Cropping, sufficiency, resolution: take three (or, thoughts after shooting with the H6D-100c)

A little reframing – not as chaotic as I’d have expected…

I’ve said a lot about cropping in the past, when I think it’s justified, and even a little bit about the proto-wimmelbild interpretation of recursion in composition. Bottom line: good/acceptable cropping is when the composition and restriction of edges is done deliberately and premeditatively before capture; you know you’re going to need to leave some stuff on the cutting floor because perhaps your finder edges aren’t precise or 100%, because you want a non-native aspect ratio, or because you didn’t bring a longer lens but composed for a tighter scene in the middle of the frame. Bad cropping is when you’re hunting for a composition after capture – it’s not deliberate at the time of initial composition and is basically trying your luck. The key differentiator here is one of intent.

Longtime readers will also know that I’ve said a lot about sufficiency and matching your output intent; we don’t all need medium format, but as good as today’s cameraphones are, they’re still often restrictive in a way that affects the final image negatively. I was initially resistant to increase resolution because of processing demands, and because I believed that having high pixel integrity would lead to visually more pleasing results than simply more (but ‘looser) pixels; I stand by that. But clients asked for more, I encountered single-capture dynamic range and tonal limitations, and eventually – landed up with medium format. Sufficient for my needs, and those of my most demanding clients? Yes. The 44x33mm 50MP CMOS sensor cameras are more than enough, and that hasn’t changed. But I spent a very brief time with the Hasselblad H6D-100c a couple of weeks ago, and have been feeling like something has been brewing ever since.

Pure spatial resolution is one thing: the ability to describe ever finer structures is great, but also provides postprocessing challenges of its own, not to mention retouching and the like. Let’s just say that I won’t be using one of these for product photography. Less obviously, resolution is also coupled inextricably to tonality: the more spatial steps you have to describe a transition, the more subtle a transition you can represent, and the more accurately you can fit what is a discrete function (the pixel grid) to a continuous one (the real world). Even when downsampling, the advantage is obvious: you have more accurate colour and tone and less noise at the pixel level due to averaging. This was the first surprise in 2014 when I compared the Pentax 645Z to my D4: under comparable conditions, a 645Z file downsized to D4 sizes was significantly superior, even though the D4 had access to faster lenses (which were used). This obviously improves with the more pixels you’ve got, assuming of course your optics are capable of sufficient resolving power.

More pixels also obviously moves us further past the sufficiency bar: if you need 12MP ‘clean*’, then having a 100MP ‘clean’ start point gives you a lot of latitude to work with. Even if you need 25MP – you can either oversample and downsize by a factor of four, effectively extending your pixel-level noise threshold by two stops (with the added advantage of better acuity and colour), or you can throw away a good chunk of your image and halve the angle of view of your lens. I’ve never had anybody say the 50MP files weren’t anything other than fantastic; why would I need 100MP (with attendant increase in demands on optics and physical weight as the larger sensor requires longer focal lengths for the same angle of view)? Furthermore, the shooting envelope shrinks considerably: with more pixels per degree FOV, the amount of tolerable shake becomes even lower. It’s not trivial, either: double the spatial resolution is 41% higher angular pixel density. Lastly: I’m happy with the angles of view and rendering style of the lenses I’ve got now. Going from 44x33mm to 54x40mm would mean working that out all over again. In short: I went through the stages of lust, but was pretty sure I’d stick with 50MP.

*Clean at the pixel level; high acuity, low noise, low artefacts etc.

I’m no longer quite so sure about that. Whilst the files themselves are even better than the 50MP – aside from the obvious resolution gains, a bit more dynamic range, quite noticeably less noise, and similar levels of color and pixel acuity – yes, there’s definitely more discipline required for handholding and focus accuracy. Not to mention handling 200mb+ files. I actually kept thinking that because there was so much resolution, I could reduce my load out and gain flexibility in one of several ways: match angles of view on old and new sensors, and then not carry a longer lens when I needed more reach; keep the same focal length and crop to 44×33 by default, but have wider when I needed; or go wider and forgo the tilt shift**. Of course, you still have the full fat files to play with when needed or when conditions are ideal. This is attractive for so many reasons: you can get away with carrying a lot less hardware, which makes up for the heavier tripod you’ll probably need; you don’t have to worry about degradation through movements, and you’ve got a more flexibility in composition.

**Hold camera level, crop off the bit you don’t need. With the 24mm on 54x40mm, I’ve got 16mm-e – which is far, far wider than the 28mm-e I work with now after the HTS’ 1.5x factor and smaller 50MP sensor.

There are but two catches, one of which I think I’ve solved. The main one has to do with cropping and visualisation: it’s difficult to know where the edges of the frame are going to lie without a hard reference line. This inevitably gets you into trouble even when you previsualize your crop: there are intrusions and imbalances and all sorts of other issues that are very easy to miss. I think the solution here would be to have a focusing screen scored with very fine crop marks – perhaps a grid that covers square, 16:9, 2.4:1, 44×33 and FF. They provide 75, 54, 67 and 40MP respectively: the smallest of those is still better in outright spatial resolution than the D810, and the D810 doesn’t have the same pixel-level performance, either. Even if the back were to be used on another camera with lenses that didn’t cover the whole sensor – Otuses, for example – you’d get a 31mm square from a 135 format lens, which should yield about 44MP. The overall effect would be like one of those action finders: you can see what’s going on outside the frame. What I’m thinking is flexibility, flexibility, flexibility…

The second big catch I haven’t solved: that’s the price tag. I may have to put the CFV-50c up for sale soon to fund it…MT

__________________

Ultraprints from this series are available on request here

__________________

Visit the Teaching Store to up your photographic game – including workshop and Photoshop Workflow videos and the customized Email School of Photography. You can also support the site by purchasing from B&H and Amazon – thanks!

We are also on Facebook and there is a curated reader Flickr pool.

Images and content copyright Ming Thein | mingthein.com 2012 onwards. All rights reserved