Apologies to everyone who has been kind enough to email me or comment recently – I’ve been planning for a while now to write a Question & Answer article to capture most of the questions, and to update the “My Workflow” page on the site, but they’ve been coming in quicker than I’ve been able to respond!
However, I’ve found the time to catch up with most of them now – there are a few recurring themes that I can cover in one go. This will be a work in progress.
Is the 7D a noisy camera?
I know here’s a perception out there that the 7D is a noisy camera. What can I tell you? As a one-time Nikon D200 user, I’m extremely sensitised to noise and simply won’t tolerate a noisy camera: it follows then, that as far as I’m concerned the 7D isn’t noisy – I simply wouldn’t be using it if it was.
Most of the serious reviews of it (from respectable sources anyway) tend to the same conclusion, incidentally. Certainly it provides no evidence of the much repeated idea that more pixels means more noise – indeed, all other things being equal, smaller pixels = less noise.
That’s surely what I’m seeing: and the nature of small pixels means that noise is less obtrusive in the first place, and – importantly – the noise generated by sensors with small pixels seems to be far less damaging to detail and edges.
The 7D is however, a frequent victim of poor comparison techniques, I think.
People will insist on comparing 100% crops from the 18 mp sensor in the 7D with, say, 100% crops from the 10 mp 40D. Obviously the 7D is disadvantaged by that comparison: but it’s a bad – and irrelevant – methodology. You need to look at the image level. Do that and you’ll find that the 7D is cleaner than the 40D at any ISO, and markedly better as ISO increases.
The real key to getting the best noise performance out of the 7D though, is selective sharpening.
I still apply light noise reduction on conversion (usually just enough to remove the chroma; and a tiny bit – default or less – of luma) but I don’t “capture sharpen” on conversion any more. Thorough testing has satisfied me that I lose nothing from the end result by missing this conversion stage out – who knew? – and instead I apply sharpening selectively in Photoshop.
This approach means that the little bit of NR done at the conversion stage is usually all that’s needed, and that once resized, the images look good.
There is a problem in some circumstances with a banding artefact in low ISO 7D files: I’ve only seen it (trivially) once in any of the 40,000-odd “Real World” images I’ve taken with my 7D (probably because I’m never below 400 ISO) but I can generate it in 100 ISO images if I crank up the shadows and look at 100%+ view (in other words, I have to make it into a problem by pixel-peeping).
But there’s a solution to this problem too. Canon’s own DPP is very effective at dealing with this banding; and a recently introduced Raw converter called PhotoDirector 2011 – which apparently uses Canon’s own RAW conversion engine – handles it brilliantly, as well as providing quite a lot of the functions that Lightroom delivers and which are absent from DPP. It is in fact fair to say that PhotoDirector owes a lot to Lr, but it’s none the worse for that, and although it’s not quite as good, it’s a damn’ sight cheaper.
My long-standing Open Source favourite Raw Therapee also has a useful debanding tool: it works in a different way to the one in DPP and PhotoDirector – you can see at 100% that it “dithers” the banding to break it up – but the adjustment range is wide, so it takes little effort to find “just enough” to remove the banding without any detriment to detail, and at the image level the solution is great.
There’s also some talk that – in general – the 7D is noisy at low ISO, especially in the sky: again, this is very converter-dependent, and – again – isn’t a significant issue at the image level. I recall that when the Nikon D300 was first released, the self-same complaints were levelled against it; they quietened right down as people got used to the camera and how to get the best from it. It’s the same for the 7D.
The bottom line in my experience is that the 7D is not only not a noisy camera, but at higher ISOs it’s the best APS-C camera available – and that includes comparison with much newer cameras like the Nikon D7000, which are great too, but no better.
What’s your workflow?
The upcoming article will spell it out in detail (this earlier piece still has value, but some of the software choices have changed) but there’s no magic to it – and as I touch on above, I consider the most important factor to be selective sharpening after conversion – and no pixel-peeping!
To emphasise this point, I’ll refer back to the first of the 6400 ISO images I linked to above.
Although I’m a huge fan of Lightroom 3 these days, primarily because of its excellent demosaicing and noise reduction technologies, this image was converted in Capture One 6 and – again – I think it’s tremendous for 6400 ISO from a “cropper”: and at 50mm, f/3.2, and 1/50 handheld, it’s properly low light. Granted, not much in the way of shadows to deal with, and it’s not cropped, but it’s still bloody good for low light and high ISO, and that’s because of selective sharpening.
I might have added a little bit of selective NR too – the Topaz DeNoise Photoshop plugin is excellent, applied on a Duplicate layer and then erased from where it’s not needed – but it didn’t need it.
Of course, some folk will argue that at 1000 px, any image can look clean at high ISO, but we know that’s not true.
Do I have a “special” 100-400mm?
Honestly, I don’t think I do.
I know there’s much talk out there about some people having exceptional 100-400mms – there’s a common view that most are so-so (or worse) and that only the Lucky Few get really good copies.
The thing is, I know quite a few folk with this lens, and I’ve tried theirs. I’ve put their lenses on my camera bodies, I’ve used their camera and lens, and I’ve put my lens on their cameras. And every single time, with every single combination, I’ve had the same results – sharp, detailed, well-focused images.
Now then. Having just implied that I think most copies of the 100-400mm are “thereabouts”, at the risk of contradicting myself somewhat I know that my lens isn’t as good say, as the one owned by Romy Ocon. But if mine is average – and I honestly believe it is – then that means there must be an awful lot of pretty damn’ good 100-400mms out there…
And I have to say that although I understand completely the arguments from some people for the 400mm f/5.6 prime over the 100-400mm, I can’t be without IS, the closer MFD and the zoom. Given the effectively non-existent Real World IQ advantage of the prime over the zoom, I’ll have versatility every time, please – the prime really does strike me as a limiting lens in some respects.
How might the 7D be improved?
Well, it’s a great camera but it’s certainly not perfect, and there are a few things I would like to see done better.
Although it doesn’t really bother me, I’d like to see the low ISO banding problem sorted out – it surprises me that it got through QA, to be honest (although any camera will tend to bending given enough encouragement – even the Nikon D700 if you push hard enough).
I’ve seen examples of the 7D artefact that were obvious enough without having to torture the file; although it’s still something I consider to be easy enough to deal with by making good conversion and post-processing decisions, it shouldn’t be there.
More importantly to me, I dearly wish that Canon would sort out their Auto ISO once and for all. Nikon has had this right since 2004 when the D70 came out, but Canon still insists on half-arsed implementations.
What I can’t do with the 7D but dearly wish I could (and can with any current Nikon – Pentax too, as far as I can tell) is put the camera in Manual, set the shutter speed and aperture I want, an upper ISO limit and an Exposure Compensation (EC) value, and let the camera ride up and down the ISO to maintain these settings.
The 7D is close but no cigar: crucially, it is impossible to adjust the EC in Manual/Auto ISO. This matters a great deal for bird photography or any other genre where the photographer has no control over lighting and subject.
I don’t get to decide the upper ISO limit either. As it happens it stops at 3200 ISO anyway, but I might want more or – just as likely – less…
Come on Canon – it ain’t hard. Hell, I could probably write the code myself.
The other improvement I’d really like to see is rather subtle – but also obvious.
I use the Zone AF mode a lot for BIFs, and – for the most part – it’s excellent, especially against a clear sky. Page 90 of the manual very clearly suggests that Zone is “inclined” to focus on the thing closest to the camera. That – you’d think – would make shooting a bird against a busy background a doddle, then.
Well, there’s a lot of room for improvement for the algorithm that recognises what’s closest!
In my experience the AF will sometimes (which I say because sometimes it works as expected) happily grab whatever’s behind the bird if there’s enough contrast there: and indeed, often enough even if there appears to be no contest between the bird and the background. In short, it doesn’t really seem to be able to tell the difference between what’s near and what’s far away.
A little bit like Father Dougal, really…
If Canon could organise a firmware update that really tightened up on “closest subject priority”, that’d be a hell of an improvement. Mind you, the Nikon D200 came with something purportedly very similar, and that didn’t work for me either – but in fairness, Nikon does warn that its implementation may not work with long lenses. And interestingly, Nikon dropped this function from its later “best” AF modules.
And y’know? I still manage.
But for me, that’s it – the only three things I can really see room for improvement in (and the first one doesn’t really bother me anyway), which is pretty impressive, I think.