Earlier in the year, many of you saw me post the image of the Hasselblad 501CM hanging off a tripod at 90 degrees near the surf line. Several asked why on earth would I need to turn a square format camera sideways; apart from the obvious answer of ‘to shoot vertically!’ there’s definitely more than meets the eye. Firstly, Hasselblad did actually produce an A16 645 format magazine for the V-series bodies; they’re relatively rare nowadays and must obviously be used with the correct focusing screen to ensure accurate composition. In addition to being better suited to the typical print rectangles, you also get 33% more images per roll of 120 (16 instead of 12, as the name suggests). I was using something a little more exotic; though like the A16, it isn’t rotatable and so requires you to turn the camera through 90 to shoot verticals. It’s not very convenient, to say the least.
I’ve dallied with medium format digital on several occasions – pairing the Leica S2 off against the D800E here, for instance; taking the Hasselblad H4D-40 for a trial here. Neither of those cameras did it for me: the Leica had perhaps the best lenses I’ve ever seen, though the whole shooting experience was just far too similar to a large-ish 35mm DSLR to make me shoot and see differently. And there wasn’t that much gain over the D800E, either. The H4D-40 felt clunky, slow, a little loose, and quite frankly, enormous. I wasn’t able to shoot fluidly with it at all.
Shortly after getting my first V-series camera – a 501C – I borrowed the CFV-39 digital back from the regional distributors, Shriro, to take for a test run. In short: I hated it. So how come I landed up buying the very same back six months later? Simple: turns out the first body didn’t play nice with the back; not only did it have triggering issues, but it also had mirror alignment issues that meant that what was passably sharp on film thanks to the thickness of the emulsion was just out by enough on the digital back to cause noticeable back focusing issues. I thought it was the back, since the film results were acceptable*.
*I’ve since dismantled and re-aligned the mirror of that camera, and both film and digital results are significantly better. The higher the resolution, the more critical focusing precision becomes – as no doubt many D800/D800E users are now aware.
The second time I demoed the CFV-39 was back at the distributors’ offices; they were having a demo/ loan gear clearance after the official discontinuation of the V series. I went in to pick up some spare A12 magazines, perhaps a meter prism and bellows. The latter two items weren’t available, but an ex-demo H4D-50 and the same CFV-39 digital back was also in the pot: out of masochistic curiosity, I tried both the H4D and CFV again – this time with a different (and newer) 501CM body very generously gifted to me by one of my readers. I shot tethered and with the camera on my heavy Gitzo 5-series systematic tripod just to rule out any possible stability issues.
Let’s just say the difference was night and day: while the H4D-50 still wasn’t doing it for me from a handling point of view, the CFV was now on song. It’s relatively fat 6+ micron pixels were delivering wonderful levels of detail and acuity, and color was perhaps the most accurate I’ve ever seen straight out of camera. My credit card made an appearance, I took the customary anti-fraud phone call from the bank (“Did you just spend X at X?”) and the back followed me home. It might not have that much more raw resolution than the D800E, but the pixel acuity is still a notch up, I feel – to say nothing of color reproduction. Sadly though, the sensor was not full frame 6×6 – and aside from the 1999 Dicomed Bigshot series, there have not been any 6×6 sensors since – it’s a 1.1x 645 with 1.5x 6×6 crop – if that sounds confusing, it is. Basically, what we have is a 49x37mm sensor with 39 MP; there’s a special etched focusing screen you need to use in order to get accurate framing. You can shoot the back in either 37x37mm or 37x49mm modes; 37x37mm mode is a 1.5x square. In 37x37mm mode, you basically have a FX 35mm DSLR with 50% more at the top**. I don’t actually think it makes any sense to shoot in 37x37mm mode unless you’re really running out of card space; you might as well compose with the square lines, shoot the full area, then crop in post. You never know when the extra might come in handy – especially for commercial work.
**Not strictly a like-to-like comparison; if you cropped square from a FX 35mm camera, you’d be using a 24x24mm capture area.
The relationship is theoretically a bit like DX to FX, but somehow, just like DX and FX, the lenses don’t quite render the way you’d expect them to. The 80mm normal I’d become used to (and which feels extremely natural on a 6×6) was now a bit too long for both the 645 frame and the 1.5x 6×6 frame; 50mm works well to match the field of view, but you lose that slight telephoto compression from the 80mm that’s part of medium format’s ‘normal angle of view’ signature. It took me some time to figure out my focal lengths again, after having gotten used to a 50-80-150mm combination on 6×6 film to replicate my preferred 28-50-90mm fields of view (in 35mm equivalent). For 645, I find that the 150 and 120 still work great; the 50 is too much like 35mm on FX 35mm format for my liking, so I only ever shoot it square; the 80mm is slightly too long for normal, and becomes a short telephoto when used square. Confused yet? So am I. I think I need a 40mm to restore my wide field, and I’ll use that along with the 150mm and pretty much leave it at that for digital. I’m sure familiarity will come in time and with more use, though.
I have to admit that I haven’t really shot with the CFV as much as I’d like to; partially because of existing work commitments, jobs that aren’t really suitable for medium format (mainly watches) and the review queue backing up – in fact, most of the time when I do have time to shoot, it’s with cameras that are for review. (It’s a little frustrating at times because I’d really like to use my own gear.) I have tried shooting some street with the CFV; it’s not as bad as you’d think other than when it comes to getting a vertical, then people really look at you strangely. Otherwise, the experience is actually very much like shooting with the film V-series – so long as you remember the crop factor, and get your focal lengths right.
That brings us full circle: the camera was sideways because it was one of the few opportunities i had to try out the back properly; instead of going on holiday like a normal person, I packed a full medium format digital kit and proceeded to do some experimentation. If I didn’t, then there’s no way I’d be confident enough to use the gear on a future job – just too many unknowns. (It’s also just as well I did, because the back had an error with it’s DSP board and had to go back to Sweden for an interlude and a new board; it was repaired under warranty and came back about a month ago.) I wouldn’t have for instance found that the same battery seems to go forever; you can get hundreds and hundreds of images out of it before charging; nor would I have found that mirror lockup makes a significant difference, even at the maximum 1/500s shutter speed, and with pretty much every focal length. It doesn’t do anything above base ISO well, and you can’t really use it handheld unless shooting with at least 1/3x the focal length – the pixel density per degree AOV and huge mirror slap make it even more prone to camera shake than the D800E. It’s also got incredible native dynamic range, but very little latitude for correction or tolerance for error; this is the kind of camera that requires shot discipline of a whole different level to conventional DSLR gear.
I also wouldn’t have a couple of really nice 30″ prints from that vacation now, either; it’s very important to note that there’s really no way you can appreciate the output in a web jpeg; even full-screen on a good 27″ monitor still pales in comparison to a good print (and here’s why). Perhaps I’ll throw these into a future print run.
During the brief period I had 903 SWC in my possession, I tried the CFV on that camera too; it didn’t work that well simply because the rear element of the Biogon is too close to the sensor, and without microlenses or telecentricity, corner results were quite poor. Too bad, as it actually made for quite an interesting street photography camera – a 28mm-equivalent, hyperfocal/ zone focus medium format compact. In hindsight, I have to admit that purchasing the CFV-39 was much more of a case of ‘I want’ than ‘I need’ – probably a good thing it wasn’t the CFV-50 – but then again, isn’t that so much of photography for most of us anyway? MT
_____________________________
Enter the 2013 Maybank Photo Awards here – there’s US$35,000 worth of prizes up for grabs, it’s open to all ASEAN residents, and I’m the head judge! Entries close 31 October 2013.
____________
Visit the Teaching Store to up your photographic game – including workshop and Photoshop Workflow videos and the customized Email School of Photography; or go mobile with the Photography Compendium for iPad. You can also get your gear from B&H and Amazon. Prices are the same as normal, however a small portion of your purchase value is referred back to me. Thanks!
Don’t forget to like us on Facebook and join the reader Flickr group!
Images and content copyright Ming Thein | mingthein.com 2012 onwards. All rights reserved
Ming Thein | Photographer 
Thanks on your marvelous posting! I truly enjoyed reading it, you can be a great author.I will always bookmark your blog and will often come back someday.
I want to encourage you continue your great writing, have
a nice afternoon!
Hi Ming,
Any update on the CFV-39 after a further time with it ?
Re the SWC: I had a CFV-16 with it’s square crop and it seemed fine for square images. Is the CFV-39 in square mode OK on the SWC ? How poor is it in full frame mode ? Can you show an example as I found the SWC/CFV a wonderful walk around combination, but I really didn’t like the crop.
1. Still not okay with the SWC even in square crop.
2. The corners are terrible – colour shifted AND smeary. You’re better off using a D800E and Zeiss 21mm for this application.
Hi i have used the SWC since 1972 the only alternative is the Alpa 12 TC with a Rodenstock HR Digaron-S 35mm and a Hasselblad HAA adapter fine with a Phaseone P+45 and you can even put a Phaseone 280 on it . The Rodenstock HR Digarons are build like Retrofocus lenses .
That…sounds expensive. How does the Rodenstock do with a digital back? The SWC+CFV is not good in the corners at all.
Selamat Pagi ,the Rodenstock High Resolution Digarons are specially designed to work with digital backs the image circle is 70mm even the filter on the CFV chip is calculated in the lens design .The HR Digaron-S 35mm has got a Flange Back of 53.5 mm and the HR Digaron-S 28mm of 53.1 mm . There are also HR Digarons-W these have a bigger image circle 90mm . The corners will be sharp since the chip of the Digi Back is further from the lens then with the SWC . I use the Alpa Max shift the Digi Back right 17mm and take an exposure and then to the left 17mm and make the next exposure stich them together in CS and you get perfect pano’s or with rise and fall and you can photograph the Petronas Towers without distortion .
You are right its not cheap its like buying a Swiss watch, just like the ones you photograph but the quality is unsurpassed . Look at their Website http://www.alpa.ch/en/intro.html . their Representatives in Singapore and Malasya is Cathay Photo Store Ltd in Singapore . Web http://www.cathayphoto.com.sg . The Alpa TC is like an SWC but beter and i would use the Leica universal view finder the on for the WATE . The Specs of the Digarons you will find at the Alpa website under lenses . I am sure you will not be disappointed . I was told that Leica used the Specs of the Rodenstock lenses as Standerd for the S lenses . The Rodenstock HR Digarons have MTF with 80, 40, 20, 10 lp/mm.
Sounds like it would work pretty well. We have other buildings here apart from the twin towers, you know 🙂
I’ll pass though. Cathay is one of the least friendly and most expensive companies I’ve dealt with…frankly, if you’re spending that much money, the buying and support experience had better be good, too.
Sorry , i don’t know Cathay i just found their name on the Alpa site , but my German dealer is very friendly and helpful . My experience with a Firm in Jakarta is the same . The Alpa TC with The 35mm and HAA adapter will be the same as a 903 SWC new price .
Sad about the SWC. I hope, though I doubt, they will ever make a digital solution for that.
If 16mp can be put on m4/3s and 20 on 1″, then surely somebody will put 45 on APS and 75 on FX. It is just a matter of time.
True, but we’re diffraction limited from f5.6, which doesn’t give you enough DOF on FX, and that kind of pixel density (resolution per degree FOV) is far too much for most people to handle.
Ming,
Love the color/tonality of the above pics!
I think there is something to be said for sensors with large (or “chubby” – as I like to say) pixels. I don’t know how to to explain it, but I’ve always liked the color reproduction of a camera like a D700 (8.45 micron pixel), despite it having a relatively low resolution. It seemed to have high color resolution / tones.
Do you think there is any chance they (camera companies) might start concentrating more on factors like dynamic range, instead of high megapixels…? I think it is a crying shame that we’ll never be able to see a lot of these pics in their full sized glory. Another victim of the limits of technology…
Buy a print 🙂
We’re getting there, I think. Pixel count took a back seat to noise for a while; dynamic range is the next thing, and perhaps color accuracy after that…bottom line is that the camera companies will try and sell the obvious stuff first – because it’s easier – before moving on to trying to explain to consumers why the subtleties matter…
Hi Eric,
I agree with your sentiments. I’m probably not as experienced as you, so mine is more like gut feeling than an experienced eye. I have something to say about color in a minute but before that, do you remember (have you seen) this? Ming found some interesting things (particularly on color). And, as with all narratives that ring true, those things didn’t add up to one simple conclusion.
I feel very lucky to own a Nikon D3. It’s the same sensing apparatus in there as in the D700, isn’t it [is it? the D700 came a little later and I often wonder if they didn’t tweak the signal processing ever so slightly]. I’d come to this camera from DX. My most recent DX was a D7000.
Before making the switch, I’d read a lot of opinion which basically translated to “FX = better colors.” Even when comparing older FX to newer DX! I couldn’t quite get why “bigger sensor = better colors” — written as though it were self-evident and obvious — though it makes more sense to me now [still, the arguments for better colors from Bayer matrices shouldn’t be as simple as “more light collected by the sensor is better” since, while that is partly understandable to me, about half the color resolution of a Bayer is mathematic and not measured. So I think much of the “better colors” is actually born from better signal processing rather than better signals]. Anyway, at the time, this “FX = way better color!” piqued my curiosity, and eager for the big body, the weight, and the finder — the main reasons I blew my entire computer savings on a D3 — I made the jump.
Now, a few months later. I am still over the moon with the D3 and consider this the last DSLR I’ll ever buy [it’s only film cameras and mirrorless/more modern formats for me, from here]. It’s without doubt my most prized possession [besides wife and kids!] and my most picked up camera. But I didn’t have the same color epiphanies as everyone else. In fact, I find the D3 puts a strong orange hue in some tones, most especially skin, that has to be constantly corrected for. I find the AWB to be a step back: it balances everything way too cold much of the time. And generally speaking, away from native sensitivity, its performance is quite normal [it wasn’t at the time, I know]. So not color ecstasy, at all. More work, not less. But still, on the whole, I’m happy with colors from the FX sensor and EXPEED one. I think that combo has rightly gone down in the books as a bona-fide quantum leap.
Looking back over some D7000 files just this afternoon, it has to be said, when they were good, they were just as good. The gamut is probably shaped differently and the tonal response being a bit flatter complicates it, but yeah, it looks to me like it was as good, if not better in some cases. We’ve talked about output mediums a bit the last few threads and perhaps screen viewing isn’t the best; but in this case, at least it is consistent. And I have to say: I’m not sure the D3/D700 stands up all that well. And it should be like that: they’d had two/three years to improve upon signal processing in the D3 by that point. And just consider what the Coolpix A [optimized D7000 sensor] can do now… I think it’d probably put even the D4 to shame, if someone were daring enough to do the test and honestly report the results.
And yet, I still stand on the side of moderate resolution [fatter pixel pitch]. And just like you, I want the res [the pitch] from a generation or two ago [and have put my money where my mouth is]; but I’d like it with the processing and sensor optimization advances we have now. That’d be mana from heaven. Like what they did for the D7000 sensor in the A, do that for my sensor in the D3… D3 –> D3s –> D3x ==> D3″ [12Mpix, weakened AA, Expeed 3, etc etc]
For me the D7000 was just *too* small a pitch to be genuinely all-round useful: at night, for instance. Casting my mind back, Ming had had a post on cinematics a few months ago, when I still had the D7000, and I remember I was trying to ape him all that week… I put a 50mm on [the best I could do] and tried some shots on my way home from work [at the very earliest 20:00] and was so shocked when I couldn’t get a halfway steady photo at even 1 / focal length shutter speeds. Often 1 / 2x wasn’t enough. This was mostly just a feature of the number of pixels per angle-of-view degrees… as that number goes up [smaller pixels] life gets more annoying, doesn’t it. Noise: the ISO performance seemed horrid to me at the time; but I have to admit my own incompetence here as with hindsight the D7000’s high ISOs, 800-1600, were totally useable, and 3200 was fine in a pinch, for small outputs [and with that much res, “small” is a relative term]. Anyway, there were good things but overall I found the D7000 resolution a killer. Switching to the barn-door/cow’s ass/ what have you sized 2007 FX pixels was just great for me. My jaw dropped when I read how many pixels they put on a D7100 APS-C sensor. I can’t imagine enjoying the shooting of that [though I can imagine enjoying the files].
The camera I’ve taken the most photos with and gotten the most keepers from is a 6Mpx Epson. “Only” 3000px by 2000px… [my iPhone can outresolve that!] And it’s also the camera I’ve gotten the most fun from. Not a coincidence.
What’s good for me isn’t good for everyone. I know. But I’m quite sure the market at large will start to move this way that you’ve mentioned Eric; where else is there to go? I doubt many would enjoy 75Mpx for example [that just *can’t* be true]. This said, as Ming was mentioning before—we need the computers to drive and the monitors and to appreciate it all on!
I have a kidney up for sale if there’re any organ traffickers in the audience…
Let me play the contrarian here …
Smaller pixels are actually better, but it’s hard to tell because there are so many confounding factors, like different lenses, improving RAW processing, etc. If we can factor all of those things out, the reason single-pixel measurements can appear worse for smaller pixels than larger pixels is because the sampling bandwidth of a smaller pixel is bigger than that of a larger pixel: it can see finer structures. Because of this, it also lets in more noise, and noise is often not specified as a spectrum of amplitude vs. frequency, but as a single number that is the sum over the measured frequency range. And that frequency range is often not specified, at least for consumer devices. In other words, smaller pixels are like having a bigger window, which can let more things in, including noise. How perceptible that noise is depends on many, many things.
An interesting implication of viewing the pixel size as sampling points is that per-pixel analysis is pretty much the wrong way to look at a sensor’s performance. Instead, we should decide on what the final output size is (eg. print size or size of display), and then figure out sensor performance based on that, because even if a single pixel has more noise (and it doesn’t necessarily, as in the next paragraph), the averaging of many more pixels over the same viewing area will reduce that noise. Computers have led us down the wrong path for sensor measurement because it is trivial to examine pictures at the pixel level on a computer.
Another poster wrote about the D700 pixels being bigger and therefore better. Comparing the full saturation values and read noise of D700 and D800’s sensors shows that the D800’s sensors have not only lower noise, but also higher dynamic range, which are actually two sides of the same coin. Look at http://www.sensorgen.info for the raw data (which admittedly also does not specify the noise bandwidth).
D700: minimum read noise 5.3 electrons, maximum well capacity of 58111 e-
D800: 2.6 e-, 44972 e-
D700 dynamic range in stops: log2(58111/5.3) = 13.4 stops
D800 log2(44972/2.6) = 14.1 stops
The OM-D E-M5 BTW has 13.6 stops of DR.
A poster in DPreview, of all places, has written up a more detailed summary of this: http://www.dpreview.com/forums/post/32064270
I don’t think we’re comparing apples to apples here. We’ve got to remove the technological evolution bias – old gen big pixel vs new gen small pixel, small pixel will likely win. The trouble is that there are no sensors with different pixel pitches and similar underlying photosite architecture to compare. The underpinning circuitry also affects noise levels, not just the size of the pixel well. So it isn’t quite as straightforward as you might think.
The D800/D800E is the current generation. The D700’s sensor is now 2007 technology; the OM-D’s 2012. Same thing – if you compare the OM-D’s pixels with the D800/D800E or D7100 (closer in pitch, same generation), you’ll find the bigger ones still win.
Yeah, as I understand it, the noise component that is simply a feature of sensing only is the “shot” noise [poisson distribution] and thermally induced signal, “hot pixels” [which is a function of exposure time, but still related to the physical capture mechanism rather than what happens thereafter]. Both effects of physics not cameras.
And that said, the shot noise only becomes shot noise when the continuous signal hits the ADC, this is where the lion’s share of noise comes from:
1) Quantization noise
2) Read Noise (inc. pattern noise)
3) Amp glow
4) PRNU [pixel response non-uniformity]
I think the case of the Coolpix A is a good one to consider. Same sensor as in the D7000; yet with tweaked downstream gubbins, you get a much better package. I’m not an expert, but my gut feeling is that this ties into the fact that we have a layer of ADC and bayer demosaic before the RAW signal is represented to us [going from voltages to data numbers or raw values]. You improve those two, the same sensor [and size sensels] give you better pictures.
This said, great post Andre and I agree with the physics of what you wrote: smaller pixels shouldn’t be the topic; perhaps full well values and sensor contrast ratios [DR] should –> I always liked the analogy to catching rainfall with buckets. You can catch an amount of rain with two big buckets or twenty tiny ones: if the amount of rain that fell doesn’t change and the total capture volume doesn’t change, all that shifts is the signal to noise ratio, because of the interesting statistics of light.
Anyway, I have to run now, but it’s an interesting discussion and I hope to chip in again later. Cheers fellas!
Agreed – they’re frequently lumped together though simply because of product positioning, technology costs/ limitations/ development timelines etc…
Andre, I’d like a second bite at the cherry as I think my previous response [made commuting to a meeting this morning] didn’t do your good post justice. Ok so:
Smaller pixels are actually better, but it’s hard to tell because there are so many confounding factors, like different lenses, improving RAW processing, etc. If we can factor all of those things out…
Yes, so if I’ve understood you right here, we’re talking about photoelectrons and converting to voltages. Not the stuff that happens downstream [as that signal goes “off” the sensor/out the amp]. I understand that, but this line then confused me:
…the sampling bandwidth of a smaller pixel is bigger than that of a larger pixel
This is well in the territory of your confounding factors as bandwidth is a property of analog digital converters, no? We’re talking about quantizing the signal [voltage] now. If I’ve understood it correctly, bandwidth is just the sampling frequency of the ADC. I was never very good at electronics and preferred calculus based chemistry/engineering as the statistical mathematics in electronics is nigh on impenetrable to my brain. So I have trouble getting a clear mental picture of “bandwidth = sampling frequency”; but I can visualize slicing channels ever thinner –> 12bit numbers into 14bit numbers, into 16bit, and so on. This image would also stand as “bandwidth” would it not? [the width over which sample values are distributed]. I don’t see how this would be a property of the size of the sensel; it just seems an arbitrary property of the quantization algorithm to me—decided by engineers and not physics [though you couldn’t indefinitely slice the algorithm bit depth down and down — well mathematically you could — but past a certain point, the difference between two neighboring values would be imperceptible to the matter of the sensory equipment itself, never mind the converter!]. Anyway, could you expand on this one, as you obviously know what you’re talking about and this is a chance for me and everyone else to get a free lesson. Why is the bandwidth of a smaller pixel larger than a bigger one?
…smaller pixels are like having a bigger window…
This one got the better of me, too. Smaller pixels make for a bigger window? I’m wondering if you meant that with the full well values larger [though they may be in a smaller sensel, i.e., deeper wells] and with more of them on a chip –> effectively we have a bigger window ==> more light captured? [But deeper wells = worse performance if I’m not mistaken.] Let’s do the bucket analogy: two big buckets or twenty small ones; both with the same volume in the aggregate. For the same incident light [same lens, same aperture & shutter time] the performance *when compared on an equal footing* should be the same… If we could have a D800 with EXPEED one onboard, say, and took a picture, then did the same picture with all the same settings with a D700, and then we downsized the D800 picture to D700 resolution => wouldn’t we have more or less the same image [remember I’ve specified the same processing gubbins; but the sensels are slightly different, they have different max and min well values as you informed us]. If we did the same test over again, now with EXPEED 1 (D700) versus EXPEED 3 (D800), I’m quite sure we’d see the D800 is a far better 12Mpx camera than the D700.
per-pixel analysis is pretty much the wrong way to look at a sensor’s performance. Instead, we should decide on what the final output size is…
Totally agree with this, and can’t for the life of me understand why camera makers haven’t stolen and implemented MT’s pixel binning proposition for better noise performance which operates on this same principle. It’s exactly the analogy above with down-ressing a D800 image to D700. Though we have to do it by hand after the fact. In-camera crops, if I’ve got this right, just kill information from sensels outside the defined sensing area—which is such a waste. I have lots of confidence in Adobe’s downsizing algos — and quietly nicked MT’s advice about bicubic smoother to Ciao Pui the other day — but I have even more confidence that the makers could write even better ones for cameras to do inline for us.
As I say, when I first got a D7000, I went out and took photos at 1600+. Then got back to the PC unloaded into LR and zoomed straight into 100% and was all “huh?” I didn’t think they were very low noise *at all.* Not knowing much [not a lot changed there!] I wrote off the camera’s high ISO instead of looking closer to home or using my head, or both. Those reject high ISO files of mine, weren’t that bad to begin with [now I’m a little more worldly photographically and have seen other samples of other cameras] and would’ve been a-OK if I downsized—which I would’ve done since I only print to screen [and the web at that!].
So I’m completely with you on this point. Per native pixel is not practically the way to approach it; final output size is. Amen.
… the D800′s sensors have not only lower noise, but…
I’m torn on this one. The D3/D700 sensor is natively a more light sensitive instrument [native ISO of 200 versus 100 for the D800], so I’m not sure we can say the D800 sensor is truly less prone to noise. I think it may be that the D800 handles noise much better than the D700, even with the D700 having a one stop advantage in native sensitivity. Am I sounding wrong here? I can’t quite parse it in my head… But just from first principles: the D800 sensor should be a little more natively noisy since its base ISO is less sensitive, and it requires more gain [amplification of noise and the consequent increase in S:N ratio in step with the rms of the electron count in the sensel well] for the same ISO value.
Great links by the way. For all the ribbing we give DP Review there are a lot of anoraks and knowledgeable people on there. Plenty of chunky info to be had, for sure. Some of the guys in the Sigma [DPM] forums are scarily complex.
Totally agree with this, and can’t for the life of me understand why camera makers haven’t stolen and implemented MT’s pixel binning proposition for better noise performance which operates on this same principle.
Not quite – I’m talking about taking smaller pixels with RGB/ND filters over them, and making up luminance and color info for one pixel. It isn’t quite the same process mathematically as downsampling – and it wouldn’t look the same, either. Think Foveon with no color casts and better DR.
Sorry. Misremembered there. But whatever… why can’t we do that
–> Think Foveon with no color casts and better DR.
/Tom panting at the other end of the internet 😛
Just jumping in for a quick hijack: no reason why you can’t do cinematics with smaller sensors. Don’t forget, even Hollywood uses ‘Super 35’ – or around APSC-size – for most of the big productions. I’ve certainly done plenty of cinematics with the OM-D and it’s even smaller sensor. Hint: it’s not about the bokeh, it’s about the light…
With you. And no reason you can’t do them on views other than flat (I think Hollywood’s choice of longer lenses was probably more to do with lighting setups and not interfering with/getting in actors’ way (if someone steps off the mark suddenly, the compressed perspective helps; actors probably do better if a 21mm isn’t 40cm from their face).
A great example: the way P.T. Anderson shot the Father-son reunion scene in THERE WILL BE BLOOD. As cinematic as it gets. I think cinematic is a mood more than a technique.
That said, I think a wide is also needed for some scenes. To me, it’s more about lighting for atmosphere, human elements and visual intent than any of the other technical qualities.
4×5 is simple and relatively inexpensive. Very high Ooof Factor, even with ‘average’ lenses. The oldtimers made up for it with SIZE. I’m using the old MOD 54 reel in a Paterson tank for dev (http://www.mod54.com/about.php). Jobo tubes work good too, but are more expensive. A box of chems, a dev bag and the reel/tank are all you need, if you scan.
A friend sticks to 5×7 simply because you can contact print, forgoing the enlarger/scanning. 8×10, obviously even better, though even more cantankerous.
It’s one of those slow, cathartic photographic processes. The camera becomes a feature of the room when not in use, too!
I’ll bet one’s hit rate is a lot higher, too!
Can your D800E film scan rig scan 4×5 film?
Do you have any examples to look at?
Nope, too big, sorry!
Great article, Ming!
So what do you think about the Pentax 645 digital? It’s pretty much the only digital MF body I can afford, and lenses are cheap enough that fire-selling part of my mirrorless system should be able to fund a small work kit.
I’ve never extensively used FF digital before, and my current DSLR lenses are Canon, which doesn’t have a good high-pixel count FF body yet. With the 5D3 being rather lackluster and “going D800E” likely even more expensive than MF, the 645D looks pretty attractive…but how does it match up to the H-series and the S2?
I haven’t had a chance to use one, to be honest. I believe there’s a crop factor (33x44mm sensor?) involved, so you might not see that much of a difference in rendering style over a FX camera, less so with an FX camera and very fast primes. But yes, the lenses are fire-sale cheap.
If you have to buy lenses the D800E may well land up being more expensive than the 645D.
Vs. H series – probably better handling, to be honest, but smaller sensor.
Vs. Leica S – similar rendering, but the Leica’s lenses are on another level. The S handles like a DSLR.
Your disliking the Hassy and then picking it up months later reminds me of my experience with the Leica M8 and CCD’s in general. Sometimes it takes 2 or five tries to get acquainted with a new camera, more if it’s a different methodology/style; even more to learn and accept its shortcomings and use it within its capabilities.
And we also change/ mature as photographers, which no doubt helps too…
Sorry if duplicating someone else–
Have you tried the Hassy back with a Horseman (or similar) view camera for studio work? I haven’t but heard of it. I bet that would be a killer studio/watch/products set-up, plus tilts and shifts!
Looking into a 4×5″ setup actually…not sure if there’s enough work at that level to justify it at the moment though.
Ming:
Thanks for the write-up. MF still has great “Ooof!” factor (the sound you make when you first see the image). My dream is a full size, somewhat affordable digital V-back that I can just slap on the back of the 500. I can always hope… But, I’ll wait. I’ll stick to the simple HB system for when it counts and do fast stuff on DSLR.
The most interesting bit in this for me is that the use is pretty much for personal work.
[i.e., not film]
[[and funnily enough, I think your film photography from the Hassle has had more professional utility?]]
Yes and no; I bought it with a view to commercial work too. And I’ve shot some stuff with it already that’s covered the cost of it (though sadly embargoed, so not releasable).
I knew (or strongly suspected) you’d end up here as soon as you picked up the Hassy all those months ago. I keep thinking of picking up a used CFV-something, myself, but just can’t make myself do it. Great work as usual.
Haha – thanks. Too bad they never made a full 6×6 back – or even a 1.1 crop – I’d sell a kidney for that. Seriously.
I saw this coming Ming 🙂 I’m thinking about the same, keep refreshing local stores’ websites with used Hasselblad gear and waiting for CFV-39 back to appear. There’s something about the rendering of MF digital backs, not sure if it’s because of CCD, because of 16-bit output, because of huge DR or just because of different AOF – or combination of all of them. But once you have a chance to play with those gorgeous files it hits you and it does hit hard.
On another note, have you considered 60mm f/3.5 lens for use with digital MF? They are quite cheap used.
You mentioned that you have mirror alignment yourself, is it difficult to do or do you have some tips?
… And just know, knock-knock and the package with PM45 is at last here – the circle for film setup is complete (well, almost, still considering purchase of 50mm f/4 lens)
Enjoy. I’ll be selling my 50/4 FLE if I can find a 40 CFE IF…
I think it’s all three, personally. Each individual factor doesn’t make that much of a difference, but combined, they do.
I’ve heard good things about the 60, but not had the chance to try one. I don’t know if it’d be different enough from my 50 to justify itself though.
As for mirror alignment – let’s just say it’s definitely not simple enough to explain in text form! I’m pretty good with this kind of thing, but there were portions where I was sweating. I would recommend leaving it to a repairman unless you are either extremely confident with mechanical devices (in which case, dismantle away, just make sure you record the order in which the bits come out) or you don’t really mind if it doesn’t quite work right afterwards… 🙂
Thank you for taking time to reply (I know you’re very busy). I’d rather send it to Hasselblad than put my hands on it, was just curious if mirror alignment is a difficult task.
Yes, it’s pretty easy to mess up…
Wonderful results Ming.
Thanks Wolfgang!
Interesting that moire is quite obvious on the bag shot… is that a downsizing artifact or lack of anti-aliasing filter on the back?
Fantastic images as always – the tonal depth looks lovely. Also interestingly I think the rectangular format isn’t rectangular enough for me apart from the last image.
It’s a downsizing artefact in this case; I obviously have to shrink them to upload/ post, then Flickr does its usual sharpening disaster. The back outresolves the fibers at the point of focus, actually – beyond that, you have falloff anyway due to DOF.
Back in the good old days, I never shot MF. The size and feel of my D800 is enough like a film SLR to make me feel comfortable.
There’s something in the way a larger sensor/ longer focal length for the same AOV renders that the smaller formats can’t match; it’s similar to the transition from APSC to FX.
Well, reduced DOF is significant. I’m just an old goofball with a camera. What do I know?
Depends on what you prefer – some types of photography work better with extended DOF, for which MF isn’t useful. Other types need that DOF control – horses for courses and all that…
I like portability. Of course there are also smaller than FX alternatives, and the so called discrete factor. Perhaps in the end, few of us can afford larger than FX digital alternatives.
I really don’t think that’s the case. MF digital being expensive is a bit of a myth – there are used backs for less than a D800; this is a good way to start. The bodies and lenses themselves are pretty cheap, too, and unlike digital, they don’t seem to depreciate much – most of them are already as low as they’re going to go in price. A Hasselblad in good condition can be had for about $800 with a lens if you look hard enough – that doesn’t even buy you a D7100 body. I know people who buy Leicas and complain they’d like to try MF digital but it’s too expensive – perhaps they haven’t looked hard enough…
I didn’t realize the backs could be had cheaply although I knew the cameras were inexpensive used.
The paradox is that the technical image quality of the cheaper backs isn’t better than the best of FX now, but the way they render is very different because of the sensor size…so there’s still something to be gained, I think. At very least, there’s little to lose if you decide you don’t like it after a few months.
Wonderful read and photos!
Thanks Eric!
Ming, have you tried phase one? would love to see your review of it comparing to Hasselblad.
Not easily available here, and it’s not the kind of thing you just buy out of curiosity… 🙂
What I’m more interested in is the Leaf Aptus 12R back with the internally rotating sensor – easier to do portraits and no risk of getting the thing dusty or accidentally dropping it (unlike the Phase backs, which can be rotated but only by removing the back).
Yes, lots of things make it very expensive. But I think with D800E and potential challenger from Canon (rumored 75mp), those who use super high pixel DSLRs will ultimately figure out the sensor/pixel size difference even if the pixel counts are the same, as they get more critical to their own image quality. With this I think DMF will come back, strongly.
75MP is unrealistic for a 35mm sensor unless it’s interpolated from a multilayer design; you’d hit diffraction limits with lenses at f4 – assuming camera shake focusing errors didn’t kill your resolution first…
75Mpx is just INSANE. Even in a layered sensor. It’s insane. I can’t believe they’d do that.
I might have my wires crossed, but I’d love to see more of the post-sensel architecture played with rather than the number of sensels. Better bit depths, sampling rates what have you in ADCs, etc. I don’t want or need more resolution. I do want and need more tones, more hues, more DR up the ISO scale. I suppose it all comes back to clarity and output medium though… I need to be able to see it.
Implies –> I might have everything I want and need right now.
[but I doubt it as when a new camera comes out like the A or the D7100, D800, etc., we read about how amazing the files look and how much better: so we can see the difference in our present output mediums. Or at least, think we can 🙂 ]
We need better output media: seriously. I can’t wait til they do a 27″ iMac with a retina display; 5.5-6k on the long axis means you’d actually need 20MP just to fill at native output size. I think we’d see the difference on something like that pretty quickly…
6K!
Took my daughter out to lunch today and had to walk in the electrics superstore across the road. Just looking just looking 🙂 Sony 4K was so lovely, about the millionth time I’ve been to stand gawp, but like 33% more than the Regza (Toshiba) next to it. I’d still buy the Sony. Just blind loyalty, nothing else.
Had a play on a 13″HD tablet too. The Cintiq? The Intuous? I get lost with these stupid names: I just want to look at a Wacom.
Some pretty badass fridge-freezers in there too.
I also put hands on a D800 on the way out. Wow, see what you mean about them *really* wanting you to buy the battery grip. Just nowhere for your little finger to go. And are the finders misty or is it just me. Like a layer of condensation is on the glass or something. But yeah, for the same money… I’d be more tempted by the giant Hitachi fridge with tasteful LCD readout on the front. Or a down payment on the 55 Sony 4K 😉
Electrics stores = Nirvana
I did say the large 4k displays were awesome. Too bad my mac mini won’t drive one at 4k, defeating the point.
13″ HD tablet is the Cintiq13. Intuos has no screen.
Finders in camera stores are almost never clean; there are always fingerprints in the eyepiece or on the lens or even on the mirror/ focusing screen (!)
The downside of haptics 🙂
For me the 4K would be for watching my films on. But I’m half resigned to defeat as it means buying the best hits *over again.* And probably not on physical media either—the delivery model for 4K I’ve seen so far looks pretty much exclusively online –> cue drm misery and not being able to do everyday human life stuff like lend films to friends [very good to be my friend: my collection beats most rental stores and services hands down].
Blu Rays can store the amount the necessary data for 4K presentation, but do support the standard so we’re looking at yet another new platform. A real groaner for people like me with fortunes sunk into existing collections. But with my photog cap on—I definitely want this technology in my front room.
As you may know, the new dustbin-shaped Mac Pro can drive multiple 4k displays, and I’ve been wanting a new computer for a while to replace my vintage 2008 MBP, but there’s this shiny new Olympus that’s coming out soon, too …
Anyway, the age of my computer was driven home yesterday as I attempted to cull some 600+ images from the 16MP NEX-5N. Basically viewing them at 1:1 to find the ones out of focus or with camera shake. It must have taken LR5 a good 10 to 15 seconds to render the 1:1 image, and I’d barely culled over 100 images after an hour. No editing yet!
Why so many imges? I’d been shooting at 1/30 f/2.8 with ISO 3200, and was shooting in high-speed bursts to minimize camera shake. Full manual control including AF, and despite my earlier comments, the camera isn’t that bad in manual … as long as you don’t need to change the EV! How I miss Olympus’s IBIS, too.
I forgot to mention that the point of my post was to ask how heavily a 39-megapixel or even Nikon’s 36-megapixel files imposes on a modern computer.
I’m running a late 2012 Mac Mini with the 2.0 quad i7, 16GB of RAM and dual 500GB SSDs – no issues at all. I can open a dozen D800E/CFV files without slowdowns, and use 2000px feathered dodge/burn brushes without lag. I honestly don’t notice much of a speed difference between handling OM-D files and CFV files other than when it comes to opening from ACR; the larger files take a couple of seconds longer, but that’s about it.
Storage is quite another matter entirely, however…
I’m looking forward to one of those, too. Rendering the videos is taking ages and locking up the computer from other tasks…
Here’s one of the problems of digital: too many images. We take insurance shots and keep shooting until we think we have it absolutely perfect, because we can. And the upshot is that we land up with hundreds more than we need; typical film outing: two, maybe three rolls of 120 – 24-36 shots. An hour to develop and scan, half an hour to convert. Probably 15 keepers. Digital: 4-500 shots, similar number of keepers, three hours to go through them all. Ironic, but film is faster in bulk for a given quality of output…for me at least.
Definitely agree that digital encourages you to take too many images. Ironically, I’m having trouble finishing off a 36-exposure roll in the Ricoh GR1 at the moment. That camera always sounds like it’s going to have a heart attack when it winds the film.
For editing video, what software do you use? My computer is very slow in transcoding (import and export stages), but FCPX is surprisingly OK for editing and grading, and doesn’t crash that much on me. (Knocks on wood.)
Another idea I saw on the Internet for quickly looking through many images is to use OS X’s Finder and Preview to go through images. It is indeed very fast, but you do need to alter your workflow since you don’t get the rating system and you need to import slightly differently. Also, the native RAW converter sometimes isn’t as quickly updated as Adobe’s, so it may not work for the latest cameras.
Haha, I thought it was just me! 36 exposures actually seems like too much a lot of the time. Two rolls of 120 does me nicely usually.
My partner is doing the video editing precisely because of the software/ computer thing and his expertise – put it this way, he works with video akin to the way I work with images…
Wonderful post. Wonderful shots. I’d love to see some of these up close printed large. 🙂
Thanks! I have printed some of these large, and they’re stunning – but you have to go really large to see the difference to the best of the FX digitals.