 world but, it seems to me that the 4 x 5 captures much more detail than even the P45 can ...Would you say more setail is captured on the film ..we are just limited by the scanner at this point ?

 

I shoot a D200 and the 4 x 5 provia scans I have blow away my D200 to my eye at least. .......Aww ...I love it all !

 

 

Dennis

[ellis_vener_photography]

"Doesnt this assume both Photos are capturing the same level of detail from the start ?"

 

Yes it does. The factors are: focus and lens properties and how stable the cameras are. Another limitation of film having to do with focus is how flat the film is held in the film holder at the time of exposure and how accurate the alignment is between the groundglass and the film --or if you using a medium format camera to hold the digital back, how accurately it can be focus. With a digital back , at least i nthe studio you should be able to work tethered so that you are judging focusing accuracy on the computer's display --you can't do that with film.

 

I would certainly expect truly professionally scanned 4x5 Provia to "blow away" D200 or even D3 and 1Ds mark 3 quality for very large prints ( 16x20 and over) -- but I've made bad scans of 4x5 and other professionals make scans that were just as bad if not worse than what a cheap 4mp point & shoot can do even when judging small prints.

 

I feel qualified to make these remarks because of the past 20 or so years I've consistently had large prints made for clients from 35mm, various medium format sizes, 4x5 and 8x10 film and also made from various levels of digital cameras. We can bandy theory all we want to but the thing that settles any argument is looking at prints. And even then that settles very little becasue digital photographs originals and film photographs are really two different mediums with their own inherent positive and negative characteristics.

[dennismcarbo]

Hi Ellis,

 

Good points, There are so many variables I was unaware of with LF Film...I do love the results when I get it right though! ..thanks for the reply

 

Happy Shooting All

[sal b]

From a review of "Manufactured Landscapes":

 

<a href="http://www.specifier.com.au/pastissues/view/detailed/c/ManufacturedLandscapes/id/15978"> http://www.specifier.com.au/pastissues/view/detailed/c/ManufacturedLandscapes/id/15978 </a> :<p>

 

"<i><b>The technology is old, and manual: a squeezebox-like 4x5 or 8x10 camera, heavy equipment, and obsessive darkroom printing. Burtynsky takes as many as 200 shots of a subject, using three or four brands of film, and prints them on three or four brands of paper. The most vital and luminous combination of these forces becomes the final work."</b></i>

[ben_rubinstein___mancheste1664880652]

Of course you can use a view camera with digitar lenses and stitch 3 MFDB files together to create a 120 megapixel 2:3 ratio photo! That must be fun...

 

There was a test shoot on Luminous Landscape shooting the same scene with a whole bunch of formats including a 39 megapixel back and 4X5 film. Although there are many who can nitpick the technicalities of the test, in the real world nothing is perfect either and it could give you a very good idea before you start thinking about the stage where you invite a representative of one of those companies to come round for a day worth of demonstration in your studio where you can make your own detailed tests based on your method of shooting, your workflow, etc. 39 megapixel backs are far too expensive to even bother discussing on forums to be honest, they need serious test driving to see whether they are as good or better - for you.

 

Some like Ellis will say that a digital file needs far less 'megapixels' to achive comparible quality due to the cleanliness of the files. Some may say that the look of the film is a necessity that cannot be ignored. You might make a comparison based on the best drum scan in the world but if you don't do all your work that way then it's immaterial. You may decide that you can achieve sufficient (a subjective word if there ever was one!) resolution and tonality for prints 'X' size from a digital back even if it were to have less resolution than a certain film size. It may be immaterial if the economics of a MFDB won't work in your favour anyway.

 

The only one who can really answer your question is you, it's just too subjective. Go out there, make the tests and then screw what anyone else says or thinks afterwards whatever your choice, what the heck do they know about your specific circumstances or what YOUR eye sees?

[erlend sæteren]

I have recently tested fuji 6x8 and provia 100f versus 34mpix digital. The 6x8 looked better. The goal was 120cmx150cm inkjet. Got more testing to do. Might improve both skanning and digitalrawprosessing. I prefer a 5000dpi filmscan over an interpolated digitalfile.

 

About ppi. I am not able to see problems at 135ppi without glasses on my nose. At 90 ppi I can see jagging if I look close, but Ihave to look close. So I think that 100 ppi is ok for very large prints.

[arlen_c._elkins]

I can't speak for a 39mp Digital back, but several years ago I was shooting a setup of a copy camera we were selling which used a Dicomed/Betterlight scan back and I was using 4x5 film. As I was using Quartz lights for illumination, I thought I would shoot the same image with the scanback and compare to the film images.

 

I had the Betterlight image reproduced at 8x10 using a Fuji Pictography and the Film images processed and enlarged normally. I was amazed by how much better the digital image was when compared to the film image. More resolution and more detail in both highlights and shadows.

 

I showed both enlargments to other photographer friends of mine, told them one image is digital and the other is film, asked them to pick which one is digital and which one is film. Everyone picked the digital image as better of the two images, but they all assumed it was the film image, as "film is better than digital".

 

Of course the scanback has much more resolution and a larger image area than the MF digital back, but I would assume the digital back would still best the film on dynamic range and as the film would have to be scanned, not only would the quality of the scanner come into the equasion, but also the fact that the image from the digital back is 1st generation, the film image is 2nd generation.

[andrew_keam]

Thanks to everyone for contributing. Despite the variance in opinions they all seem

equally valid and like Ben Rubinstein has stated, I guess I need to do some real testing

and see what is acceptable for me. The only problem with this is getting my hands on

some of the equipment talked about ( rental of digital medium format definitely aint cheap

). As for the Burtynsky prints i saw in Fremantle, they were a mixture of ( mainly ) aerial

shots and landscapes and were specifically commissioned for the festival. As I said

before, it stated in the local paper that these were shot digitally and after shooting 5 x 4

aerial, I dont see why you wouldnt. He also had some of his earlier 'Shipbreaking' series

there and they did look different. They did have a certain nicer 'film' look to them but the

new series did actually look sharper, despite them being bigger. I would assume at his

level, that he would have been using top of the line in both regards ( digital and film ) even

though the 'shipbreaking' series was significantly older. What i am actually interested in is

how you actually bring a digital file up to the size of 50inch x 40 inch? Is it just a matter of

bringing the ppi down to a smaller enough size to allow this or is there something I am

missing?

[User_503771]

Andrew, I think you need to backtrack, maybe read all these posts again and make sure you understand what each is saying -- then run some tests for yourself at smaller sizes to determine what pixel dimensions will allow for what you feel is good print quality. Then you can do the math from there.

 

To give one direct answer to the pixel question: You can either start out with a large enough file to give a certain printed resolution per inch, or you can use software to add pixels to give a more acceptable final rendering at large sizes.

 

But this issue has so many variables involved, from the shooting to the scanning to the after-processing in Photoshop (or equivalent).

 

I think Ellis Vener made a very good point though, above: "We can bandy theory all we want to but the thing that settles any argument is looking at prints. And even then that settles very little because digital photographs originals and film photographs are really two different mediums with their own inherent positive and negative characteristics."

 

I wish I had put it that way myself! In discussions with friends, I've been saying for a while now that I think that film and digital really are two different media, and that I've been really enjoying my own experiments/experiences in scanning color film and printing digitally -- which creates a third, combination medium! And I've also been maintaining that the print is the result we're all after, isn't it?

 

But rest assured! No matter how many opinions or facts there are about all this, the ultimate judge has to be the artist, regarding how the work is to be produced and how it is to look. It may be praised or panned, and the artist can always change his or her mind, but the artist is always the first person who has to be happy with the print.

 

I know this doesn't give any final answer -- except to say that you are the answer!

 

But really, I'm convinced this is true. As an example of this, just take a look at a painting by Monet, then one by Warhol. Both are famous artists of high repute.

 

Or take a look at prints by Ansel Adams, then prints by Garry Winogrand.

 

See what I mean? So run some print tests, see what happens.

[roger krueger]

The only online test I've seen was at Luminous Landscape, and he engaged in massive

sabotage by downsampling the LF scan to the 39mp pixel count. So they came out

pretty much the same. Well duh!

 

So much depends on your workflow I'll join in with those saying to do your own tests. I

know I can blow the doors off my 1dsII with my 6x9, 4x5 should have an even larger

advantage over 39mp. Wish I had a 39mp to test...

 

But there are a lot more ways to screw up film, especially in scanning. It's also harder

to get the post right--you have to be more careful sharpening due to grain, even more

so if an insufficient-res scan has aliased the grain to be bigger than it really is. So,

YMMV.

 

In general, digital's strength is glassy, noiseless smoothness, big film excels at

resolving high-frequency detail.

[ben_rubinstein___mancheste1664880652]

Downsampling makes pictures look sharper so I don't see how it was sabotage eventhough I have little to no respect left for MR after some of the fudging he has done in the past to avoid admitting that he was wrong.

[erlend sæteren]

About the difficulties when testing. If you are used to scanning 4x5 inch Velvia in super quality, you will realize that it is impossible to get the same nice colours from a digital file. But if you are very good and works a lot with the right rawconverter you will after a while come really close. That was colourspeaking. I was able to get close in Lightroom.

About expectations. I will expect that the film will be better due to its larger size. How much larger it must be to be better you can find out. I will try to do some test during the summer to find out myself. So far it looks like twice the filmsize is better than digital, but this is only a firstglance nonscientific evaluation.

[danielleetaylor]

<P><i>The only online test I've seen was at Luminous Landscape, and he engaged in massive sabotage by downsampling the LF scan to the 39mp pixel count. So they came out pretty much the same. Well duh!</i></P>

<P></P>

<P>As I understand it, he resized both for printing to 30x40 inches at 360 ppi. So he was actually enlarging the 39 MP digital file to a 155 MP final size. I don't know if he had to shrink or enlarge the film scan to obtain the same size and resolution.</P>

<P></P>

<P>The magnified crops on the web page show that the film did resolve more detail. He said as much, but noted that he would be happy to exhibit a 30x40 inch print from the 39 MP back. Just to clarify, the comparison was written up by Charles Cramer, not Reichmann, and Cramer did more comparisons on his web site (http://www.outbackphoto.com/artofraw/raw_28/essay.html).</P>

[roger krueger]

Ben: Downsampling may make an image pixel-for-pixel sharper, but it destroys fine

detail.

<br><br>

Daniel--I hadn't seen the Charles Cramer article, yeah, that appears to be using a lot

more resolution. Not sure why he couldn't have made the images the same size on

screen, though. Was the 4x5 composed that much looser? And he doesn't say what he

scanned at, only what he interpolated to. Probably an insignificant omission, but if MR

made the same omssion I'd be suspicious.

<br><br>

But the Luminous Landscape test I was referring to:--<a href="http://www.luminous-

landscape.com/essays/back-testing.shtml">Measuring Megabytes</a>

<br><br>

has this to say about how they arrived at final sizes:

<br><br>

<i>Comparing different sized images:

All captures were made such that the 15.5 inches wide "goal posts" in the test scene

approximately filled the width of the frame, but to make comparisons I still had to

compensate for differences in image size and offset. To accomplish this I started by

setting Photoshop's image interpolation preferences to bicubic. Then I chose the

cropping tool and set the crop size to 8000 x 6000 pixels at 360 dpi. Then for each

developed capture I stretched the crop width to just touch the "goal posts" on the left

and right. After setting the crop width, I then centered the crop on the color checker

cross bar. After performing the crop I dragged each of the nine images into a new layer

of a Photoshop comparison document. </i>

<br><br>

So my memory was slightly faulty, they went to 48mp, not 39mp, but that's still

well under 2000ppi on the 4x5 film, easily far enough from preserving all the detail to

count as sabotage in my book. And this one does look like it was written by

Reichmann, not Cramer.

[erlend sæteren]

Grain versus interpolation artifacts.

I have seen a 34 mpix digitalfile interpolated to ca 900megabyte in 8 bit. I think the

pattern constructed by interpolation was ugly, but I believe it must be possible to

interpolate better. Scanning a 4x5 velvia to the same size shows some grain, but that

is not a problem. Actually I can accept much larger grains, because I expect grains in

a photography. In fact I believe that a grainy 35mm Tri x in Rodinal (Ralph Gibson

style)can produce a nicer 2meter print than an ugly interpolated 39mpix digitalfile.

[gerard_bynre]

Work-flow is the critical factor - how can you get the best balance of control, quality,

consistency and usability at a price you can afford? That will vary for everybody. For

me it's with 5x4 film, because I can't justify the outlay for the digital backs & new

lenses that compare to my film set-up. In time that might change.

 

One argument that's been trotted out time and time again which I think is a total red

herring is the 'digital is first generation, whereas a scan from film is 2nd generation".

If one wants to be philosophical about it, it's all 2nd generation for starters - its a

photo, stupid! To talk more practically, I've found that 16 bit drum scans at

sufficient resolution can out-resolve my T-max 100 5x4 originals, and I feel that the

tonal range rendered in the scan far exceeds what's needed to produce a print, on

the proviso that the original neg was properly exposed and developed. So where's

the 'generational' loss in quality? Recently I had a 5x4 T-max 100 neg scanned, -

some sharpening applied to a small area I felt needed to be sharper, and written out

onto an 10 x 8 neg. (FP4) using a Kodak film writer. That's at least 2 more

generations then the original neg, and printed side by side on 20 x 24 fibre paper the

10 x 8 is better - why? well just because it's sharper where I wanted it to be sharper!

 

You've got to figure your own work-flow. There are different ways of producing what

you want, and multiple factors at work.

[gerard_bynre]

"That's at least 2 more generations then the original neg, and printed side by side on 20

x 24 fibre paper the 10 x 8 is better - why? well just because it's sharper where I

wanted it to be sharper!"

 

Or to be clearer - there was no visible 'loss' over the multiple generations, but there

was a visible advantage to the 10 x 8 -the sharpness problem (human error) was

resolved.

[kelly_flanigan1]

Here is a 35 megapixel shot with a Phase One. The camera was a 4x5 speed graphic. The lens was an Apo ronar 150mm F9 off one of our old copy cameras. The shot was captured before any digital lenses were created; designed or marketed. The tethered computer for the scsi card was a 1996 vintage Pentium Pro with a 200mhz Cpu and 512 megs of ram and windows 2000. <BR><BR>NONE OF THIS EQUIPMENT IS NEW; and the captured 35 megapixel image captures fine details; cracks in the fireplace mortar. The canned dogma on photo.net is that buying a new lens will make a worlds difference in results. Here the 150mm Ronar and 135mm Componon have been used for over a decade now to scan fine artwork with the 35 and 50 megpixel scan backs.<BR><BR> With a modern 39megapixel back on a MF camera; the pixel pitch is finer; and here a more modern lens might help.<BR><BR>One can debate paper versus plastic; digital versus film; wine versus beer until the end of time. Real paying clients dont have the time for this confusion of the image maker; they have actual deadlines. <BR><BR>One person one this thread might live next door to a fine 4x5 lab that also has fine drum scanning and optical enlargement; another might have to FedEx all 4x5 lab work for a client that has a rush job. Follow where the money is flowing. Folks who sell lenses will say one must use a viewcamera and digital lens. Folks who do scanning will hawk their services; scanner holders; drum scanning. Folks who sell cameras will hawk wooden bodies; film and holders. You have to find what works for your applicatuon and rise above the marketing BS and froth.<BR><BR> Not all high end capture is about resolution; many folks jsut want fine tonality; smoothness thats from moderate; conservative; low enlargement ratios. With artwork one can place color reference swatches to tweak the colors. <BR><BR><img src="http://i4.photobucket.com/albums/y148/ektar/scanback/PhaseOneScanBack.gif?t=1211637047"><BR><BR><img src="http://i4.photobucket.com/albums/y148/ektar/150mmRonar/Ronar150mmFull.jpg?t=1211637617"><BR><BR><img src="http://i4.photobucket.com/albums/y148/ektar/150mmRonar/Ronar150mmWWVCLOCK.jpg?t=1211637679"><BR><BR><img src="http://i4.photobucket.com/albums/y148/ektar/150mmRonar/Ronar150mmStones.jpg?t=1211637733"><BR><BR><img src="http://i4.photobucket.com/albums/y148/ektar/150mmRonar/Ronar150mmDOG.jpg?t=1211637796"><BR><BR><BR>A 4x5 trany can hold more info than a 35 or 39 megapixel back. With a digital back one can tweak focus spot one and cheat the ground glass to film plane errors many folks have. One can also provide clients with images; they really might have deadlines. May Company in Los Angeles was using the Phase One 4x5 scan backs 11 years ago for clothing shots.<BR><BR>One can debate and worry about 4x5 ffilm versus an XYZ digital capture until the end of time. Folks can post best case film scanned images of maps shot with cameras attached to granite blocks. Amateurs often get lost in this debate becuase they have no specific goals; not clients with deadlines. The digital back is always going to be expensive to folks with no actual clients to help pay for the cute tools. <BR><BR>If you are out backbacking then that fine landscape shot can be done with 4x5 trany film; you can drum scan the best case great ones. With an advert agency client one might not have the luxury of a massive time frame to shoot film and drum scan it.<BR><BR>There is alot of stuff shot today with high end full frame dslrs that is very acceptable for clients; at a radically lowere cost than using a LF; MF or 4x5 scanned image.

[kelly_flanigan1]

One rarely reads anything on these film versus digtial threads about design margin; ie having alot more than is required for the task. With an artwork settup here I want more than just barely capturing an original details; I want to have abit of margin so that the task is easier. the whole world doesnt revolve around best case thinking.

[jed_wormhoudt1]

While the image quality of 4x5 film versus 39MP digital debate has received

considerable attention- another, perhaps equally important, might also be usefully

addressed: the relative performance of camera movements when working with a 4x5

inch sheet versus the very much smaller (very slightly sub 6x4.5cm) P45 sensor. I

don't know whether or not any of the Burtynsky work was done with movements- but

certainly Adams and Sexton have used them with film. Perhaps someone has direct

experience of a P45 on a 4x5, using tilts & swings, and shifts- and could comment.

Regards, Jed Wormhoudt