Site Sponsors
Results 1 to 14 of 14

Thread: Translating square inches to megapixels

  1. #1
    Senior Member Lars's Avatar
    Join Date
    Nov 2007
    Location
    Sunnyvale, California
    Posts
    1,811
    Post Thanks / Like
    Images
    19

    Translating square inches to megapixels

    A question that sometimes comes up in a discussion re large format photography - and is related to the title of this forum - is "How many megapixels can you capture with that 8x10 of yours?"

    While this is of course not the only consideration for using large format, the question does have some merit. Some scientific bench tests using superfine grain b/w film and lenses at their optimal aperture give a lpm value corresponding to 800 Mpx on 8x10" film, but in real life we rarely shoot measurement targets and brick walls.

    I have recently examined some of my drum scans, scanned at 2000 dpi for a 16000 x 20000 px image. The finest details I can find are edges between 1 and 2 pixels wide (no sharpening applied). Another example is grass straws which are down to 3 pixels wide. (As far as I can recall, these scans were not made using the finest aperture, so it should be possible to resolve slightly more at the expense of more graininess.)

    This is of course only when the focus plane is set correctly and the subject is not moving. As some of you know, this is not always easy (or possible) with larger formats. In addition, landscape and photography is three-dimensional in its very nature (pun intended).

    So... I figure that a real-life number for my 2000dpi drum scans would be somewhere between 320 Mpx (16K x 20K) and 80 Mpx (half of that resolution). (The original film does indeed have plenty more information, but then scans get ridiculously large, a 4000dpi scan lands at 8 GB and has no real practical application except capturing what is mostly film grain.)

    Of course, such resolution is rarely needed in actual output, but that is another discussion.

    Any thoughts? What is a proper number to use at a coffee table discussion?

    Lars

  2. #2
    Sr. Administrator Jack's Avatar
    Join Date
    Oct 2007
    Location
    Los Altos, CA
    Posts
    10,486
    Post Thanks / Like
    Images
    1031

    Re: Translating square inches to megapixels

    Well if we use your 2 pixels as the limit in a 2000 LPI scan, we are essentially saying the lens resolved 1000 lines total or 500 line-pairs per inch. Dividing by 25.4, this translates to 20 LPmm (line-pairs per mm). Looking at it empirically from this point of view, even pretty basic lenses by LF standards can achieve that level of resolution. (However, I certainly agree that all of the variables are usually working directly against the LF photographer actually getting all of a lens' resolving power on the film! )

    Coming at it from a different angle, I would suggest that most modern 4x5 lenses can achieve 50 LPmm at the center and 35 to 40 at the edges, while 8x10 may drop a bit to 45 centrally, 30 at the edges. If we average those values we get an approximate average image resolution across the film plane of 45 LPmm for a 4x5 and 38 LPmm for 8x10. Multiplying those by 25.4 to derive Lp/inch, we get 1143 LPin for 4x5 and 965 LPin for 8x10. Doublnig these to get to LPI, we get roughly 2300 LPI for 4x5 and 1830 LPI respectively on 8x10. These seem like fair numbers to me, especially considering my own results show them under the resolution limits for most emulsions. (For most film, I've noted that scanning above 2400 LPI generates little more than better detail of the film grain, so I agree with you on this point too!)

    So for 4x5, 2300^2 is roughly 5.3 MP/sq-in of film. Taking this to the total, we have roughly 19 square inches of useable film area due to the slight loss at the border. This gives us a total of 19*5.3 MP or essentially 100MP of equivalent capture resolution on 4x5 film.

    Doing the same for 8x10 we square 1830 and get 3.35 MP per sq-in of 8x10 film. Accounting for borders, we have about 77 square inches of usable film area, times 3.35 MP/sq-in for a total capture resolution of about 258 MP.

    So these are the range I'd put them:
    *4x5 ~= 100 MP effective resolution
    *8x10 ~= 260 MP effective resolution
    Obvously pretty darn close to the range you stated above
    Jack
    home: www.getdpi.com

    "Perfection is not attainable. But if we chase perfection, we can catch excellence."

  3. #3
    Member
    Join Date
    Dec 2007
    Location
    Bath, U.K.
    Posts
    148
    Post Thanks / Like

    Re: Translating square inches to megapixels

    Quote Originally Posted by Lars Vinberg View Post
    A question that sometimes comes up in a discussion re large format photography - and is related to the title of this forum - is "How many megapixels can you capture with that 8x10 of yours?"...
    Any thoughts? What is a proper number to use at a coffee table discussion?

    Lars
    Well my non-scientific argument (for 54 in the not too distant future) would go along the lines of...

    "One megapixel is not the same as another megapixel"...and of course then go on to explain that even though their camera phone can produce a 100mb file its not the same quality as their compact camera 100mb which is not the same quality...etc etc!

    Then they'd go out and quickly shoot some great pictures whilst I was still levelling my 54 on the tripod!

    Marc

  4. #4
    FrankF
    Guest

    Re: Translating square inches to megapixels

    Quote Originally Posted by Lars Vinberg View Post
    A question that sometimes comes up in a discussion re large format photography - and is related to the title of this forum - is "How many megapixels can you capture with that 8x10 of yours?"

    A friend of mine told me of his experience:

    One 4x5-customer of his was ready for a test a while ago:

    The chose the same table top setup and lighting and ran the Nikon D2X with Micro Nikkor 70-180 against a 4x5 Linhof with some fine Schneider or Rodenstock Glass of that time. They had the film professionally drumscanned, the D2X-RAW processed in Nikon Capture + Photoshop and sent both files to the best lab they could find to make big prints, as far as I remember it was more than one meter wide.

    The D2X-Files delivered by far the better results. The test Betterlight scanback vs. Canon 1DsMK2 deivered about the same quality (on screen).

    What do we conclude?

    4x5 worse than 12 Megapixels? Nope

    I say: The lenses make the whole difference. There simply is no Digitar/Digaron that covers 4x5 with movements not to speak of 8x10, so the current Nikkors win in 24x36 format and the Digitar/Digarons win in 36x48 format, the Leica S2 line might win in the new 30x45 domain.

    My guess: If you take 4x5 or 8x10 with the best available glass and film as of today, and then wetlab it through to the print (no digital interludum) and then scan the film, digital will have some tough time. If you built a 10.000-Euro-4x5-Digaron/Digital (RED will have to do this soon), film will still win.

    My question: If you build a 4x5 chip based on the Nikon-D3-pixelquality, who will win with the same glas in front???

    Frank

  5. #5
    Senior Member Lars's Avatar
    Join Date
    Nov 2007
    Location
    Sunnyvale, California
    Posts
    1,811
    Post Thanks / Like
    Images
    19

    Re: Translating square inches to megapixels

    Frank,

    That sounds like a test that was not properly executed. I've had a D2x and some fine Nikkor glass, and I also shoot LF 8x10", 4x5", 6x17cm, 6x9cm. WRT resolution the D2x is like a toy compared to those formats. Good LF lenses used properly outresolve just about anything in total resolving power over the entire image circle.

    Re lenses, my Apo-Digitar 5.6/120 from Schneider does cover 4x5 (I use it on 6x17 cm), and easily outresolves most Nikkors - this means some 3000 dpi in resolving power, or some 180 megapixels on a 4x5 (if you can find the film to match). I sold my Apo-Ronar 480 for 8x10" to get a Cooke XVa - both lenses more or less match a Nikon 50/1.8 in lp/mm, and both lenses cover 400 mm or more in image circle. That's gigapixels in resolving power.

    A conclusion that 4x5 cannot resolve more than 12 megapixels is not just slightly flawed, it's not even close. The testers must have made some serious mistakes.

    Lars
    Monochrome: http://mochro.com

  6. #6
    Senior Member
    Join Date
    Dec 2007
    Location
    Reykjavik, Iceland
    Posts
    2,310
    Post Thanks / Like
    Images
    9

    Re: Translating square inches to megapixels

    Agreed. I like digital too, but 4x5 is so far beyond my D3, or even my E54LV (22mp) that it is not worth even testing. It's just whether you need that capability, whether it works for your application, and whether you have the technical equipment required to extract that data (the simplest and most effective is a good enlarger and lens). I also don't doubt that digital may look "better" in some ways than 4x5 at small print sizes. Digital has a different look and usually 35mm lenses are contrastier than 4x5 lenses, so at even moderate image sizes, it may have more pop. But if you start making really large prints, the difference will be apparent (assuming proper technique).
    My photos are here: http://www.stuartrichardson.com and more recent work here: http://stuartrichardson.tumblr.com/ Please have a look at my book!
    My lab is here: http://www.customphotolab.is and on facebook

  7. #7
    FrankF
    Guest

    Re: Translating square inches to megapixels

    Thank you Lars & Stuart!

    I just put together my digital 4x5 via stitching D3-Files for LF-Tabletops.

    I guess for a full scan I need 18 shots, which is not confortable at all, but gives me full movements at 180 MP.

    For most applications I do not need the full scan, just three horizontal in the middle which should help me to a nice 36x65 square millimeters.

    I chose the following components: Sinar F2, Nikon D3, Nikkor AM ED 5.6/120mm LF-APO-Macro, just-together-Nikon-F-adapter. To start with this seems sensible and economical to me.

    I think it is very interesting to me that the Schneider APO-Digitar 5.6/120mm covers more than the 48mm-format.

    I just decided against the Rodenstock APO-Sironar digital 5.6/90mm because it does not cover 4x5 AND does not play with the just-together-adapter.


    All I read in this forum and I read a lot seems very knowlegable and polite.


    I still value the practical experience of my friend very much.


    What do you think of my hypothesis that an all analougue chain plus scan from the print gives much more than scanning the negative and then printing?

    Frank

  8. #8
    FrankF
    Guest

    Re: Translating square inches to megapixels

    Quote Originally Posted by Stuart Richardson View Post
    I also don't doubt that digital may look "better" in some ways than 4x5 at small print sizes. Digital has a different look and usually 35mm lenses are contrastier than 4x5 lenses, so at even moderate image sizes, it may have more pop.

    Just what he said.

    The customer wanted large prints with relatively low resolution for his booth at a trade fair. The D2X had more "POP" as you say, more contrast, was flashier. The customer chose the "digital look" and with the D3 instead of the D2X now even chooses to print the images for the booth up to 2.5 meters wide.

    I have a 1-Meter-print here of a 1250-ASA-Shot from my D70 I took in a pigs stable with all the lively little pigs running around my feet. Grweat shot, no chance for Large Format, but possible with the Leica S2 I guess.



    So, to come back to the opening question of the thread:

    It depends. You get 0 Megapixels from 4x5 in that piggy-stable-situation but might get 180 Megapixels with very patient table top items in a studio with big light.

    Frank

  9. #9
    Senior Member
    Join Date
    Dec 2007
    Location
    Reykjavik, Iceland
    Posts
    2,310
    Post Thanks / Like
    Images
    9

    Re: Translating square inches to megapixels

    "What do you think of my hypothesis that an all analougue chain plus scan from the print gives much more than scanning the negative and then printing?"

    A lot of people disagree with this these days, but I find myself agreeing. I use an Imacon 646, as well as a traditional enlarger. I find that when used with a properly aligned enlarger, glass negative carrier and top notch enlarging lens (preferably apo), the results I get in analog fiber prints are still better than scanned images, even with full resolution on the imacon. There are always other considerations though, such as consistency (if necessary), ease, speed, the need for retouching and localized edits etc. But in my opinion, if you have a good negative and a very good analog imaging chain, it will usually look better. In the same way that the average CD sounds better than the average LP, but a great LP with a top notch turntable setup will sound better than any CD...analog is analog -- there is more "grain", but also higher fidelity to the original scene (whether auditory or visual).
    My photos are here: http://www.stuartrichardson.com and more recent work here: http://stuartrichardson.tumblr.com/ Please have a look at my book!
    My lab is here: http://www.customphotolab.is and on facebook

  10. #10
    Senior Member Lars's Avatar
    Join Date
    Nov 2007
    Location
    Sunnyvale, California
    Posts
    1,811
    Post Thanks / Like
    Images
    19

    Re: Translating square inches to megapixels

    Quote Originally Posted by FrankF View Post
    I think it is very interesting to me that the Schneider APO-Digitar 5.6/120mm covers more than the 48mm-format.
    ...
    What do you think of my hypothesis that an all analougue chain plus scan from the print gives much more than scanning the negative and then printing?
    Frank,
    The 120 Apo-Digitar (not the macro) is specified by Schneider as covering 110 or 125 mm depending on which specs your read. These specs are very strict WRT focus plane flatness and resolution. Outside this circle, out to 190 mm, it illuminates but flatness and resolution is not guaranteed. For my 6x17 work this is normally not a problem as I rarely shoot with it wide open. on 4x5 corners might possibly be slightly soft although I have never seen a hint of a problem. This lens resolves so much more than a 4x5 sheet needs. At the same time, I suspect the current 120 Symmar-L to be very close in performance. and coverage. The interesting thing with both these lenses is that the rear element is as wide as the front, this is what allows for the wide image circle. The other Apo-Digitars are not like that.

    If you scan a print you might gain some subtle detail if the print is huge, but the loss of shadow detail will be drastic. Not a good idea unless you want to reproduce a finsihed darkroom print.
    Monochrome: http://mochro.com

  11. #11
    FrankF
    Guest

    Re: Translating square inches to megapixels

    Quote Originally Posted by Lars Vinberg View Post
    The interesting thing with both these lenses is that the rear element is as wide as the front, this is what allows for the wide image circle. The other Apo-Digitars are not like that.
    Lars,

    Symmar = symmetrical design.

    All LF-macros are symmetrical, but most do not cover 4x5 up to infinity. The symmetrical 120-APO-Macro-Nikkor I chose covers 23,5 cm at 1:1, but does not cover 4x5 at infinity. For my tabletop work it is the right choice though. I hope it will have the resolving power to feed the D3-Pixels with enogh information.

    Frank

  12. #12
    Senior Member
    Join Date
    Dec 2007
    Location
    Reykjavik, Iceland
    Posts
    2,310
    Post Thanks / Like
    Images
    9

    Re: Translating square inches to megapixels

    Just another consideration -- it is one thing to make a huge scan and print out a 30x40 or larger print on an Epson 9800, it is another entirely to print 30x40 or larger in a conventional darkroom -- you either need very tall ceilings, ability to project onto the floor, or a horizontal enlarger. Once you have that, you still need to either roll the print up and process it in long narrow tanks, or you need huge trays (and the space they take up). Either way, you need tons of chemicals. I have given up on printing larger than 20x24 in the analog darkroom (or at least in a darkroom I would build for myself...if I can rent one that can do it, that's fine). Jumping size in the conventional darkroom is a exponential leap in difficulty. With digital printing, it's the same button...you just need access to a big inkjet, Chromira or lightjet.
    My photos are here: http://www.stuartrichardson.com and more recent work here: http://stuartrichardson.tumblr.com/ Please have a look at my book!
    My lab is here: http://www.customphotolab.is and on facebook

  13. #13
    Sr. Administrator Jack's Avatar
    Join Date
    Oct 2007
    Location
    Los Altos, CA
    Posts
    10,486
    Post Thanks / Like
    Images
    1031

    Re: Translating square inches to megapixels

    Quote Originally Posted by FrankF View Post

    Symmar = symmetrical design.
    If we want to split hairs, original Symmars were fully symmetrical, but the later image taking designs are now sort of quasi-symmetrical plasmats instead. And during the course of revisions, we went from front and rear element being the same size since they were symmetrical, to rear elements being smaller since they were now only quasi-symmetrical, and now back to rear elements being unnecessarily larger in the quasi-symmetrical design to allow the same filter size at each end. What the quasi function really means, is you can reverse the cells on a contemporary non-macro plasmat and get better optical performance at sub 2:1 reproduction ratios.
    Jack
    home: www.getdpi.com

    "Perfection is not attainable. But if we chase perfection, we can catch excellence."

  14. #14
    Senior Subscriber Member
    Join Date
    Nov 2007
    Location
    San Francisco, CA
    Posts
    1,306
    Post Thanks / Like

    Re: Translating square inches to megapixels

    Quote Originally Posted by Lars Vinberg View Post
    Frank,
    The 120 Apo-Digitar (not the macro) is specified by Schneider as covering 110 or 125 mm depending on which specs your read.
    This is at infinity though, with a level back. At anything closer you get a progressively bigger image circle, and if you tilt the back you stretch the image a little. At least I personally can't remember the last time I shot 4x5 focused to infinity.

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •