How good is non crosstype point? Share your thought :)

Some more test shots from yesterday's outing.
These 5 shots were tracked by the top center AF point. Two shots, then I half lifted my finger from the shutter, allowing it to continue tracking, then another 3 shot burst. (This explains the size jump between image 2 and 3).

So top center AF point, AI Servo, thin DoF at 300mm f/2.8, just to challenge the camera more.
I'd love to repeat this test with a Canon 300 2.8 IS. That tracks a little better (according to TDP) than my Sigma. Still not bad.

Full frame of the first image:

gabebalazs wrote in post #16423459
Wow, that really sucks. Now I'd really like to know what improvements Canon has implemented on the 6D. Like my photos show in that other thread, maybe I'm lucky or Canon did actually improve the outer points on the 6D (of course the real improvement would have been all cross-type but even the current outer AF points seem to be better than 5DI and 5DII).

Gabe,
I believe my first target had too little contrast to be a good test. I reshot it with the center point and the results were equally poor. I reshot the center point AGAIN with a high-contrast flower garden and got 16 keeps of 19 shots. I will re-shoot the outer points with a better target. Please disregard the results you quoted. I have edited (deleted) them.
Tom

You talk of the swallows, which I try and take inflight every year as I know a place where the congregate and hunt bugs low to the ground. I only use center point on my 1DMkII, and use a 100-400, usually at 400 or 375 and set and locked. Out of probably 500 shots or so I probably have only 10-15 that I would call clear. For the 1st few years I shot them I had my shutter speed down at only 1/1200, and it just wasn't fast enough. 1/1600 seems to work much better, then again I have been experimenting with zone focusing, where I pick a spot the birds will be going by and shooting when they come past, and I get fair amount of shots in that way as well. The 100-400 has a pretty hard time keeping up, which is why I tried the "zone" focusing as well.
The shots that aren't keepers but I can see detail show an amazing amount of maneuverability, just inches above the ground. I hope to one day find a combination of lens and camera that can shoot these little acrobats with some reliability.
I'm thinking maybe 5dIII or 1DMKIV at this point, once I can afford it, but I doubt that 100-400 will be able to keep up, so maybe a different lens as well.
Not many eagles here, wish there were, but for now the swallows are a blast to try and get images of.

Bsmooth wrote in post #16423595
You talk of the swallows, which I try and take inflight every year as I know a place where the congregate and hunt bugs low to the ground. I only use center point on my 1DMkII, and use a 100-400, usually at 400 or 375 and set and locked. Out of probably 500 shots or so I probably have only 10-15 that I would call clear. For the 1st few years I shot them I had my shutter speed down at only 1/1200, and it just wasn't fast enough. 1/1600 seems to work much better, then again I have been experimenting with zone focusing, where I pick a spot the birds will be going by and shooting when they come past, and I get fair amount of shots in that way as well. The 100-400 has a pretty hard time keeping up, which is why I tried the "zone" focusing as well.
The shots that aren't keepers but I can see detail show an amazing amount of maneuverability, just inches above the ground. I hope to one day find a combination of lens and camera that can shoot these little acrobats with some reliability.
I'm thinking maybe 5dIII or 1DMKIV at this point, once I can afford it, but I doubt that 100-400 will be able to keep up, so maybe a different lens as well.
Not many eagles here, wish there were, but for now the swallows are a blast to try and get images of.

I guess once you master swallows, you can shoot anything

There's no difference in accuracy or any other performance issue with linear AF points. After all, a cross-type point is just two linear points looking at the same area of the image.
The difference is just that since the point is linear, it can't detect all possible orientations of the contrast you may find in an image, and thus it may become less accurate from that reason. When the angle between the contrast and the AF point becomes small, accuracy deteriorates. Had it been a cross type point, the other part of the pair would have taken over and maintained the accuracy.

The picture xhack posted shows not only the orientation of the AF points, but also that the rectagles depicting them show how the AF point is laid out. It's a little more obvious in the actual viewfinder than it is in the picture.

IMAGE LINK: http://www.flickr.com …/91007789@N05/8​426162425/

From other thread. My 6d center top most point refuses to lock on this (In AI servo mode).

ISO6400, f2.8, 1/50

Per the OP most of that makes sense, but really all it comes down to is having contrast in the right direction for the non-cross type points(as mentioned by several already) size only makes a difference if helps to put intersecting contrast on the AF point.

I don't know about the subject is usually smaller when using a telephoto lens, it's only smaller if the photographer makes it smaller by choice or error(didn't get close enough or chose to short of a lens). Now subject does play into it as we've all heard the you never have enough reach for birds (although I've had a few times that I had to much reach) but it's hard to control your distance with these subjects especially if it's BIF. It does help to shoot loose for moving subjects when using really long lens(it's hard enough to find the subject in the view finder at something like 600mm, keeper something that's moving and nearly feels the frame is nearly impossible to not clip part of the subject out from time to time while tracking).

Anyway I normally try to only use cross type since you don't have to worry about having contrast in correct direction(the fact that I shoot a lot of things that move fast I don't always have time to put a non cross type on proper contrast). I usually just leave the non cross type disabled on the 5DIII since it has a very good spread with the cross type. While on the 1DIV I leave them enable but try not to choose them when using single point but they're on for when expansion is desirable(if a non cross type has proper contrast it will lock and AF should be good, if it can't I have cross type that mostly can).

Speaking of swallows(those things are extremely hard to shoot). There was a pair of Rough-winged Swallows doing loops right about the water at my local pond one day I was down there with camera. They were too far to get any good shots, but I was playing around with trying to track them. I put on 45 point expansion was surprised to each AF point the swallow passed under light up. The condition weren't optimal for me to think it could even do this, sun low and to the side/back, lost of reflection on the water and the swallows just inches about it.

Anyway I didn't really try to get any shots of the swallows other than a couple of snaps to see if the AF was actually locking, it actually did fairly well for how things were. Here's one of the shots:

IMAGE LINK: http://www.flickr.com …83602155@N04/10​675392676/

YashicaFX2 wrote in post #16423413
If you saw this before I edited out everything, apologies. I reported really bad results with a 5D/28-75 and 5D/Σ70-200 OS. I have determined that my original target did not have enough contrast. How? I reshot it using center point with expansion points. That sucked, too. I got much better results with the center point on a higher-contrast target. I will reshoot the Sigma with a better target, later. No sense in re-shooting the Tamron. The 5D/28-75 combination fails regularly with outer points in good light in One-Shot.

xhack: thanks for posting the diagram, i looked all over for that

You don't need a separate graph - the view finder have that graph integrated.

An AF point is a one-dimensional image sensor.

So when the viewfinder shows a horisontally stretched box, then the AF point is 1 pixel high and multiple pixels wide - and being a horisontal line it can only detect vertical features, i.e. features that crosses the horisontal AF sensor line.

And when the viewfinder shows a vertically stretched box, then the AF point is one pixel wide and multiple pixels high. And can only catch horisontal features that crosses the vertical sensor line.

pwm2 wrote in post #16423859
You don't need a separate graph - the view finder have that graph integrated.

An AF point is a one-dimensional image sensor.

So when the viewfinder shows a horisontally stretched box, then the AF point is 1 pixel high and multiple pixels wide - and being a horisontal line it can only detect vertical features, i.e. features that crosses the horisontal AF sensor line.

And when the viewfinder shows a vertically stretched box, then the AF point is one pixel wide and multiple pixels high. And can only catch horisontal features that crosses the vertical sensor line.

That depends on the body. Non of my current bodies give you any indication of what type of point it is. 5DIII all of them show up as small squares, 1DIV all show up as vertical rectangles and the 40D well in only has 1 type.

bobbyz wrote in post #16423829
From other thread. My 6d center top most point refuses to lock on this (In AI servo mode).

So someone from that thread asked why one needs 40+ cross points for shooting a toddler, that is why.

AI servo mode is designed for the specific purpose of photographing things that are moving. It is specifically designed to detect and predict movement. If you point your camera at something that isn't moving in AI Servo mode the AF will behave unpredictably.

I reshot my 5D/70-200 OS. The best lit series I had was a white pickup with a black grill, with chrome bars (lots of contrast) coming straight at me at 30-40 mph. I used the right-most point. In a 13-shot series I got ONE sharp photo.

I shot the flower bed again again with top-most and right-most points. I still only got about 1/3 keepers. The 'in-focus' area was constantly behind the AF point. I was panning and the AF could not keep up. The performance was not significantly better with the center point. Perhaps I was just asking too much of the camera/lens combination. Neither is known for stellar AF performance, especially in AI-Servo.

Snowyman wrote in post #16423901
AI servo mode is designed for the specific purpose of photographing things that are moving. It is specifically designed to detect and predict movement. If you point your camera at something that isn't moving in AI Servo mode the AF will behave unpredictably.

That makes no sense to me. If it can detect movement, by definition, it can detect the lack thereof. As such it should be able to focus accordingly.

lovemyram4x4 wrote in post #16423884
That depends on the body. Non of my current bodies give you any indication of what type of point it is. 5DIII all of them show up as small squares, 1DIV all show up as vertical rectangles and the 40D well in only has 1 type.

The 40D only has cross points.
The 1D4 only have 6 of the 45 that isn't full cross so probably too full to try to graphically indicate.
Basically the same with the 5D3 too.

One Shot is the proper mode for subjects like lens caps on carpet. It is not moving, and one shot will nail it.

Put your camera in AI Servo and point it at a non-moving object. The focus will continue to change in and out of focus as it's looking for movement that isn't there. We all know that one shot is the focus mode for non-moving subjects, and AI Servo is for moving subjects. Seems like some are trying to force the issue by purposely challenging the focusing drive modes. You wouldn't use One Shot while trying to catch the olympic swimmer approaching head-on, just as one shouldn't use AI Servo for a completely stationary inanimate object.

Some low-light non-crosstype outer-AF point work from a 5DII:

IMAGE LINK: http://www.flickr.com …/41802269@N03/6​064692892/

IMAGE LINK: http://www.flickr.com …/41802269@N03/5​898755764/

Snowyman wrote in post #16423901
AI servo mode is designed for the specific purpose of photographing things that are moving. It is specifically designed to detect and predict movement. If you point your camera at something that isn't moving in AI Servo mode the AF will behave unpredictably.

Only top center point. Others working nicely. On 5dmk3 no issue whether subject moves or not, I always shoot AI servo and no problems. Before that I use to use 1dmk2 and same no problems. But this is off topic. Issue is cross point or not. With cross point at least I get better focus and better tracking.