I’m reposting this article from 2012 for the simple reason that I’m still getting far too many emails from people obsessing over equipment with bigger numbers or higher specs solving compositional and creative deficiencies. I think I’ll continue to do this on a regular basis so long as those communiques keep coming in…
The never-ending photographic arms race got me thinking recently about sufficiency: how many pixels, fps, AF points, ISO settings, etc. are enough? The troubling thing is that I thought I used to know the answer: I’m no longer sure it’s quite as clear cut. See, the thing is that if you’re viewing images online, in theory, anything close to your screen resolution (leaving space for UI elements, text, menus etc.) should be sufficient – 1000px wide is more than enough for most purposes. The images on this site are mostly 800px wide, for reference. In theory, that should mean an iPhone is overkill. Yes and no; just because resolution sufficient, it doesn’t mean that we’re going to have enough dynamic range, or color depth (or accuracy).
This raises a hypothetical question: suppose we could have say a ‘perfect’ 3MP – in a compact camera, with a reasonably good zoom lens; if we put those 3MP into a 2/3″ or 1/1.7″ sized sensor, we could probably get a fast-sh 24-120mm f2-4 equivalent into something that would be reasonably pocketable. The relatively low pixel density would mean several things – good acuity, low noise, good color accuracy, and much higher forgiveness of the lens quality. With today’s technology, I don’t see why you couldn’t get a clean ISO 6400 and useable ISO 12800. Even with a lens of moderate speed, that’s a more than sufficient shooting envelope for most photographic purposes. Add a good optical stabilizer, fast AF, 14 bit RAW, responsive buffering and controls, and most people would be set. And with raw files that small, you could probably get 10,000 of them onto a 32GB card.
But, nobody would buy it and it would be a commercial failure. Only 3MP? Really?
That’s the kicker. If you’ve ever had a really high quality, but small-ish file, then you’ll know that you can actually do quite a lot with it; I remember seeing some images from the 2005 Wildlife Photographer of the Year competition – prints, at 40×60″ or so – which were shot with a Nikon D1H: that’s right, all of 2.7MP. Did they sing? You bet. Did they look grainy, or pixellated? Not especially so, but I’m sure you’d see the difference if you shot the exact same scene with a D800E. The shot that won would no doubt still be a great capture even twenty years from now. A hundred years from now. Does the equipment matter? Insofar as it didn’t get in the way of the capture, no.
In a more realistic scenario, a proper 300dpi print (actual pixels, not printer ink dots) looks extremely sharp. 200dpi is still acceptable; do you know what most computer screens are? Closer to 110. Sharp images still look sharp, don’t they? And the reason things appear pixellated is because the pixel mask of the screen gives each pixel a hard edge. If the edges of each dot were just a tiny bit fuzzy and overlapped the next dot, like in a print, then things would look just fine. The reality is that short of sticking yourself a few inches away from the display, you’re not really going to see the individual pixels. I have to be honest; the Macbook Pro’s 220dpi ‘retina’ display doesn’t look all that different to my 27″ Thunderbolt Display from healthy viewing distances. But it requires four times the amount of graphics power to run, because guess what: every graphical element requires four times the pixels.
Most people print no larger than 6×4″, or perhaps out to 8×12″ for special images; survey a group of enthusiasts on the largest print they’ve made, and you’ll probably find 13×19″ a good end point. There are two reasons why: cost, and lack of display space. Enormous prints are great, and actually printing work is important, but you’d also better have enormous walls to hold them.
So what does a 1500×2000 pixel, 3MP file get you? At 110dpi, 13.6×18.2″ – lo and behold, that’s pretty darn close to 13×19″! The one problem with this scenario is that it assumes that all photographers using the ‘sufficient’ camera have the shot discipline to get things perfect at the pixel level, and that they’re really getting the full resolution of the sensor. Sigma/ Foveon shooters will know what I mean by this – the files may be small, but the real resolving power is pretty darned high. Higher than the sensor’s pixel count would lead you to believe (but not as high as Sigma’s marketing department would like you to believe).
Even for use on a next generation ‘retina’ display, you could still get a 7×9″ image out – comfortably fitting the display size. You could probably interpolate it a little too, and not see too much degradation in quality.
Let’s put this all into context. A decent 20MP file has a whopping 5500×3600 or so pixels, or enough for a 50×33″ print – I don’t have enough space in my house to hang more than one or two of those, and at that size, I’d struggle to think what image I wouldn’t mind looking at for hours on end. 20MP is available in ultracompacts (not very good pixels); large-sensor compacts like the RX100/II (much better pixels); we’re not far off for M4/3, have exceeded it in APS-C and any larger format. Yet there are still people crying ‘not enough resolution!’. At that sensor size, pixel density is so high that critical focusing and shooting discipline become very important; I notice that the stabilizer seems a lot less effective than lower-pixel count cameras, but I’m now starting to suspect it’s because it’s got to work quite a lot harder to maintain perfection at the pixel level.
Take the argument a notch further: okay, so there are still reasons to have something with interchangeable lenses. And those things always come with bigger sensors (okay, so not the Nikon 1 cameras or Pentax Q, but they haven’t exactly been a commercial success – expensive, small sensor, low pixel count; even if the pixel quality is reasonably high and the cameras are extremely responsive). Bigger sensors mean more expected resolution. I’m going to bypass M4/3 and APS-C for the time being, because they all top out at reasonably similar resolutions; there isn’t that much linear difference between 16 and 24MP – remember, area scales with the square of length. Let’s go to full frame – one of the most common questions I’ve received via email in the last few months is ‘what do you think of the A7/A7R?’ The fact that that question is even being asked signals that the marketing department has done a good job. They’ve sold you on the lure of the enormous number of pixels and the necessity of full frame to make a ‘proper’ image; you haven’t bothered to do a bit more reading to find out that a) you’re going to need much better lenses; b) shot discipline once again becomes critical; c) stability is a hot topic; d) the back end file handling becomes very time consuming indeed. Frankly, if you’re not making 20×30″ or larger landscapes (for the fine detail) on a regular basis – do you really need the heartache that comes from having to revaluate all of your lenses, and shunt around 50MB RAW files? Probably not. But, the bragging rights are a different matter altogether.
One of the reasons why the photographic industry is still growing despite passing saturation point some time ago is because of the photographers themselves: they don’t know when to stop. Canon will happily sell you a 50/1.2 for three or four times the price of the 50/1.4, which itself is three times the price of the 50/1.8 – simply because there are people who will pay for that extra half a stop. (And it may be an extra half a stop of aperture, but it’s almost certainly less than half a stop of transmission. Let’s not even talk about Leica’s 2.8-2.5-2.0-2.0 APO-1.4-0.95 lineup.) Nikon recently followed suit with it’s 58/1.4, which is three times the price of the 50/1.4 – but it’s still f1.4! Let’s be realistic: the camera’s metering doesn’t gauge scenes that accurately; I routinely make adjustments of +/- 1 stop, and with smaller sensor cameras, I’ve got to use the spot meter or risk an unsalvageable image. With the degree of adjustment latitude our modern sensors give us, whole stops changes are what you have to be making before you see any appreciable differences in image quality.
Here’s another curious thing: equipment with more conservative specifications often performs better than more extreme gear, even though the extreme gear frequently costs several times more. This is because we’re dealing with known technology, with greater tolerance for error; it’s a Swiss Army Knife rather than a scalpel. This is why there are many superb 50/2s or 50/2.8s that perform excellently wide open, but few 50/1.4s and faster that do – the Zeiss Otus is one of the exceptional few. It’s also why lenses with higher resolving power – take the macros for instance – have modest apertures. It’s easier to correct for smaller apertures. Same thing with sensors – it’s easier to get good dynamic range and a high signal-to-noise ratio out of a sensor with a larger pixel pitch; the supporting electronics for each photosite has to be of a certain size, and if there isn’t much area to go around in the first place, that circuitry starts to make a significant and noticeable difference to the light-collecting ability of the sensor.
The bottom line is very Delphic: Know thyself. Specifically, think carefully about which pictorial limitations are due to the photographer, and which are because of the camera. Low light performance? It’s probably your gear. Poor composition? More pixels isn’t going to help you. Blurry images? More pixels definitely aren’t going to help you; examine your technique first.
The recent posts illustrated with images from a phone and compact respectively were posted specifically to illustrate this point: just because a camera is old, it doesn’t mean that it’s no longer capable of producing good images. If somebody could use it to get the shot when they had no other choice, it means that you can still do that. The important thing to remember is that most of the time, the limitation is in the user, not the equipment. So next time, before buying more gear, think about spending money on improving your skill level first. I guarantee that you’ll see a much bigger improvement in your images. MT
2014 Making Outstanding Images Workshops: Melbourne, Sydney and London – click here for more information and to book!
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!
Images and content copyright Ming Thein | mingthein.com 2012 onwards. All rights reserved