I don't want this to be a debate about which one is better, that's not what I'm looking for. Medium format is clearly better . . . but the reason for this post is a little help on deciding if it's better for my application.
Since the MF digital stuff is coming down in price, I have a special upcoming 9 month long project and I'm trying to decide if I should make the next jump up to make sure I'm getting the absolute best quality image. Subject matter will include landscapes/architecture, and I need the best color and dynamic range as possible, since these will go through some moderate post-processing as well.
Anyone here experienced with the H3D-39, or even the likes of the P30+ / P45+ backs from Phase One?
I wonder how many purchasers of medium format digital (and I am NOT implying you Mr. sboerup), who have more money than sense, jump into medium format thinking "it will be just like my 5DII, only bigger and even better IQ", not realizing these systems are really designed for photographers skilled at controlling lighting ratios, for large studio use etc. I would bet some think-"wow-ISO 3200 on this big sensor will be REALLY great!", not realizing they poop out after ISO 400, by design, on purpose even When I was first learning digicam technique (I did film of all sizes for 30 years before that), I know I had such thoughts in my head, until I learned the "shocking truth"
That being said, I could actually make decent use of one, since I am a 95% of the time tripod bound landscape shooter, and have the skills to control lighting scenarios properly.
Yes. Make sure you know the ramifications of your subjects - shooting MF "the wrong way" will make you spend countless tedious hours in post-processing, and make you wonder what the fuzz what all about.
What you need is stupid amounts of light, or the option to use very long exposure times. For architecture, and also for a lot of product work you need to stop down past F32 to make the moire go away. No, it is NOT "fixable in post-processing".
You will also learn to hate the clunky and very slow handling of either the P or the HB. I prefer the P45+, since the H3D is even slower and more convoluted to use. I've only toyed with the P645 (Pentax) for a very short time, but you might want to look into that also.
The point that kind of determines your need for medium format is if you're going to shoot stuff that has movement all over the frame (wind in trees, landscape). A much cheaper (and maybe better, from a picture quality PoV!) way is to learn to use either a tech frame or a modern T/S on a camera like the 5D2 or the D3x, and stitch. All-in-all, I would actually think that takes less time per finished picture than shooting MF. But you lose the "one frame, one shot" flexibility.
A landscape-orientation D3x 4-stitch takes 30-60s to execute, 2 minutes to stuff through the computer, and the DR will make the MF backs look like 2004 technology (which they are from a sensor point of view). This will result in a 36-45MP frame, depending on your needs.
If you want a really noticable difference with MF, you need to go 39MP+. And they're still pretty expensive for what you get. But if you really need it, there's no other way to go.
First off, I certainly don't have more money than sense This would be a huge financial commitment (that frankly I'm not sure I can even swing yet), but for the project that I'm about to undertake, I'm just trying to make sure I would REALLY need the upgrade in quality.
I'm quite familiar with shooting architecture, had a number of photographs published in LUXE magazines in the region . . . so I'm very comfortable with what my 5D2 gives me for those projects, and the ne Canon TS-E lenses look phenomenal. At the same time, the Canon sensors have always made me yearn for something that offered more. Not purely with resolution, but the MF files that I've had the rare occasion to examine are just a whole different quality.
Rarely have I ever shot ISO800+ with any of these shoots. Occasionally, but that was because the 5D2 could handle it and I'd prefer to shoot quicker rather than increase the shutter speed. The subject matter and photographs I would be creating would be in daytime, and even at night it's very much illuminated enough for ISO200 speeds, or even 400.
I've created 60mp+ stitches from a 5D and the resolution is astonishing (lens was an 85L too), even greater with a 5D2. I guess I'm just wanting to hedge my bets and make sure I get the very best quality, rather than wish I had invested a bit more to get what I'm looking for.
With the MF, I'd probably be stitching photos as well, granted I probably won't be using any TS lenses though . . .
At the same time, I'm also going to be doing some extensive manipulation to some of the photographs . . . adding additional elements, merging photos, so the photographs in their final printed form won't necessarily be "true" to the scene, but more illustrative art pieces (hopefully)
PS - these have the potential to be printed BIG (for an art gallery).
Thanks for the time to write, both great comments.
theSuede wrote:
A landscape-orientation D3x 4-stitch takes 30-60s to execute, 2 minutes to stuff through the computer, and the DR will make the MF backs look like 2004 technology (which they are from a sensor point of view). This will result in a 36-45MP frame, depending on your needs.
I'm curious on this statement, how you say the DR of the D3x will make the MF backs look like 2004 tech. From what I've seen and handled the MF backs have incredible latitude that no DSLR comes even close. Granted I've never touched a D3x file, but this just doesn't sound right.
I'd love to be surprised though I thought of a D3x, but using a 5D2 with the new TS glass seems it would be just as good.
sboerup wrote:
PS - these have the potential to be printed BIG (for an art gallery).
How big? Since you used capitals is that meaning really big, like larger than say 70 x 100cm?
If yes you should really think about large format film and a drum scanner, as any of the digital MF cameras will not give you enough headroom in resolution for extensive image manipulation. There is a reason why artists like Andreas Gursky and Thomas Ruff are still using large format film most of the time.
There's an article over on the Luminous-Landscape forum right now where Miles H posted RAW files from a D3X and the pentax 645D. The pentax clearly beats the D3X and everyone seemed to agree. You should check it out because having the RAW files will illustrate the differences for you.
I shot both DSLR and MF. Even my 2004 technology Phase P20 back beats my Canon 5D2 for a lot of the work. The 5D2 prints just look flat compare to those from the p20. It has less pixels but gets more detail too. MF files look different than DSLR files partly because of the larger sensor size and partly because of the CCD sensors. Often people that don't know will say DSLR's win easily because they read DXO mark data for DR or whatever (which isn't accurate for MF for a number of reasons). But most people that shoot MF will be able to tell you its better for a lot of work and DSLR's are better for other things. If there is a lot of light and you can shoot at low ISO, MF is going to win every time. If you have to shoot at ISO 800 or higher then DSLR should be your choice. If you have fast moving subjects then DSLR is going to be your choice every time. AF sucks on the MF cameras.
anscochrome wrote:
That being said, I could actually make decent use of one, since I am a 95% of the time tripod bound landscape shooter, and have the skills to control lighting scenarios properly.
So how do you control the lighting of your landscapes?
Grad filters? Apart from a truck load of Brute arcs, that's the only way I know!
I hope the Suede comes back to discuss this more; I want to know more about the DR being better on a D3x than on the MF backs aspect; and the use of a "tech frame" for stitching (never heard of one of these).
And personally, re. stitching, I use portrait orientation, five or seven images and a panorama head when I want to stitch big images; I am getting into a Sony A850 for the "poor man's D3x sensor" experience.
Guy Mancuso and Jack Flescher over at GetDPI have just been testing a Phase IQ180 that makes interesting reading and might be helpful.
Kit Laughlin wrote:
and the use of a "tech frame" for stitching (never heard of one of these).
I'm guessing "tech frame" means a technical (large format) camera, with a DSLR attached on the back. It becomes a sort of scanning back because the back allows the camera to be shifted around the frame to make shots which can be stitched together. The Zoerk panorama adapter is a simplified implementation of this (for MF lenses).
If your looking for something that will give you anything close to the quality that 8x10 film is capable of in a six foot print, I think your only complete digital options are to stitch or to use a true scanning back such as the ones from Betterlight:
My MF experience also echoes that of EricH above. For post processing flexibility with files shot at low ISO, I don't think there is any comparison between MF and 35mm DSLR. The MF files are just so much more robust and malleable.
There's one, and only one way that medium format can beat the D3x for DR and file flexibility, and that's when you downsample the 39MP + backs to 24MP. And that's not the case if you need larger, maybe stitched files - then you compare pixel for pixel. And then there's very few things that MF does better.
You obviously get a more homogenous field, since you're only using one frame (well, this point is also moot if you're going to stitch with MF too)... One problem with stitching is that you're using several parts of the field of the used lens at several places in frame. This can make the defocus parts of the scene look different depending on where in the frame you look. In a "one-shot" frame the lens aberrations are perfectly symmetrical. The coma in one corner is the exact mirror aberration of the coma in the opposite corner. In a stitch the coma can have different directions in the frame depending on what original part that frame section came from.
You could also compare evidence, so that you don't mix personal prejudice and projected expectations into the comparison. So, the MF file is more "flexible"? In what way? Compare two frames with the same pixel resolution (1000 pixels > 2 meters of the target in reality). Start pulling the files - upwards & downwards in exposure, up & down in saturation. Does the MF back behave any better? Nope. In fact, it has less headroom, and more noise in the shadows. The very definition of lower "DR".
I don't have the P45+ here anymore, but I could do a H3D39 vs D3x comparison late this week. I can already now say that some will say that the HB has higher "per pixel resolution", but I can also show that to be a total hoax. The reason you see the AA-less sensor as having "more detail" is that it contains a huge amount of digital artefacting. Yes, this means that the average pixel>pixel contrast will be higher, but it ALSO means that that contrast is totally unrelated to any reality in front of the lens.
There are few MF owners around here. Check the MF forums on getdpi.com and LL, for more first-hand experiences with the systems. I owned a Sinar e54LV for some time and have not been able to get the same level of quality since trading it in again. Having said that, if you work harder with a good DSLR, you can get very similar results. You might need to stitch and use HDR techniques to get there, but it is possible.
Tanks for the replies, appreciate it. I do try and follow GetDPI and LL forums, but their web design is just much harder to read than FM
I'm slowly starting to lean back towards the "DSLR is good enough" mindset at the moment. Stitching isn't that hard for me and I've obtained great results, but if what a lot of people here (and on getdpi and LL are saying) is true about the DR just not being that much better, then it wouldn't make much sense to go MFDB for me.
I just stumbled across the gigapan Epic Pro device, which would make stitching even easier (although still only stitching 6-10 frames, not 900). Will give it a whirl.
Some backs do have better DR, but I would say that the biggest difference is that these backs are tuned for great colour, and have no AA filters. The results are really a noticeable step up, but there are lots of gotchas and it is hard work to get the most out of the systems. I will probably get one again at some point in the future, but not instead of a DSLR.
sboerup wrote:
I just stumbled across the gigapan Epic Pro device, which would make stitching even easier (although still only stitching 6-10 frames, not 900). Will give it a whirl.
Yeah, I thought so too until I bought one. It works like it should but it is heavy and bulky and not very portable.
So are you happy with what it does? Are you only disappointed in the size/portability? It doesn't seem overly portable to me, but I don't think that will be a problem . . . yet.
theSuede wrote:
So, the MF file is more "flexible"? In what way? Compare two frames with the same pixel resolution (1000 pixels > 2 meters of the target in reality). Start pulling the files - upwards & downwards in exposure, up & down in saturation. Does the MF back behave any better? Nope. In fact, it has less headroom, and more noise in the shadows. The very definition of lower "DR".
When I did a similar comparison a few years back specifically with B&W in mind (using raw greyscale mix, Shadow/Highlight masking and curves) using an older MF system (Leaf Aptus 17/ Mamiya 645) and my then Canon 5D, the MF file was able to take extreme abuse without falling apart whereas the Canon 5D file quickly began to show issues. I shot the exact same subject (same time, same light), cropped the Aptus to match the 5D file, abused the files similarly, etc. The difference was not subtle at all. The MF file was so much more malleable then the 5D file, it was not even funny! Worth noting is that the 5D file resolved about the same amount of detail but simply could not withstand local tonal adjustments nowhere near the amount the Leaf file could. The 5D file was "thin", almost like a thin negative and the Leaf file was dense and robust. Granted, this is with older equipment but unless the MF systems have somehow gotten much worse in their ability to be manipulated as MP have increased , I have not seen a 35mm DSLR file with that sort of flexibility even today.
When I was first learning digicam technique (I did film of all sizes for 30 years before that), I know I had such thoughts in my head, until I learned the "shocking truth"