Luminous Landscape Forum
Equipment & Techniques => Digital Cameras & Shooting Techniques => Topic started by: button on December 29, 2008, 04:53:57 pm
-
Could someone who has experience with 4x5 photography as well as a high megapixel 35mm DSLR (ie, canon 5d2) provide an estimate of the number of DSLR frames required to achieve the resolution of 4x5 film (such as velvia 50)? This assumes a given object (fixed frame) and same perspective, with allowance for using longer lenses on the DSLR. For example, I could stitch a 4 frame rectangle with a 50mm lens, or a 16 frame rectangle of the same object with a longer lens. I guess this question boils down to this: how many DSLR megapixels are needed to equal 4x5 resolution?
Thanks,
John
-
This may be helpful...
http://www.luminous-landscape.com/essays/back-testing.shtml (http://www.luminous-landscape.com/essays/back-testing.shtml)
Michael
-
fwiw, I was in SW USA in Nov this year, shooting with a D300, only a mere 12mp. Came upon a scene where I wanted more then just an HDR or NEF file, and decided to use my 28mm ais Nikor, instead of a zoom and shot a 5 frame panorama. I had been there last bout 10 yrs ago with a Linhof Super Tech V and lenses. Overlapping the frames bout 30-40%, merging in CS3, acheived a file about 200mb. It is below in jpg form, looks at least as detailed as the 4x5 sourced 30x40+ prints[from 10 yrs ago] I have in my office when I have made cropped images to compare. I figure this combined image amounts to about 32-36MP in scale. There is some artifacting on top right center at edge that is not visible in prints, but essentially, after seeing this, it has made me want a 24mp camera more. I figure I would then be able to shoot a two frame stitch and come close to the same pixel dimensions and quality from the previous images I shot with 4x5. Pixel peepers need not comment, this is from looking at prints from 18 inches or further, which is still close for a print this size. YMMV...
-
Thanks guys!
Michael, I've been looking for that shootout now for days (I guess I wasn't using the right search terms)- much appreciated.
Don- that's very helpful. I just got the 5D2, hoping (like you) to cut down on my frames for stitches. I'm hoping for 4, and no more than six. My goal is to get a sequence done in a minute or less, to catch that fleeting magical light. I figure that 60-100 megapixels (allowing for overlap) should at least come close.
John
-
fwiw, I was in SW USA in Nov this year, shooting with a D300, only a mere 12mp. Came upon a scene where I wanted more then just an HDR or NEF file, and decided to use my 28mm ais Nikor, instead of a zoom and shot a 5 frame panorama. I had been there last bout 10 yrs ago with a Linhof Super Tech V and lenses. Overlapping the frames bout 30-40%, merging in CS3, acheived a file about 200mb. It is below in jpg form, looks at least as detailed as the 4x5 sourced 30x40+ prints[from 10 yrs ago] I have in my office when I have made cropped images to compare. I figure this combined image amounts to about 32-36MP in scale. There is some artifacting on top right center at edge that is not visible in prints, but essentially, after seeing this, it has made me want a 24mp camera more. I figure I would then be able to shoot a two frame stitch and come close to the same pixel dimensions and quality from the previous images I shot with 4x5. Pixel peepers need not comment, this is from looking at prints from 18 inches or further, which is still close for a print this size. YMMV...
I Have a 1DsmkIII, 1DsmkII and Kodak SLR/n, I know I'm an odd ball and see things differently, but the more MP's I have in a camera the better film looks to me. That AA filter does more than reduce moire, I think it takes away some picture depth, it levels the field between the sharp and less sharp. MP's when they reach the DR end can go ugly in the transition, film goes black or clear. Resolution wise I keep trying various Canon lenses, I'm mostly disappointed with the results, either distortion edge performance etc. Plus I would find stitching a 1/2 dozen frames more of a pain than scanning.
Shooting 5x4 is a different experience to a DSLr, the pictures you take will be different because your state of mind changes, your priority changes. No doubt if I had a Phaseone my opinion could change, at the moment I don't see 35mm digi's as any answer to the journey and creation of making LF images, lighter to carry, cheaper (possibly) to run and easier yes........... as good or better? not to me.
Kevin.
-
I Have a 1DsmkIII, 1DsmkII and Kodak SLR/n, I know I'm an odd ball and see things differently, but the more MP's I have in a camera the better film looks to me. That AA filter does more than reduce moire, I think it takes away some picture depth, it levels the field between the sharp and less sharp. MP's when they reach the DR end can go ugly in the transition, film goes black or clear. Resolution wise I keep trying various Canon lenses, I'm mostly disappointed with the results, either distortion edge performance etc. Plus I would find stitching a 1/2 dozen frames more of a pain than scanning.
Shooting 5x4 is a different experience to a DSLr, the pictures you take will be different because your state of mind changes, your priority changes. No doubt if I had a Phaseone my opinion could change, at the moment I don't see 35mm digi's as any answer to the journey and creation of making LF images, lighter to carry, cheaper (possibly) to run and easier yes........... as good or better? not to me.
Kevin.
That's a point of view that I wouldn't have expected, Kevin, and now I'll have to A/B my 5D2 against my 40D to try to see what you are seeing. I assume that your comments refer to single frames from your DSLRs. What is your impression of the results you've obtained from stitching?
John
-
John - to say 4x5 is a different experience is obvious and true, used 4x5 for over a decade and MF for two decades. That said, it is a matter of taste and personal choice, I for one would not go back to film ever, and a mfdb or 4x5 digital back is way too much. For me, I like having a more portable system, more responsive system, a more multipurpose system, and a dslr is thing for me. As I said, YMMV. I think you will be good with your 5D2 and less stitching. Does not mean you can not stitch when you want to. I frankly could go either Canon or Nikon, have pretty comparable systems in both, but have personally had too many focusing issues with Canon, went through 3 5d bodies before I got one that focussed predictably, consistently. MF adjustment is just not my thing, I hope Nikons inclusion of this function is not a sign that future models will be more problematic. So far my d300 is fine with all my lenses I currently own. Personally I think 60-100MP is a bit overkill, but you may view you prints closer than I do. Although friends of mine think I am way to picky as is. I can tell you just from what i said before, that 30+mp is fine for 30x45 prints, the image I included above looks great at 30x60 inches. Looking at cropped sample images, at this size, I can only imagine what a image with twice the pixels would look like. Hope you 5D2 works for you, you will for sure, be able to stitch less and still acheive large file sizes, at this point I am waiting for a D700x model to arrive, to accomplish the same thing. IF you are surely, going for 60-100mp, you better make sure, your computer is up to top form in RAM and hard drive space....have fun, going to Greece in March and will surely shoot more stitched images and hdr....with whatever camera I have by then....
-
I know this has been done to death, but i have a 1DS3 with some very fine Zeiss glass and have to say if i had to rely on this to produce fine prints for the wall that i was proud of i think i would change my career. Honestly my Pentax 67 pisses all over it and looks beautiful, and the 1DS3 gets used for stock only, and even for stock i'm shooting more and more colour neg in the Pentax because even on the Getty website i can see how much better the film pics look, but economically it makes sense to have both, and that's the way of the world i'm afraid, but out of choice i know which one i'd be sticking with.
-
I know this has been done to death, but i have a 1DS3 with some very fine Zeiss glass and have to say if i had to rely on this to produce fine prints for the wall that i was proud of i think i would change my career. Honestly my Pentax 67 pisses all over it and looks beautiful, and the 1DS3 gets used for stock only, and even for stock i'm shooting more and more colour neg in the Pentax because even on the Getty website i can see how much better the film pics look, but economically it makes sense to have both, and that's the way of the world i'm afraid, but out of choice i know which one i'd be sticking with.
Very good points. Similarly, my Mamiya 7 used to beat the heck out of DSLR. Stitching though did help a bit - this is a D300 HDR stitch (click here (http://flickr.com/photos/markhout/2845147470/) for tech data). I have tried HDR to get the same smooth dynamic range as I would have with the M7 or 4x5. I'm not there yet, but I'm getting close. Wonder if a full frame would have helped here.
(http://farm4.static.flickr.com/3255/2845147470_244a468898.jpg)
-
Very good points. Similarly, my Mamiya 7 used to beat the heck out of DSLR. Stitching though did help a bit - this is a D300 HDR stitch (click here (http://flickr.com/photos/markhout/2845147470/) for tech data). I have tried HDR to get the same smooth dynamic range as I would have with the M7 or 4x5. I'm not there yet, but I'm getting close. Wonder if a full frame would have helped here.
(http://farm4.static.flickr.com/3255/2845147470_244a468898.jpg)
Mark, that's a helluva shot there- you could have impressed me with that even if it came out of a camera phone! Tell 'ya what I'm going to do: I'm going to take my 5D2 and shoot some pics of a dollar bill framed the way the crops are framed in the above linked "shootout" thread, with a variety of focal lengths to see what length is required in order to get resolution similar to that seen on th 4x5 image. If the test goes well, I'll post the pics.
John
-
Mark - I will second that as a beautiful shot....nice mono conversion......Don
-
I'm going to take my 5D2 and shoot some pics of a dollar bill framed the way the crops are framed in the above linked "shootout" thread, with a variety of focal lengths to see what length is required in order to get resolution similar to that seen on th 4x5 image. If the test goes well, I'll post the pics.
John
The thing is - this not so much about resolution than about brilliance / dynamic range / "that 4x5 feeling". Resolution wise I think that Michael's tests show that there is very little difference in practice between the camera and lens combinations tested. And modern sharpening methods can I think easily mimic a 10MP file of a dollar bill into a high resolution file. The OP asked about resolution, but I still think the 4x5 vs. digital is about pixel size rather than pixel count - am more than happy (certainly financially) if anyone can prove me wrong...
Thanks Don & John for your kudos...!
-
Simple math:
2400DPI scanner equivalent.(true pixels)
2400 X 5 = 12000
2400 X 4 = 9600
115.2 MP.
If your 35mm camera does, say, 6000*4000(say it's a top end model), then basically 4 shots. If it only does 16MP, then it might take 6-8 shots.
But, there is a dirty trick:
100-200DPI dye sub is just fine for printing things that will be viewed from a distance. This means that you can basically get away with 1/2 the resolution of 4 X 5 to have the same result to the audience/person viewing the printed results. All larger formats do if you're dealing with digital and printing with an inkjet or dye sub printer is get you more room to enlarge. The quality per linear inch is roughly the same no matter the MP.
The actual math: 200dpi dye sub X 1.5(Bayer losses compared to scanned film)=300DPI digital. 12000 X 9600 film equivalent is 40*32 with zero loss in quality that's discernible to the viewer. At 100dpi, though, that's 80X64 (150 dpi digital = 100dpi dye sub) Inkjets of course have no
But if you have, say, 6000*4000, that's still 40 X 32 at 100DPI printed. Most people wouldn't notice the difference from more than a few feet away.
-
But if you have, say, 6000*4000, that's still 40 X 32 at 100DPI printed. Most people wouldn't notice the difference from more than a few feet away.
That really depends on the subject IMHO.
South West landscapes are relatively forgiving since there are often few mid distance items with small details.
Woodscapes are a different matter from my experience.
On the other hand, I have seen 50x75 inch prints from a D200 that looked real good, but they were shot with a 85 T/S tilted so as to minimize DoF and only a few cm were in focus in the final print. Blurred areas are basically just as nice in large print whether they are from 8x10 or from a 10 MP DSLR if the RIP does a good job.
The bottom line though is that this is a matter of personnal standards. Whether you are happy with selling good prints, or whether you are trying to achieve very good prints. Since prints will rarely been displayed side by side, it ends up being a matter of personnal ethics more than anything else. Here also, how many MP you need to achieve a very good print at a given print size is highly subject/media dependant.
Another thing to take into account is possible future usage of the image, including foreseable improvements in display technologies. Will your images fare well in a slide show displayed on an A1 100MP LCD screen if the previous image was a stitch made from P65+ images?
Cheers,
Bernard
-
Thanks guys!
Michael, I've been looking for that shootout now for days (I guess I wasn't using the right search terms)- much appreciated.
Don- that's very helpful. I just got the 5D2, hoping (like you) to cut down on my frames for stitches. I'm hoping for 4, and no more than six. My goal is to get a sequence done in a minute or less, to catch that fleeting magical light. I figure that 60-100 megapixels (allowing for overlap) should at least come close.
John
I should add a bit more to my statement. I don't see any real quality improvement from a MP point of view from the Kodaks 14mp to the Canon's 21mp, shooting 6x7 film I think just gives me more. I now shoot 99% digital for commercial work and film 99% for my own. Having now got 21mp, to me it just shows how good film is, instead of closing the gap on film and MF digital, I think those 21mp highlight the short falls of 35mm digital.
For commercial work I'm shooting from aircraft, I find it strange that my old 6x7's shot at 100 iso with a Pentax that had a flapping garage door for a mirror, slower lenses etc look much sharper than my digital at 400 iso with faster lenses and often lens stabillisation. I have stitched images from a helicopter, not easy as they are never stationary, with great results.
Yesterday I was out for fun shooting 5x7 sheets, 4 hours I only took 2 images yet to be processed, if I had taken a digital camera I would of shot dozens and viewed them by now. That sounds like an argument for digital right? Not to me, it was about the journey to the image as much as the image. It's easy to fill a HD with digital files if that's the goal. The pleasure of the LF thought process and joy of the rare occasion of getting it right are worth more than a HD of images. It's not just resolution either, I try to shoot my 5x7 with a Soft focus when possible, I mean how daft is that right, all that gear to get blurry pictures.
And I would swop all my digital images for them on 6x7 film if I had a magic wand. But that's just me, most would not.
Kevin.
-
Simple math:
2400DPI scanner equivalent.(true pixels)
2400 X 5 = 12000
2400 X 4 = 9600
115.2 MP.
The pixels from a 2400dpi film scan are not the same quality as pixels from a DSLR image. This is the same type of math that people used to quote when arguing that you need a 20mp DSLR to match a 35mm film. Anybody who's compared actual prints knows that this is nonsense.
-
Around 60 MP. So a three image (portrait mode) stitch from a Sony A900, D3x or Canon 5DMk11 or 1Ds III should get you close enough - albeit the image would look quite different in non-resolution related ways.
Comparisons with scanned film pixels are misleading. Beyond a certain point all you are getting by scanning at higher resolutions is more grain.
A 25mp dslr should be close to 6x7 drum scanned film.
Quentin
-
albeit the image would look quite different in non-resolution related ways.
Such as?
-
The thing is - this not so much about resolution than about brilliance / dynamic range / "that 4x5 feeling". Resolution wise I think that Michael's tests show that there is very little difference in practice between the camera and lens combinations tested. And modern sharpening methods can I think easily mimic a 10MP file of a dollar bill into a high resolution file. The OP asked about resolution, but I still think the 4x5 vs. digital is about pixel size rather than pixel count - am more than happy (certainly financially) if anyone can prove me wrong...
If anything, you will have more DR with a modern DSLR than with slide films. Brilliance is now being defined as micro contrast, and yes, you have more potential for good micro contrast and natural looking images with a large file like that provided by 4x5.
Too much DR results in flat looking images closer to what we used to be getting with color negatives. It is of course always possible to increase the contrast of a wide DR image after the fact by applying a steeper curve, but the problem is that many of us as loosing the ability to anticipate when shooting.
Slide on 4x5 forced us to focus on images where the light was such that the subject would fare well with limited DR. Now a DSLR enables us to address a wider set of subjects, but many of these subjects will not be suitable for the application of a steeper curve in post processing...
Now on the look... we have at least 4 factors impacting IMHO:
- aspect ratio (4x5 is more square)
- DoF (limited or infinite)
- geometry of the image as a result of proper T/S functions usage
- quality of the lenses, including the near total lack of distorsion
My personnal view is that it is possible to emulate these fairly well with 35 mm gear.
- aspect ratio (4x5 is more square) -> through flat or cylindrical stitching
- DoF -> through DoF stacking or the usage of a T/S lens when applicable
- geometry of the image as a result of proper T/S functions usage -> through PS post-processing or T/S lenses
- quality of the lenses, including the near total lack of distorsion -> various PS plug-ins, starting with DxO and PTlens, can help deal with this.
Cheers,
Bernard
-
Brilliance is now being defined as micro contrast...
Kind of like that achieved with the "clarity" slider in ACR/LR?
-
Such as?
Depth of field, field of view, etc. It is a lot less easy to visualize the final image with stitching in my experience, whereas with a view camera WYSIWYG applies.
Quentin
-
My personnal view is that it is possible to emulate these fairly well with 35 mm gear.
- aspect ratio (4x5 is more square) -> through flat or cylindrical stitching
- DoF -> through DoF stacking or the usage of a T/S lens when applicable
- geometry of the image as a result of proper T/S functions usage -> through PS post-processing or T/S lenses
- quality of the lenses, including the near total lack of distorsion -> various PS plug-ins, starting with DxO and PTlens, can help deal with this.
Cheers,
Bernard
[/quote]
Bernard,
Yes sort of, most of those will also degrade the image in other aspects also. Most techniques there involve possible stretching or dumping information.
More importantly the state of mind is reversed from getting it right when you press the shutter to fixing it later. I don't know of any photographer that does not think they have got more lazy after shooting lots of digital, it's so easy to think shoot and fix later, shoot a few more to be sure. Resolution and technology aside, commercially it often makes sense that way round.
I think you will not get the same image on many occasions with LF and 35mm digital side by side. I'm not saying one is always better than the other, the discipline forces you to think differently and work differently.
The great beauty of LF for me is the release from technology, pixels and the computer, like I said before, the journey to the end result for me is more enjoyable and immersive with LF. Pixels, colour space, jpg's, up ressing, auto this and that, %100 viweing, ftp, farting about in Photoshop, that's for the day job.
The OP asked about resolution related to stitching, I think that is a small part of the difference between working 35 mm digi compared to LF film.
Kevin.
-
My feeling (and it's a feeling, not a scientifically substantiated theory), is that it is he *excess* presence of micro-contrast and acutance which gives digital its 'look', and that the reason many still prefer prints from film is the smoother and more continuous tone of the analogue process.
While it undoubtedly makes prints look better in some respects, the 'clarity' slider in LR (and the equivalent processes) is a major 'offender' in this respect.
What I doubt, however, is that film taken through any sort of scanning and digital printing printing process can produce a truly different appearance than a comparable digital source file. Conversely, I would be very curious to try creating inkjet negatives and printing them on analog photo paper.
It's a fun age!
- N
-
> ... A 25mp dslr should be close to 6x7 drum scanned film.
Here's a comparison that might be useful. With a Bronica RF 645, 45 mm lens, and Technical Pan film, I took a shot of a resolution test chart. This negative was scanned with a Nikon 8000 scanner. The resulting digital file indicates the MF negative has about 45 lp/mm of accessible resolution.
A Canon 5D2 with 90mm TS hits about 60 lp/mm on its sensor, using the same chart. But, of course, the MF negative is 1.6 times the size of the 5D2 sensor. So, the 5D2 prints would have about 80% of the MF resolution at an equivalent print size. These are real world numbers that include all the defects of the lenses, film, and scanner. As a practical matter, it looks like the 5D2 can make 16 x 20 inch B&W prints that look perfectly sharp.
Paul
www.PaulRoark.com
-
> ... A 25mp dslr should be close to 6x7 drum scanned film.
Here's a comparison that might be useful. With a Bronica RF 645, 45 mm lens, and Technical Pan film, I took a shot of a resolution test chart. This negative was scanned with a Nikon 8000 scanner. The resulting digital file indicates the MF negative has about 45 lp/mm of accessible resolution.
A Canon 5D2 with 90mm TS hits about 60 lp/mm on its sensor, using the same chart. But, of course, the MF negative is 1.6 times the size of the 5D2 sensor. So, the 5D2 prints would have about 80% of the MF resolution at an equivalent print size. These are real world numbers that include all the defects of the lenses, film, and scanner. As a practical matter, it looks like the 5D2 can make 16 x 20 inch B&W prints that look perfectly sharp.
Paul
www.PaulRoark.com
Paul,
Interesting but in my view not relevant for a number of reasons. The numbers might favour film but subjective impression tells a different story. Also Tech Pan is a discontinued and quirky high resolution B&W film not typical of film available and used now.
I stand by around 60-70mb being roughly 4x5 LF comparable in terms of subjective quality.
Quentin
-
Yes sort of, most of those will also degrade the image in other aspects also. Most techniques there involve possible stretching or dumping information.
More importantly the state of mind is reversed from getting it right when you press the shutter to fixing it later. I don't know of any photographer that does not think they have got more lazy after shooting lots of digital, it's so easy to think shoot and fix later, shoot a few more to be sure. Resolution and technology aside, commercially it often makes sense that way round.
Kevin,
I love shooting 4x5 myself too, and am not trying to convince you that digital is better or anything like that.
Still, the correct application of the techniques I mention require a very strong shooting discipline when taking the image. There is indeed a key post-processing step, but good quality will only be achievable if the intend was clear when shooting. I don't see these techniques as fixes, I see them as ways to expand the enveloppe of a single digital capture.
I can confidently tell you that doing all that in the field is at least as complex as shooting 4x5, and probably more so. Just a recent sample done with DoF stacking from 9 frames. The stacking was done with Helicon Focus and didn't require any additional post-processing (based on the correctly shot images, the stacking took less than 1 minute).
(http://farm4.static.flickr.com/3095/3155532077_ddf0d747aa_o.jpg)
This was shot with a 180 mm lens and would have been impossible to shoot with 4x5 since my 4x5 camera doesn't take lenses longer than 400 mm... Ideally, this should have been shot with a 300 mm lens and stitched from 3 frames to reach an even higher level of quality, but considering that I would have had to crop the 4x5 slide about 50% to get to this framing, the quality is about the same even without stitching.
Cheers,
Bernard
-
The pixels from a 2400dpi film scan are not the same quality as pixels from a DSLR image. This is the same type of math that people used to quote when arguing that you need a 20mp DSLR to match a 35mm film. Anybody who's compared actual prints knows that this is nonsense.
I used 2400DPI as that's actually the standard resolution for film on a typical DLAB's internal scanner. It's also the common "lower quality" setting on most dedicated slide scanners these days. I think it's a perfectly fine number, actually. Though, to be honest, I can't tell any difference most of the time at 2000 vs 2400. Just 2400dpi is pretty industry standard. 2400 DPI is also typical quality for a lithograph from a printing press. You don't need any more DPI, really.
The reason you need so many pixels more for digital, though, is purely because of Bayer pattern losses and the other ugly processing tricks that the sensor has to have done to it in the camera. Microlenses and AA filters make it worse as many people have noticed. I rate most consumer DLSRs at about .60 ratio versus a near perfect 0.65-0.66 for a DLB.
Oh - 35mm film(6-7MP optical) is equal to about 13-14MP with the same calculations. 2400dpi plus Bayer losses. Not even close to 20MP, which is way over 35mm. You;re right - 20MP is loony. Thankfully, 12-16MP cameras are everywhere, essentially making 35mm film moot. 24-25MP models now are approaching 645 actually. Quite impressive, considering the cost.
But 4 X 5 is insanely larger than 645 or even 6 X 6. 100MP is what you get if you turn on your scanner at 2400dpi and crank out a file. For a digital camera to match it, it would need quite a few more MP actually. The reason we can get away with less than 200MP backs for 4 X 5 is because the printers allow us to fudge things greatly. Because ink bleeds a bit when it's deposited. Also, as I stated, because you can't SEE more than about 150-200 lines per inch at a range of more than a few inches.
It's more of an issue I think of people saying that 60MP is the same as 4 X 5 because our printers are not quite up to the task of reproducing every last tiny detail like our monitors can if we're pixel peeping. ~100MP *is* 4 X 5 on a scanner. But you can easily kludge a fraction of the MP out of a digital camera on a printer to where you'd need to put your nose right up near the print to really tell which is which. And, that 36*48 inches is about as large as prints get without spending money none of us have on our printers. All of that extra resolution - in fact, much anything beyond 645 scanned film - is wasted unless you print on special equipment. You'd basically need a full digital printing press to make lithographs off of in your basement to see large format improvements at reasonable sizes. Got a spare few hundred thousand?
-
If anything, you will have more DR with a modern DSLR than with slide films. Brilliance is now being defined as micro contrast, and yes, you have more potential for good micro contrast and natural looking images with a large file like that provided by 4x5.
Too much DR results in flat looking images closer to what we used to be getting with color negatives. It is of course always possible to increase the contrast of a wide DR image after the fact by applying a steeper curve, but the problem is that many of us as loosing the ability to anticipate when shooting.
Slide on 4x5 forced us to focus on images where the light was such that the subject would fare well with limited DR. Now a DSLR enables us to address a wider set of subjects, but many of these subjects will not be suitable for the application of a steeper curve in post processing...
Now on the look... we have at least 4 factors impacting IMHO:
- aspect ratio (4x5 is more square)
- DoF (limited or infinite)
- geometry of the image as a result of proper T/S functions usage
- quality of the lenses, including the near total lack of distorsion
My personnal view is that it is possible to emulate these fairly well with 35 mm gear.
- aspect ratio (4x5 is more square) -> through flat or cylindrical stitching
- DoF -> through DoF stacking or the usage of a T/S lens when applicable
- geometry of the image as a result of proper T/S functions usage -> through PS post-processing or T/S lenses
- quality of the lenses, including the near total lack of distorsion -> various PS plug-ins, starting with DxO and PTlens, can help deal with this.
Cheers,
Bernard
Mmm... I bought a Shen-Hao 4x5 large format camera in Shanghai today. What led me towards large format was initially a desire to flat stitch using my Leaf Aptus 65, only to realize that it would be more simple to plug in a Quickload and snap a shot on 4x5 film, such as Fuji Velvia 50 . Large format is a slow moving project to me... we shall see how it works out. It it fails the camera will be decorating my house . However, I am already clear that large format lends me to see the entire image and compose it in far more great detail compared to any digital stitch using 645 camera or DSLR. Why go through so much problem to accommodate using a small sensor??
Most awesome today was when I looked at a 6x6 color slide using a Schneider 6x7 loupe. Awesome... my Aptus is pale in comparison...
Of course different tools for different purposes... but why use a DSLR to patch an image towards large format?? Nature is not always static either..., and we should not always believe new technology is superior... My experience was that Mamiya ZD 22MP 48x36mm sensor about equalled my Mamiya 7 6x7 Fuji Velvia slide in resolution... but pale in colors...
Regards
Anders
-
but why use a DSLR to patch an image towards large format?? Nature is not always static either..., and we should not always believe new technology is superior...
Well, true, but a DSLR and stitching kit is a lot more compact and offers far greater autonomy than a 4x5 kit.
Cheers,
Bernard
-
Thanks to you all for an interesting thread. I used 4x5 cameras many years ago in college, and have just switched from a D3 to a D3x. I think that the D3x is better than anything I ever got with Pan X film in the old days. Even focusing with live view trumps the black cloth...
How much of the comments really come down to:
1. Shooting style being deliberate and methodical vs. quick - e.g. the false economy of clicking the digital shutter actually messing up the brain
2. Aspect ratio
3. Post-processing flows and complexities. It is not easy to get the effects I want in PP much of the time. Film has different DNR and tonalities that can be either good or bad. The high res film scan will have the film's color space baked into it and the dynamic range will be more compressed than most digital sensors produce. (although my first look at the D3x files show an amazing ability to recover highlights and push shadows - at least as good as the D3).
-Cecelia
-
> ... The numbers might favour film but subjective impression tells a different story.
I'm actually not defending film here. On the contrary, I thought my numbers were rather more favorable to digital than is generally believed, at least in my little B&W fine art niche.
By using objective numbers based on actual image information (resolution), including all the softening effects of the optics, etc, as opposed to the manufacturer's stated scan resolution figures, for example, I was trying to provide some more truthful comparisons. If, for example, a LF film is scanned with a flatbed scanner, most of the huge number of pixels will simply be mush. So, I think the numbers I came up with (645 Tech Pan B&W film only about 20% better than Canon 5D2 frame) are way more favorable to the digital approach than most of the numbers the film fans throw around. B&W film shooters mostly believe that digital is not close at all unless you're talking about a very expensive medium format back.
> I stand by around 60-70mb being roughly 4x5 LF comparable in terms of subjective quality.
What I was actually trying to point out is that for B&W, which I think is a much tougher comparison than color film, I think the 5d2 can, at 16 x 20 print size, take on 4x5 now -- if and only if the very best lenses and shooting procedures are used. For larger prints 4x5 with a drum scan will have an advantage, but image information that is beyond what we can see is not much use.
Where I think today's top DSLRs are losing the most compared to 4x5 is in the optics. It's very hard to get than much information onto a small sensor compared to a big sheet of film. There are, however, some very good optics out there. Contrary to what most of us experience, not all lenses have terrible color fringing with DSLRs. If the very best lenses are used, as well as a tripod, DSLRs are getting very close to 4x5 right now, depending on the print size, of course. I expect to be able to go to 22 x 28 without too much trouble, and 16x20 should be virtually indistinquishable from LF.
Paul
www.PaulRoark.com
-
It is a lot less easy to visualize the final image with stitching in my experience, whereas with a view camera WYSIWYG applies.
Quentin
For this reason, I have been taking a sheet of cardboard with cutouts of various aspect ratio rectangles into the field to act as my "viewfinder." Once I find a composition and perspective that I like, I set the tripod up to put the DSLR where my eyes were. Then, I wait for the best light and fire off my shots at various stations. I'm sure this isn't as precise as composing with a LF camera, but I've been pleased with the results.
I think I'm at least partially emmulating the LF experience, as I spend a fair amount of time setting up for a shot- I usually don't get more than 3 subjects photographed in a single outing. I'd much rather have one great image instead of one hundred good ones. I'll let you know when I find it!
John
-
Where I think today's top DSLRs are losing the most compared to 4x5 is in the optics. It's very hard to get than much information onto a small sensor compared to a big sheet of film. There are, however, some very good optics out there. Contrary to what most of us experience, not all lenses have terrible color fringing with DSLRs. If the very best lenses are used, as well as a tripod, DSLRs are getting very close to 4x5 right now, depending on the print size, of course. I expect to be able to go to 22 x 28 without too much trouble, and 16x20 should be virtually indistinquishable from LF.
There is that issue as well. Optics get to the point where that's yet another limiting factor unless you print absolutely enormous prints. But for 22 X 28, The printer just isn't capable of producing more than about 645 quality at any reasonable viewing distance. At most, the large format shot might look a tiny bit more seamless at a few inches, but we're talking about several layers of blur and kludging and tweaking in the process:
- Optics - All of the problems.
- Scanning and file conversions and adjustments.
- Printing rescaling and processing losses.
- Ink/dye dot bleeding
- Viewing distances
So it all kind of equalizes in a way. True, 4 X5 is much better and preferred for, say, projection or similar uses(but who HAS a 4x5 projector?), but it's overkill for home printing. Just save some money and use a 6X7 format camera or a 25-40MP DSLR or DB.
-
Things like microlenses and AA filters are getting better - the D3x seems to have an AA filter that few if any cameras can match (the amount of detail per pixel is remarkably high). I've used a variety of digital setups and film formats over the years (from the original Canon D30 - not 30D, the original 3 MP D30!) to the D3x for the past couple of weeks. My serious photographic work is fairly traditional landscapes, including quite a bit of work much closer in than many landscape photographers work.
Subjectively, here's how a list of cameras I know well come out (when I say low ISO film, I mean Velvia, Extachrome 100, Tmax 100, etc - not Tech Pan or other exotic ultra-fine definition films). When I'm comparing film to a digital setup, I'm referring to film scanned at 4000 DPI on a Nikon 5000 or 9000 (a consumer flatbed wlll have significantly less resolution, and an Imacon may do somewhat better, although the Nikons manage to scan grain, so an Imacon can't be that much better, except perhaps in dynamic range and other non-resolution factors). I shoot everything serious in RAW at maximum bit depth at or near base ISO, and I have rarely used any film faster than ISO 100. This is a rather random sampling of cameras I have owned or used extensively over the years.
Canon D30 (3 mp) - less resolution than low-iso 35mm film, but noise less than grain on 35mm film (overall IQ fairly similar to good 35mm film) - prints 6x9 inches very comfortably, 8x12 in a stretch. Dynamic range of low-DR slide film at 5-6 stops (nail exposures and be careful with subjects).
Sony 6 mp CCD (was in a ton of DSLRs for a while, still in Nikon D40) - resolution more or less equivalent to 35mm, prints a little bigger than I've ever been comfortable with from 35mm due to noise advantage, especially in its newer incarnations (8x12 easily, 11x17 possible). Dynamic range better than most slide films, not close to print film (in the range of 7 good stops).
Nikon D200 (10 mp) - resolution significantly better than 35mm (between 35mm and 645). Overall image quality approaching 645 (which I'd say still has the edge). Prints 11x17 easily, but 16x24 is a big stretch (I've done it, am not terribly happy with the results). Dynamic range similar to 6 mp sensor.
Canon EOS 1Ds mkII (16.7 mp) - resolution nearly equivalent to 645 film, with overall image quality probably slightly to somewhat ahead of 645. The first digital camera I have used that really plays in medium format (film) territory. Dynamic range improved over any previous digital camera I had used by at least a stop (8 or more really good stops in a raw file). Prints 16x24 fairly easily, but gives up before 24x36.
Nikon D3x (24.4 mp) - resolution well into medium-format territory, close to 6x9 cm scanned film (much sharper per pixel than 1Ds mkII due to improvements in sensor/AA technology). Overall image quality significantly better than scanned 6x9 cm Velvia! Dynamic range appears to be over 9 stops, maybe 10, while remaining completely noiseless. The only files I've seen that are definitively better are scans from large-format film. Prints 24x36 inches (a 25x enlargement) very comfortably, even examining the print from a few inches away. Files appear sharp and detailed on screen at 100%
I'm sure that MF digital is even better than the D3x (although I'd be surprised if the 30ish mp variety were a big jump). 60+ MP MF digital would be approaching well-scanned 4x5 film image quality very closely, if it carries at least the same amount of information per pixel as the D3x. The few Hasselblad H3D II/31 files I've seen are in a similar league to D3x files, with the Hasselblad's edge being roughly the resolution difference (20%). I have not seen a Hasselblad file with enough subject dynamic range to make a meaningful DR comparison (it's certainly not less than the D3x, and could be significantly more).
-Dan
-
Hi,
I just went from 12.5 MPix APS-C (Sony Alpha 700) to 24.5 MPix full frame (Sony Alpha 900). I can see a tremendous improvement when pixel-peeping on screen, but I cannot tell A2-prints apart between A700 and A900. I can scan the prints and see a significant advantage to full frame, but not with the naked eye. This may differ from print to print and I guess that with time I'm going to learn the see the differences. But I guess that around 12 MPixels is OK for A2 prints and that full frame/25 MPixels would be OK for A1.
A2 is 16.5" x 23.4"
A1 is 23.4# x 33.1"
Erik
There is that issue as well. Optics get to the point where that's yet another limiting factor unless you print absolutely enormous prints. But for 22 X 28, The printer just isn't capable of producing more than about 645 quality at any reasonable viewing distance. At most, the large format shot might look a tiny bit more seamless at a few inches, but we're talking about several layers of blur and kludging and tweaking in the process:
- Optics - All of the problems.
- Scanning and file conversions and adjustments.
- Printing rescaling and processing losses.
- Ink/dye dot bleeding
- Viewing distances
So it all kind of equalizes in a way. True, 4 X5 is much better and preferred for, say, projection or similar uses(but who HAS a 4x5 projector?), but it's overkill for home printing. Just save some money and use a 6X7 format camera or a 25-40MP DSLR or DB.
-
I'm not at all surprised at the A1 print result (that ability is one of the great things about the D3x, and the Alpha 900 has the same resolution, although a different AA filter), but I am a bit surprised that the A700 does that well at A2. I can certainly tell the difference between a 1Ds mkII print (16.7 mp) and a D3x print at A2 - the 1DsII print looks fine until it's next to a D3x print, and then you notice the extra detail. I haven't made a D3x print as small as A3/11x17 yet, and I wouldn't be surprised at all if it got harder to tell at that size(where the 1DsII is native 300 dpi), except for the D3x's extra dynamic range and smoother tonal scale.
-Dan
-
Hi,
Someone with better sight or more experience probably could see a difference. I scanned the prints at low resolution (300 PPI) and the A900 was definitively better in the scan, which is a proof that print holds more detail than what I can see. I'm going 'to test my prints' on some more folks when I get back to office. It's a pity we cannot post prints over the net.
Erik
I'm not at all surprised at the A1 print result (that ability is one of the great things about the D3x, and the Alpha 900 has the same resolution, although a different AA filter), but I am a bit surprised that the A700 does that well at A2. I can certainly tell the difference between a 1Ds mkII print (16.7 mp) and a D3x print at A2 - the 1DsII print looks fine until it's next to a D3x print, and then you notice the extra detail. I haven't made a D3x print as small as A3/11x17 yet, and I wouldn't be surprised at all if it got harder to tell at that size(where the 1DsII is native 300 dpi), except for the D3x's extra dynamic range and smoother tonal scale.
-Dan
-
Hi,
I just went from 12.5 MPix APS-C (Sony Alpha 700) to 24.5 MPix full frame (Sony Alpha 900). I can see a tremendous improvement when pixel-peeping on screen, but I cannot tell A2-prints apart between A700 and A900.
Erik,
A few questions if you don't mind:
- What kind of subject was that?
- How did you sharpen your images?
- Where both cameras shot as to optimize sharpness (tripod, MLU, good lens close to optimal aperture,...)?
- Was base ISO used?
Per my experience, you should see a clear difference at A2 on a subject with many details shot optimally with both cameras.
Cheers,
Bernard
-
I used 2400DPI as that's actually the standard resolution for film on a typical DLAB's internal scanner. It's also the common "lower quality" setting on most dedicated slide scanners these days. I think it's a perfectly fine number, actually. Though, to be honest, I can't tell any difference most of the time at 2000 vs 2400. Just 2400dpi is pretty industry standard. 2400 DPI is also typical quality for a lithograph from a printing press. You don't need any more DPI, really.
The reason you need so many pixels more for digital, though, is purely because of Bayer pattern losses and the other ugly processing tricks that the sensor has to have done to it in the camera. Microlenses and AA filters make it worse as many people have noticed. I rate most consumer DLSRs at about .60 ratio versus a near perfect 0.65-0.66 for a DLB.
Oh - 35mm film(6-7MP optical) is equal to about 13-14MP with the same calculations. 2400dpi plus Bayer losses. Not even close to 20MP, which is way over 35mm. You;re right - 20MP is loony. Thankfully, 12-16MP cameras are everywhere, essentially making 35mm film moot. 24-25MP models now are approaching 645 actually. Quite impressive, considering the cost.
But 4 X 5 is insanely larger than 645 or even 6 X 6. 100MP is what you get if you turn on your scanner at 2400dpi and crank out a file. For a digital camera to match it, it would need quite a few more MP actually. The reason we can get away with less than 200MP backs for 4 X 5 is because the printers allow us to fudge things greatly. Because ink bleeds a bit when it's deposited. Also, as I stated, because you can't SEE more than about 150-200 lines per inch at a range of more than a few inches.
It's more of an issue I think of people saying that 60MP is the same as 4 X 5 because our printers are not quite up to the task of reproducing every last tiny detail like our monitors can if we're pixel peeping. ~100MP *is* 4 X 5 on a scanner. But you can easily kludge a fraction of the MP out of a digital camera on a printer to where you'd need to put your nose right up near the print to really tell which is which. And, that 36*48 inches is about as large as prints get without spending money none of us have on our printers. All of that extra resolution - in fact, much anything beyond 645 scanned film - is wasted unless you print on special equipment. You'd basically need a full digital printing press to make lithographs off of in your basement to see large format improvements at reasonable sizes. Got a spare few hundred thousand?
I tend to roll my eyes when a photography debate turns to a battle of mathematical formula interpretation. I do often relate to a couple of personal experiances, one I was thumbing through various books (mostly landscape) in a book shop, nothing really grabbed my attention, they all had that out for a walk look and grabbed a few shots at twilight feel. Then I picked up a book and even though it was less than A4 in size it screamed LF and the LF thinking had carried on to the repro and layout, you know the get it right mentality. Looking back at the other books you could see a more that will do attitude. Not to say it's impossible to print a digital capture well or LF badly , of course it is not. I do think you will treat a sheet of film with more respect than a infinitely dupable digital file and IF the repro house is any good they will to. I also think shooting with film is a different mentality to shooting digital and I bet everyone here shoots differently now they use digital, a simple question to ask yourself is do I shoot more since digital, do I shoot 5 or more digital variations when with film you would shoot just one or two? The train of thought is different not necessarily worse but different.
My other experience is with a professional landscape shooter friend of mine, for years everything was shot MF, 617 or 5X4, he has a portfolio many would kill for. He did a few trips with a digital 35mm, looking at the results I had to say great shots, but they look like everyone else's digital landscapes you see. The ease of picking up the camera hung around his neck and thinking I'll shoot a few, wide angle tele and various exposure and decide later was to great to resist, anyone here not do that?
I shoot 99% digital it makes commercial sense. When I have the time I shoot film and that's the big difference to me in favour of digital, time and effort not quality, DR or resolution. If you take time and effort out of the equation film has a lot to offer and that time and effort needed can be a positive influence. For me digital 35mm does not put time and effort to the forefront at the taking stage, I do think if I was using a Phaseone or it's like I would be thinking a bit more film like. Last year I nearly went the MF digi route, fortunately I did not as the economic slow down has all but killed my business this year. That does not alter the fact that the more I use high pixel count 35mm the less impressed I get with it.
Kevin.
-
Bernard,
Thanks for the asking!
- What kind of subject was that?
A big painting on canvas hanging outside an art school.
- How did you sharpen your images?
LR "Landscape preset" and LR "Output Sharpen for glossy medimum setting" 480 PPI
- Where both cameras shot as to optimize sharpness (tripod, MLU, good lens close to optimal aperture,...)?
Yes. Velbon Sherpa 630 CF, RRS BH40, MLU and 2s self timer, Sony SAL 16-80/3.5-4.5 ZA on APS-C and Sony 24-70/2.8 ZA on 900, both at f/8.
- Was base ISO used?
Yes, ISO 200
Now, there is an incredible difference on the input resolution, check this link (both images resampled to 360 PPI/70 cm width that is 9920 pixels wide):
http://www.pbase.com/ekr/image/107619976/original (http://www.pbase.com/ekr/image/107619976/original)
I have also scanned the prints at 300 PPI (and also 600 PPI) a small crop of the 300 PPI scans is below:
http://www.pbase.com/ekr/image/107823207/original (http://www.pbase.com/ekr/image/107823207/original)
I'll take my prints to office when we start working and ask a couple of younger or more experienced guys to have their say.
In this experiment I don't really have A2 prints but 50% linear crops printed on A4, I don't think that matters.
The conclusion I draw from this is essentially that the A700 is good enough for pretty good A2 prints. I expected a small advantage for A900, specially after doing some "pixel peeping", but it was less than what I expected. At least on these samples. I'm happy with the A900, no question about that. I guess from what I can see that I can make decent A1 prints from A-900. Not that I do that, I don't really have the wall space, but it's nice to have good images. I many cases I also need to crop and it's nice to have some margin for that.
As a personal notice, I enjoy both your pictures and your writing on this forum.
Best regards
Erik
Erik,
A few questions if you don't mind:
- What kind of subject was that?
- How did you sharpen your images?
- Where both cameras shot as to optimize sharpness (tripod, MLU, good lens close to optimal aperture,...)?
- Was base ISO used?
Per my experience, you should see a clear difference at A2 on a subject with many details shot optimally with both cameras.
Cheers,
Bernard
-
I tend to roll my eyes when a photography debate turns to a battle of mathematical formula interpretation. I do often relate to a couple of personal experiances, one I was thumbing through various books (mostly landscape) in a book shop, nothing really grabbed my attention, they all had that out for a walk look and grabbed a few shots at twilight feel. Then I picked up a book and even though it was less than A4 in size it screamed LF and the LF thinking had carried on to the repro and layout, you know the get it right mentality.
Yes, skill does have a lot to do with it. LF and MF tends to be a bit more old-school compose the shot feel to it. And I like it myself, to be honest. I think it's really a matter of having one lens and one camera and pretty much having to work with it as-is. I've been known to wait minutes to get the right shot. A lot of stuff in books now does look like they took 5 seconds to shoot it.
As I said, the difference seems to be between professional printing and home printing. In a high quality book where they are using a lithograph/4 color process at 1400-2400dpi, yes, you'll see 4 X 5 showing more detail. And if you do that sort of work, by all means go as large as you feel comfortable with. Film and larger formats shine here, to be honest. But a lot of that has to do with the 2 ft or less viewing distance as well.(arm's length)
But for the rest of us, ink jets and dye sub printers now lag greatly behind the level of detail in the typical camera. They have grown at an astonishing rate and we're still using decade old technology to print it out. So it's not surprising that on a home printer, the lines get blurry and it all starts to look the same unless you print stupidly large prints. ~24X36 is typical painting on the wall size, after all...
Note - I can see the difference between an ink jet and a lithograph at 5-10ft, but it's an overall impression and not actual detail difference. I think the reflectivity and smoothness isn't the same on an ink jet, though it is getting closer. I suspect that there is a great difference between the inks that the manufacturers aren't quite coming clean with us about. In any case, it's certainly a case of the cameras being good enough - the printing is now lagging behind.
-
I've been known to wait minutes to get the right shot.
I wish I were good enough to have to wait just a few minutes. I usually have to wait longer than that.
-
I wish I were good enough to have to wait just a few minutes. I usually have to wait longer than that.
Jeez. I've been known to get up hours before sunset, hike to a remote site, spend 15 minutes setting up the tripod and camera for a 4X5 shot, wait another half hour for the light to be just right, and only expose maybe 3 sheets. Then find out I forgeot to pull the dark slide on one of them. And eventually prefering a DSLR shot that I made while waiting for the 4X5 ones.
-
Yes... I have also waited stupidly long times... But you get my point.
Too much of what I now see passing as "art" is basically some guy going click happy all over the place and selecting the one or two that look good.
-
In any case, it's certainly a case of the cameras being good enough - the printing is now lagging behind.
I agree with your general point, but you might be impressed by some well made prints on Harman FB Al paper. The results are more detailed than I've seen on other papers. I made the first print ever that looked LF the other day.
A3+ size, Epson 3800 printer, Harman FB Al, printed using guidelines from Eric Chan
http://people.csail.mit.edu/ericchan/dp/Epson3800/index.html (http://people.csail.mit.edu/ericchan/dp/Epson3800/index.html)
My image wasn't particularly great, but it was a 65 Megapixel stitch made with an Aptus of a local landscape (fairly bland).
However the image had elements which lent themselves to that LF look. Lots of microstructure vis. tree branches. Some small black cows that in the distance had a high contrast and distinct shape.
Following this I have made a number of test shots and prints at the same resolution and printed identically, but they don't all have the same 'Mojo' showing that there is an important relationship between scale, subject matter and printing.
Still, this is the first time I've made an inkjet print at that size which has the mojo and I mainly attribute this to the paper being able to hold much more detail because of its extremely smooth glossy texture.
-
I have recently been using a D700 and stitching rather than using 6x7 for landscapes and after seeing the results have new respect for scanned 6x7. I also recently saw some very impressive 30x40" prints from 6x7 at an Annie Leibovitz exhibition.
I want to do a direct comparison at some point but based on what I've seen it takes maybe 6 12MP D700 frames as a rough guess to get close to imacon scanned 6x7. Stitching is pretty wasteful, you need a 40% overlap in both directions and often a decent amount of cropping after the software has finished warping and aligning the frames.
The real advantage to stitching is dof blending. I shoot 6x7 with an ebony rsw45 and even with lens tilt things can get difficult at times.