Autofocus Performance: Technique and Technology

I received an email recently that thanked me for the AF Micro-adjustment info, but also asked about how to test the AF function in general on the 50D. They already select the AF point manually (a key first step), but some images are in focus while others are not; also – if they shoot a sequence, there are also issues with some images being in focus and others not.  I responded via email, but figured this would make a great post.

A few things to consider:

What AF mode are you using?
You are shooting w/ a set focus point – that is key, but also ensure you are in AI Servo mode. One Shot will almost never work w/ any sort of movement. (Canon terminology, but Nikon has similar modes)

Is it out of focus? Or is it motion blur from either the camera motion or the movement of the subject?
Are you running a fast enough shutter speed? The rule of thumb for handheld for *stationary* objects is is 1/focal length… so if you are in w/ a 400 mm lens, you should be shooting handheld at 1/400th or better if the thing you are shooting is standing still… then if you add motion into the mix – you need to have great panning and a steady hand, and IS, and a faster shutter speed.

What is the color/pattern of where the focus point is?
The af systems are all based on contrast and the ability to find enough contrast to check for focus (the main reason that low light AF sucks). If your subject is consistent in color, or the light isn't great, that may give the AF system fits…

Which AF point are you using?
Keep in mind that (with the 50D – but this applies to other cameras as well) the center point has the best sensitivity – All work at f5.6 but only the center point has additional sensitivity for lenses of f 2.8 or better…  Consider using the center point, keep zoomed out just a bit from how you would crop in camera, shoot the pic, and then crop during post processing to get the final picture you want.

How fast is the thing you are trying to shoot moving, and how far away are you?
You can be shooting something fast that is far away but coming at you, or you can be shooting something slow but is close, and you can have the same problems – relative speed.  Consider this… Someone moves 10 feet closer to you in 1 second as you try to keep them in focus…  If they are 100 yards away, not a problem, you're looking at only 3.33% of the distance between you two lost… If they start 40 feet away – they've moved 25% of the way towards you… but are still moving the same speed.

Is it moving towards you or side to side?
Towards you is the AF system, side to side is your panning technique.

If towards you, is it moving towards you linearly? 
I'm not positive, but I think most of the AF routines are geared towards linear movement… I track a lot of cyclists at a velodrome.  Depending on where I am and they are on the cycling track, they may be speeding up or slowing down relative to my position at different rates, even though they are moving at a constant speed.  Then couple that with the varying distance to the subject and relative amount of distance closed per second… and the brain can lock up thinking about it all…

Which lens is key too!
How fast (bright, low f-stop) is the lens, how fast the AF motor in the lens is, etc.

The difficult thing in all of this is finding ways to test that are consistent… so that you know it is your technique or the camera, or the situation that is the issue.  Finding players in the same uniforms, in the same light, at the same distance, and so on…

I found that a lot of practice was what really helped increase the number of "keepers" I had out of any photo shoot that involved sports photography.  Practice, practice, and then practice some more. The whole time trying to focus (no puns intended) on one or two of the above items during each session. If you try to include them all early, you'll just drive yourself batty.

Camera companies all (yes, both Canon and Nikon) put their better systems in their pricier bodies. Not too surprisingly, I found much better AF performance when I started using a used 1D Mark IIn (considered by some to still have the best AF performance out there).  The additional focus points, the better AF system, coupled with a great fast lens like the 70-200 f/2.8 L IS, make for even better performance – but mainly after all the other things are sound to begin with. I've loaned my 1D to some friends who still have lousy keeper rates w/ it.

At the risk of repeating myself – practice… practice… practice.

When you are shooting a sequence by holding the shutter down, keep in mind that all of the above is still very much (if not more so) in play; especially if the subject is close. If there is little contrast to the uniform or color where the AF point is targeting, the AF system may not recognize that the subject is even moving. You get the idea…

Oh – practice… practice… practice!

Thanks for reading!

5D Mark II and 1D Mark II N…

Gee – where to begin.

Since I last posted about equipment, I've sold my 50D, the grip, and my two EF-S lenses and that covered the cost of getting a 5D Mark II.  I had been planning on waiting a bit, but a friend of mine who had picked one up but a week later got her hands on two 1Ds Mark III bodies to use – suddenly found her new gripped 5D Mark II sitting on a shelf. An attempt to sell on ebay yielded a "US" buyer who wanted it shipped to Russia – riiiiiight… then she was on a few projects. Just as I was sure I was going to be good and wait a month to get my 5D2 after a few other things I'm selling (a few bikes and other various stuff that has been gathering in my garage), she pings me w/ an offer I couldn't refuse.

So – the 5D2 arrived the day before I was going to take the new (to me) 1D Mark II N out to the Superdrome to see how it performed for the first time during some night races.

So – my conundrum… shoot w/ the 1 or the 5?

Shootin' the Pooch… aka DIW and BIF

Part of the decision was made for me when I took my black lab to the neighborhood canals for a swim and took the 5D2 along to document the occasion. As we neared the canal – a snowy egret took off and repositioned itself just down from where it was – this gave me a chance to try catching an un-planned BIF (bird if flight) shot. My 100-400mm f/4.5-5.6L IS was on the 5D2 and I got off 8 or so shots and wasn't sure how well I'd done.

Soon the DIW (dog in water) opp was distracting me – the pup (well – he's 13 now) was too busy having fun chasing a bumper in the water as I fired off rounds and realized that I've been a bit spoiled with the 1D's 45 AF points… While the 9 had seemed plenty when I was shooting the 400D, 30D, 40D and 50D, the 1D opened my eyes to what true "great" AF performance and selectability is… I (like my friend Liz who sold me the 5D Mk II) think I've been spoiled…

Back at the house later, I was pleasantly surprised at how well the 5D2 had done with the egret. Having read about how the AF on the 5D2 was based on other 9 point AF systems from Canon - I wasn't sure what to expect. I had set it on Program, had done a quick white balance with my ExpoDisc as I walked over, and was able to track the bird with the center point fairly well. The fact that the bird was staying about the same distance from me helped; the images were tack sharp and the amount of detail – even on an all white bird - *without* highlight tone priority on blew me away.  But come to think of it – even the first shot was tack sharp. The photos of the dog came out great as well… Again – a bit spoiled by the AF point selection, I found I was able to find a good AF point and just keep zoomed back just a touch so I'd be able to crop down the pic just slightly to get the framing I wanted w/ the specific part of my pooch in focus… Again – the image quality of the camera and the detail I could see when pixel peeping was amazing.  This is the up-side of 21.1 MP images… the down side was soon apparent at the track that night.

At the Superdrome

That night I got to the Superdrome in Frisco, TX to find a huge turnout for the first race on the new surface. The facility had just undergone a $100,000 face lift for the track and other parts of the facility. The few weeks of open riding on this buttery smooth and fast surface brought out quite the crowd.

My plan was to mount the 17-40 f/4 on the 5D2 and have the 70-200 f/2.8L IS on the 1D2n. I ended up putting the 24-105 f/4L IS on the 5D2 for a bit as well.

I was soon in my routine of shooting the warm up, some shots in the pits (nicely packed w/ bikes and racers), and once the racing got started, the racing itself. Moving around on the apron for different angles, I also moved outside the rail to shoot some of the action so that the ligting would be a bit better for available light work.

I bring a 580EXII flash for the fill work, and will even push a longer exposure w/ wider aperture with the flash to get some motion blur with a clear image in the mix… You can see the images from that night here. I'll see about getting some of the BIF and DIW up when I have some more time.

The Down Side to 21.1 MP

So, it seemed intuitive that the large MP rating of the 5D2 would bring some immediate issues with the resulting file sizes – but there was another I'd not really counted on, but had an easy work around for.

I had been keeping an 8GB CF card in the 1D2n – which yields around 1100 or so images before I have to swap cards. I usually would take my main camera and put the 8GB UDMA card in it – then use a 4GB UDMA card to tide me over while I dumped the images from the 8GB into my Vosonic image storage device.  Then I'd just swap back after the download was complete.

I still figured I'd use the 1D2n the most (and I did) but didn't realize how quickly you can burn through 125 or so photos to fill a 4GB card… Not to worry – another card held me over while I moved files from the 4GB down to the vosonic.

The unexpected size issue arose when I'd finished working through the files overall – I sync my cameras so they have the same time on them, and I end up dumping all the photos into one folder when processing them in Lightroom (2.3 64bit). When I did the export to push to my sales site – it was easy to see which were from the 5D2 (the 10+ MB jpgs) and which were from the 1D2n (3-4 MB jpgs).  I started to upload and realized how long it was going to take… D'oh!

A quick cancel update command – back into LR to restrict max width to 4000 pix – re-export, and the files were soon posting quick as a wink.

I'll post more as I have some more time – till then…

Thanks for reading!

Canon Crop Comparison: FF vs 1.3x vs 1.6x – Bokeh

Who hasn't wondered what the hubbub is all about. All this "crop factor" talk. Full frame, 1.3x, and 1.6x (well – for the Canon-ites out there).  I think we all understand how it effects the reach of our lenses, but how does it effect the bokeh of our images?

A Quick Re-wind…

For those who may not know the first part of that, it is pretty straight forward.  The crop factor, multiplied times the focal length of your lens, yields an "effective" focal length number, or "35mm film equivalent" number.  A full frame sensor – sized the same as a negative – has a no crop factor, or really a value of one. For Canon, this would be the 5D, and the "s" flavors of the 1D series – the 1Ds, 1Ds Mark II, and the 1Ds Mark III.  Most of the entry level (and again – I'm typing "Canon" speak here) and mid level cameras are 1.6x, (Digital Rebel, 400D, 20D, 30D, 40D, 50D, etc) while the Canon 1D (non "s") cameras toe the line at 1.3x (the 1D, 1D Mark II, and 1D Mark III).

So – in practical numbers (and an easy one at that) if you put a 100-400mm zoom on each of the different crop factor cameras, set it as wide as it could go (at the 100mm position), you would get photos out of them that appeared to be taken at 100mm (the full frame), 130mm (the 1.3x), and 160mm (the 1.6).  This can be a bonus if you are looking for extra reach out of your lens, but can be a detriment if you are looking to get wider angle shots.

Back to the Point of Today's Post…

Okay – the sharp end of today is what about those middle zones… what really is the difference? Does the crop factor make any difference to the photos that would be framed the same in all three ratios?

Let's say three people, one with each crop factor type, see the same subject and want to photograph it. We'll also say that all three people stand the same distance away from that subject and frame their shots the same way through their viewfinders. Okay – and – miraculously enough, all three are using the same lens and decide upon the same settings for f-stop and shutter speed. How will their pictures be different?

Sure – we need to throw out the differences in the sensor's sensitivities, color saturation, etc… those really are not a factor of the crop factor itself, just a part of the construction of the sensor elements and image processing software and firmware.

How the Test was Done…

I have a 1D Mark II N and a 50D, so I had the 1.3x and 1.6x covered; luckily a friend – Kevin – just picked up a 5D. Voila – gotta love a plethora of bodies to test with. Now that the gear was covered, it was pretty easy to configure a test that would minimize differences and allow for valid comparisons.

A few of my vintage cameras soon were on a table with a magazine propped up against the wall to serve as a more defined thing to get out of focus in the background. I mounted my 70-200 f/2.8 L IS to my tripod, and switched off the IS.  To get the same net image, the effective focal lengths had to be the same. For the FF and 1.6, the math was easy, shoot the 1.6 at 100mm shown on the lens, and shoot the FF at 160mm. Some quick math (160/1.3) yielded 123mm for the 1.3x.  We ended up shooting w/ the 50D first, so we metered with it and chose the exposure of 1/10th at f/2.8 (we chose the f stop to minimize the depth of field). Other tidbits – all the cameras were set to "neutral" picture style; RAW;  I shot gray cards to set custom white balance for all three cameras; all were set for mirror lockup and a remote trigger (wired) was used.

After shooting, the images were brought into Lightroom, I did boost the vibrance on the shots from the 50D as the colors looked very different from the other two bodies (more on that later - as in another post). They were then exported as 800px wide images, 70% quality, sharpen for screen (for the "full image" pictures posted here); and as 100% quality, scaled to match the smallest (the 1D) image dimension so that when cropped, they would all be the same size, no sharpening.

The Results…

The three photos below (click to enlarge) are the three "as they filled the viewfinder" images. The 1.6x (50D) at the top; the 1.3x (1D Mark II N) in the middle, and the full frame (5D) at the bottom.




Okay – to sum it up – there is a very noticeable difference between the three photos. Now the distance for these was about 7 feet from the camera to the cameras, and another 8 inches or so to the magazine. The two areas that seem to illustrate the bokeh difference for me are the forearm of the rider and the lettering of Velo News.

On the forearm, the word GARMIN is sorta readable in the 50D (1.6x crop), is a bit blurrier on 1D (1.3x crop) and is darn near un-readable on the 5D (full frame).  In the lettering of the N, you can see how much the orange color expands and blurs as the crop factor gets smaller.

Now, some might think – wow – what if I don't want that much bokeh… I should avoid full frame… Nope – just don't shoot at f/2.8.  You can add depth of field by going to f/4 or f/5.6.

The key thing here is, if you are looking for MORE bokeh, moving to a full frame sensor camera body will do the trick. It isn't just buzz… it is very noticeable.

Thanks for reading!

Canon EOS 1D Mark II N – First Impressions…

Well, it has been far too long since I've posted to my blog. What better occasion than to discuss some new gear.

I've been toying with which "Pro" body to step up to for quite a while. I've jonesed for both of Canon's 1D Mark III variants – one for sports, the S for architectural work and portraits, but just couldn't rationalize the money for a) the amount of shooting – my photog biz is steady, but not my main gig; and 2) [sic] because technology is changing soooo quickly right now, I didn't want to jump on the bandwagon and have buyer's remorse.

I know waiting was the right decision, primarily as the 5D Mark II has whet my appetite for the next gen of sensors from Canon, and because of the (you make the call about the severity of the) focusing "thing" with the 1D Mark III.

What I did need was better AF performance at the velodrome. With racers coming at me at varying angles and ever changing relative speeds (the joys of tracking fast folk flying by from one spot on the track's apron), the 40D was okay, the 50D had been a bit better; but I wasn't getting the raved about 1D Mark II "best of breed" AF performance.

As I mulled over my decision, I'd kept my eyes peeled on prices, and recently a friend sent me a link to a 1D Mk II N that I couldn't resist. After you take out the included 2 day air, the included L-bracket (no – not a RRS – but a nice Kirk none the less) and the extra battery's cost - the body only set me back around $1200 or so – a screamin' deal as I've seen abused bodies sell on eBay for well over $1,600.

As advertised, the body is in great shape cosmetically, has (according to exif data - which isn't infallible – but I trust the seller's rep – so I still think it is accurate) about 46k actuations, and other than a wee bit of dust on the sensor, is in amazing shape.

I've had it since Friday, and started learning it Saturday; so far about 500 shots in the "cards" so to speak…

A few obvious (and some not so) tidbits so far:

  • AF is worlds better than anything I've shot before - can't wait to get to my first race
  • Faster to pick up new menu/controls after initial "this is nothing like my xxD series controls at all – I'll never learn this" impression – LOL
  • Need to use firewire cable, not USB, to control or config this camera. This changed for the Mk III series, but seems the driver isn't windows 64 bit compatible either. So I've had to put a fw card in my old desktop to adjust the personal settings on the camera
  • The heft to this thing just feels substantial and bomb-proof
  • I dig the ability to make my own file prefix (subtle and geeky, I  know… but I do dig it)
  • Looking forward to 5 and 7 shot brackets (xxD limited you to 3)
  • The sound of the shutter is mesmerizing
  • IQ at 1600 is (as I understood from research) pretty amazing and very usable, haven't tried the 3200 push yet.
  • There is a richness to the images that just isn't there in the xxD bodies I've shot. Suffice to say – you get what you pay for!
  • Sharpness of my L glass is more apparent with this body.

That's about it for now. I'll be blogging about my reasons for buying used (I've been a new gear guy, for my bodies at least, in the past), comparisons (as unfair as they may be) to my 40D (while it lasts) and my 50D; bokeh diffs between a 1.6 and 1.3 crop (and I'll have FF to work with as well on loan); and whatever else may come up from the next few weeks of learning this beast. I've also just picked up a nice copy of the EF 17-40 f/4 L – so there will be info about it as well.

Thanks for reading!