Autofocus Drill-down

Long Lens Autofocus Considerations.

If you read my previous post about the 1Dx sensor you will have seen that I mentioned my, as yet unfinished, tome about long lens autofocus for wildlife photography.  It’s a frustrating project because I keep having to change various bits to make them simpler, re-order certain paragraphs etc.

But I thought I’d blog-post something here that I expand on in the project, and it’s something an awful lot of people NEVER take into consideration.

As a Nikon user I’m used to the vagaries of the Nikon AF system and I manage to work with it just fine – I have to!

But photographers who don’t shoot wildlife, and don’t use 400mm or 500mm lumps of glass as their “standard lens” might not find the vagaries I bitch about quite so apparent; indeed some might not come across them at all.

As a wildlife photographer I shoot in crappy light, I shoot with slow lenses (both in terms of f-number and focus speed), I shoot low contrast subjects on equally low contrast backgrounds, I’m constantly shooting brown-on-brown, grey on grey etc, I shoot stupidly small subjects….the list goes on!

For years, good wildlife photography has been done by pushing camera/lens capabilities beyond their performance design parameters; and this particularly applies to our “expectations” of our latest and greatest AF system – be it Canon or Nikon.

I find so many people who come to my workshops etc. are not even aware of this one simple fact – sharp focus requires more work AND increased speed of work by the lens AF motor the closer a subject is to the camera.

Just try looking at the delineations on the focusing ring of a lens:

D3C6337 900x599 Autofocus Drill down

Canon 200-400 focused at 20 meters. (Lens porn WARNING: This lens will cause movements in the front-of-trouser department).

Look at the scale and note the distance between 20m and 50m marks – that distance is indicative of the amount of work required of the autofocus controller and motor to move from 20m to 50m or vice versa.

Now look where the 10m mark is – it requires FAR MORE work from the focus controller and motor to move from 20m to 10m, than it did to move the 30 meters from 50m to 20m.

On top of that extra work, if we are tracking a subject moving at 10 meters per second the lens takes 3 seconds to move from 50m to 20m, but then has to move a lot FASTER as well to cover the extra workload moving from 20m to 10m in just 1 second.

Then you wonder why your Nikon D40 + Sigma 50-500mm is crap at doing “birds in flight”; you never realise that your autofocus system is bag of spanners and powered by a hamster on a wheel…….it’s just not fast enough kids

Autofocus accuracy is nothing without speed if you are wanting to do productive wildlife photography.

As I alluded to before, as a photographer of the old wildlife I, and YOU will always encounter problems that users in other photographic disciplines may not, or if they do then the problem has a lot less impact than it does for us.

Think of it this way – a sports photographer will use a 500mm f4 to photograph a 6 foot tall overpaid git who’s 25m to 70m away, on a sunny Saturday afternoon or under a squillion watts of flood lighting; and he’s looking for a 6×12 for the back page of the Sunday Sport.  I’ll use the same lens to photograph a cute Red Squirrel at 5m to 7m in a gloomy wood in the middle of winter and I’m looking for a full size, full resolution image for stock.

SquidgerPor 900x620 Autofocus Drill down

Red Squirrel – this is basically the FURTHEST DISTANCE you could shoot at with a 500mm lens and still get a meaningful composition. Click for larger view.

Note the distance – 631/100 – that means 6.31 meters. Aperture is f8, so DoF is around 7 centimeters.

The image is UNCROPPED as are all the other images in this post

We don’t really want to be any further away because “his cuteness” will be too small in the frame:

The factors effecting subject distance choice are:

  1.  lens resolving power – small, fine details need to be as close as possible.*
  2.  sensor resolving power – we need as many pixels as possible covering the subject.*
  3.  auto focus point placement accuracy – if the subject is too small in the frame, point placement is inaccurate.
  4. general “in camera” composition

*These two are inextricably intertwined

I’ve indicated the active focus point on the above image too  because here’s a depth of field “point of note” – autofocus wastes DoF.  Where is the plane of focus? Just between the eyes of the squirrel.

Assuming the accepted modern norm of DoF distribution – 50/50 – that’s 3.5 centimeters in front of the plane of focus, or indicted AF point, that will be sharp.  Only problem there is that the squirrel’s nose is only around 1 centimeter closer to the camera than the AF point, so the remaining 2 .5 centimeters of DoF is wasted on a sharp rendition of the fresh air between its nose and the camera!!

Now let’s change camera orientation and go a bit closer to get the very TIGHTEST shot composition:

SquidgerLs 900x370 Autofocus Drill down

Red Squirrel – this is basically the CLOSEST DISTANCE you could shoot at with a 500mm lens and still get a meaningful composition. Click for larger view

The subject distance is 5.62 meters. Aperture is f6.3 so DoF is around 4.4 centimeters.

Now let’s change photographic hats and imagine we are a sports photographer and we are spending a Saturday afternoon photographing a bunch of over-paid 6 foot tall gits chasing a ball around a field, using the very same camera and lens:

HUPor 900x608 Autofocus Drill down

He’s not over-paid or chasing a ball, but this is the CLOSEST distance we can shoot at with this orientation and still get a “not too tight” composition of a 6 foot git! “Shep’s” not a git really – well, not much! Click to enlarge

The distance for this shot is 29.9 meters. Aperture is f6.3 so DoF is around 1.34 meters.

HULs 900x602 Autofocus Drill down

And here we are at the CLOSEST distance for this horizontal camera orientation – still not too tight. Click to enlarge.

The distance here is 50.1 meters. Aperture is f6.3 so DoF is around 3.79 meters.

So with this new “sports shooter” hat on, have we got an easier job than the cold, wet squirrel photographer?

You bet your sweet life we have!

The “Shepster” can basically jump around and move about like an idiot on acid and stay in sharp focus because:

  1. the depth of field at those distances is large.
  2. more importantly, the autofocus has VERY little work to do along the lens axis, because 1 or 2 meters of subject movement closer to the camera requires very small movements of the lens focus mechanicals.

But the poor wildlife photographer with his cute squirrel has so much more of a hard time getting good sharp shots because:

  1. he/she has got little or no depth of field
  2. small subject movements along the lens axis require very large and very fast        movement of the lens focus mechanicals.

So the next time you watch a video by Canon or Nikon demonstrating the effectiveness of their new AF system on some new camera body or other; or you go trawling the internet looking for what AF settings the pros use, just bear in mind that “one mans fruit may be another mans poison” just because he/she photographs bigger subjects at longer average distances”.

Equipment choice and its manner of deployment and use is just not a level playing field is it…but it’s something a lot of folk don’t realise or think about.

And how many folk would ever consider that a desired “in camera” image composition has such a massive set of implications for autofocus performance – not many – but if you put your brain in gear it’s blindingly obvious.

Please consider supporting this blog.

This blog really does need your support. All the information I put on these pages I do freely, but it does involve costs in both time and money.

If you find this post useful and informative please could you help by making a small donation – it would really help me out a lot – whatever you can afford would be gratefully received.

Your donation will help offset the costs of running this blog and so help me to bring you lots more useful and informative content.

Many thanks in advance.


Canon 1Dx Sensor Noise

What seems an age ago now I did a two-day workshop for Calumet at Drummond Street in London, and a chap turned up there armed with a Canon 1Dx PLUS a 200-400 f4.  The lens had only just been launched, and he’d been out and spent a truck-load of cash on both lens and body.

Yours truly was all over him like a severe rash, and I ripped it from his poor old fingers, stuck it in Case 2 and dived outside and started ripping through frames of vehicles passing the store!

I was smitten from that very moment – but it was weird all the same.  I was joyous at how the lens and camera performed; pissed off that I didn’t own it; and seething at Nikon AF and the poor distance performance of their own 200-400.

Not that Nikon is crap – far from it; it’s fantastic – but this was just SO much better, and child that I am, THIS was what I should be using and all else was just pants.

Begrudgingly I handed the old chap his camera back, satisfied my dour mood with a cursory “not bad….” and carried on with the workshop.

Later in the day I stuck the images I’d shot with the Canon 1Dx into Lightroom on the 27″ iMac I use for workshop presentations – and was immediately a little happier – they looked “iffy” to say the least!

The Nikon 200-400s’ distance resolution problem has always hacked me off – 10m or less it’s epic, but 75m and further I hate it, and in between well, sometimes I like it and sometimes I don’t. And it’s bad with teleconverters, it really is…

Scanning through all the Canon 1Dx shots I was still amazed by the lens – it was delivering tack sharp high-resolution images at all focal lengths and distances, with and without  teleconverter; basically it was kicking the Nikon into the last century simply by NOT displaying ANY of the same faults.

But I was having to look past – in comparison to Nikon – a thin veil of sensor noise, and I was also aware of a reduction in sensor Dynamic Range when I looked at the shots and noted the popped highlights that experience told me my Nikon wouldn’t produce.

Since then I’ve had a few more occasional chances to use the lens and body, and my results have continued to generate the same response – great lens, shame about the sensor IQ; but I’ve always been using other folks cameras and you don’t like to mess about with them too much, so I have always assumed that things “could be made a bit better” with some fiddling about.

Last year, hand on heart, I can honestly say that I was responsible, in whole or part, for at least 6 sales of Canon 200-400’s to existing 1Dx owners, and the lens-envy has always been there when they’ve been and bought it.

Since the first day I handled the lens I’ve been of the mind that it would be the ultimate lens for my Eagle workshops in Norway.  I was thinking of trying to take one, plus the required 1Dx, over there in June last year; but seeing as the my clients were all Nikon I thought I’d best not!

But I have a “mixed bag” of clients booked for my Winter trip in a couple of weeks time, so seeing as I was of the mind that a few folk owed me a few favours…..

Upshot is that for the last two weeks, thanks to Reece Piper at Calumet, I’ve had a Canon 1Dx sat in my office; and many thanks to my favourite Geordie lass June Lown, a 200-400 f4 to go with it.

When I picked up the 1Dx from Calumet I swiped a 100mm f2.8 macro while I was at it, as I had been tasked with a high speed action shot featuring makeup brushes and I thought we’d go nuts and do the shot whilst exploring the 1Dx in a bit more depth.

GX2R6630NTM 800x900 Canon 1Dx Sensor Noise

Canon 1Dx + 100mm f2.8 macro. Nikon SB800 flash & Calumet ProSeries wireless.

Prior to picking up the Canon 1Dx I’d done a few test shots on my own Nikon gear just to get the lighting and flash timing sorted out, but I’d been using some different brushes:

D4D0013 599x900 Canon 1Dx Sensor Noise

Okay, so here is the base .CR2 raw file for the finished image:

GX2R6638 600x900 Canon 1Dx Sensor Noise

Base .CR2 raw file

Now, I’m going to get to the point of this post topic!

As a standard retouching procedure on this type of shot I always overlay a custom Curves Adjustment layer with a sine-wave curve – it helps show up all those little imperfections you can’t see when you view the image without it:

Screen Shot 2015 01 30 at 15.09.38 Canon 1Dx Sensor Noise

Custom Curve for retouching

The main purpose in this particular case is to check for dark imperfections in that black background – yep, proper retouching is all about the minutia if you want perfection.

I’m trying to put together a video course on retouching that’ll be available in my store a little later this year – email me for details

Because the powder velocity is so damned high as it leaves the brush bristles I needed 1/32nd output power on the SB800s in order to freeze absolutely every grain of powder, so the shots(both Nikon and Canon) were at 400 iso just to give me a working aperture of f14.

When checking the test shots they looked like this with the customised Curve Layer:

D4D0013 Edit 599x900 Canon 1Dx Sensor Noise

Nikon D4 400 ISO 1/250th @ f14 test shot with custom curves layer.

Check out how clean the black background is.

So now all we do is swap the brushes, and change from Nikon D4 to the Canon 1Dx – I make no changes to either the lights or the background, and the exposure settings are exactly the same – 1/250th, 400 ISO, f14 – and I’m expecting gold…

But throw the CR2 file into Photoshop and stick the custom curve over it to see the comparrison:

GX2R6638 Edit 600x900 Canon 1Dx Sensor Noise

Canon 1Dx 400 ISO 1/250th @ f14

Sweet Jesus………….!

Now don’t run away with the idea that it’s the “normal” noise you think of – luminance noise.  In fact from that point of view it’s no better or worse than the Nikon D4 sensor.

But what you can see here is PATTERN NOISE/READ NOISE – see my Sensor Noise post from a while ago HERE

Don’t get me wrong, you can barely see it at 100% magnification, and a lot of folk won’t notice it AT ALL:

fullREZ 900x614 Canon 1Dx Sensor Noise

Canon 1Dx/Nikon D4 comparison at 100% magnification viewed normally. CLICK to view larger

But if you want BIG prints, or you sell your images for stock, then you need to check them a lot more thoroughly at higher magnifications:

400x 900x614 Canon 1Dx Sensor Noise

Comparison at 400% magnification CLICK to view larger

At 400% the noise is just about visible – because it’s a dark error/fault on a basically slightly darker background.  But keep it at 100% and put the custom curve over it and:

fullREZ2 900x615 Canon 1Dx Sensor Noise

Custom Curve layer at 100% – now that pattern noise on the Canon sensor is obvious.

…now you can see what you have got to take care off in retouching.

Got a 1Dx?  Then this pattern noise is in YOUR images – FACT.

But if the image has a more “normal” tonality to it then it certainly won’t be obvious to you – but it’s there nevertheless.  Just try looking in your shadow areas.

Why the 1Dx sensor should be so much noisier than the D4/D4S is beyond me to be honest. Yes I know it’s an older mark, but the then current Nikon D3 and D3S were far better than this; in fact they were, and still are, only marginally worse than today’s Nikons for pattern noise.

In reality the images are of course eminently usable – as the millions of 1Dx images used daily world-wide testifies; but they do need a teeny bit more effort when processing than files from a top-end Nikon camera, if the final images are to have the same degree of quality in terms of “clean-ness of file”.

There is also the question of a clipped Dynamic Range, but that’s an easy walk-around in most cases – neither Highlight Tone Priority or Safety Shift are the answer though IMO; the former just under-exposes the shot, and the latter drives me nuts, though it’s a damnably good idea in principle.

So this noise thing truly is my ONE AND ONLY gripe about this camera – up until this last week I had a few others based solely on my usage of other folks cameras, but those are now well and truly GONE.

On the “pros & cons” side of things, noise and clipped dynamic range are my only cons, and there are many pros that cancel them out – the real big one for my is the autofocus system which, at least when used with the new(ish) 200-400 and the latest firmware, is truly EPIC and seriously kicks Nikon into a cocked hat in terms of tractability, speed, accuracy and user control.

I’m working on a large pdf document all about autofocus with both Nikon D4/D4S and Canon 1Dx bodies that has wildlife photography and long lenses as the main bias, but it will give a lot of valuable information and knowledge to non-wildlife photographers and 5DMk3 owners as well. Again, email me for details – BUT IT WON’T BE FREE!

If I had the dough I’d buy a 1Dx and a 200-400 f4 tomorrow – perhaps I’d even dump Nikon all together for long lens action/wildlife photography.

But I haven’t, so unless a miracle happens and Canon suddenly feel like sponsoring someone who actually “knows about stuff” then there’ll be tears when this rig has to go back I can tell you…:(

Would I dump Nikon for all my photography where speed and autofocus are not required, like macro or landscape – not on your bloody life!

Many thanks to Reece Piper & Calumet UK, June Lown, and Chuck Westfall of Canon USA


Please consider supporting this blog.

This blog really does need your support. All the information I put on these pages I do freely, but it does involve costs in both time and money.

If you find this post useful and informative please could you help by making a small donation – it would really help me out a lot – whatever you can afford would be gratefully received.

Your donation will help offset the costs of running this blog and so help me to bring you lots more useful and informative content.

Many thanks in advance.