1ds3 focus

This is not nature photography ( I haven't time to find a duck) but topical .. Phil_S et al may be interested in this and the rest of the www appears to be devoid of any other examples, so ...

I finally had the camera in the car today with a wee bit of time to spare. Whilst I accept this not ideal I simply stuck the camera out the window, handholding it, the lens a 300mm at F2.8 @ about 300 to 500' second, on servo, and pressed the button. No other camera settings have been set away from default.

The following pic is taken straight from the raw and shows the crop area on the number plate of 720 * 100 pixels. No sharpening or alterations have been applied.

The car would have been doing about 40mph (fast duck in flight) (this is 18mps, which means that it covered approx 1.7m between shots) and the depth of field of this lense at the median car distance @F2.8 would be about 0.25m - so the focus was doing something

Following this is the crops of the number plates 7 in total.

I'll leave it up to you all to make your mind up, it was bright (ish - this is the UK) and 1'C.


T

now the detail ..

comments are welcomed, obviously these are posted at pixel level; is this good enough or would you expect better?

3 from 7 look pretty dodgy, but could this be motion blur?
You're the man who bought the camera, so it's your call whether its acceptable or not.

John

3 from 7 for me as well-about the same as the 1D3 Time will tell Tom, I had mine a few months before i started to see problems as i wasnt doing moving subjects, Then i went to California and had all sorts-Maybe its America!!!!!

Dave

Motion blur is a distinct possibility, handheld, and the car would have travelled approx 4 to 6cm during the exposure. For comparison the smaller letter height would be about 0.8cm

I am reasonably impressed given the cirmcumstances they where taken in, but hey I am no sports photographer - for that I think the world is awaiting what RG has to say about the D3 and 1ds3 focus...

T

I hope you were not driving at the same time Tom

Hi Tom. Err, it all looks 'acceptable' to me, but maybe I have gone video blind from spending too much time in front of this thing To be fair to the camera, that crop is tiny, and three 3/7 on any moving target, with any camera is good (especially handheld and moving yourself). I'd be happy with it.

Thanks for the additional info Tom. I shoot only landscapes, so autofocus ability isn't all that important to me. Currently, I don't even have AF on my MF camera. I'm sure others like to know about it though, so thanks for posting.

I'm going to have to find someone local to Denver who has the camera and go out shooting with them. I'd like to get some properly processed RAW files on my hands to compare directly to my Pentax 67ii, Velvia or Provia film, scanned with a Nikon 9000 ED.

I'd be pretty confident that this is motion blur.
When I handhold the 3 or 400 with slowish shutter speeds(sub 1000) at any moving subject I'll get a few that are not as crisp as they maybe should be.
With the 1D3 I was finding that the highest incidents of bad focus were with shutter speeds in excess 1/2000th and that if I reduced the frame rate the problem would all but disappear. I would think that the 1Ds3 would be better in this dept. simply because it is not throwing a fit when you hit the shutter.
I too await some user feedback about the D3, when I'm ready to update my gear I'll not be looking for a Mk3 again and I'm not sure how long Canon will be in developing its replacement. I'll need to look out for some sandals on Ebay if I'm to go down that road though

J

J,

I haven't as yet calibrated my lenses, but interesting it has been said in a number of places that if you set the AF menu adjustment on the 1D£ to -1 it performs better all round.

I wonder if the car owner is sat at home waiting for the speeding ticket...

I'll need to look out for some sandals on Ebay if I'm to go down that road though

I'm going to have to get a boxful and put them on e-bay, there's that many considering buying them!!

Looks good to my eye Tom { i also notice three being fractionally less sharp than the rest}.

Not sure how this would relate to defining acceptably sharp {or near perfect} auto-focus with a wildlife subject though, as the number plate is one dimensional, whereas a focus area placed over a three-dimensional 'head and beak' of an on-coming bird etc is a different ball game.

Looks promising though!!

Alan

Only 2 of the frames look OK to me. Remember that AI Servo mode is supposed to be predictive and if it can't cope with a large contrasty object with lots of edges moving at a constant velocity, I'm not sure how it is going to cope with a small subject, moving erratically with no nice edge contrast to lock on to.

I had a long chat with a guy from Canon a few days ago about the AF issue and the fix. Taking into account what he said, and the results posted by Rob Galbraith, I'm now convinced that both the 1D3 and the 1Ds3 are both fatally flawed and that there is no way to ensure that the frame that you need to be in focus will be so - unless perhaps you use manual focusing.

I get my 1D3 back from Canon tomorrow and then I'm off to Tanzania so I will put some more mileage on it there in bright light conditions. Much as I would like the extra resolution of the 1Ds3, I have decided that I am definitely not going to be buying one until the AF design issues are fully resolved.

Hi Tom

Looks pretty good to me - I don't think with current technology that any camera on the market would track and nail a subject like this with pin point accuracy on every frame and my feeling is that allowing for a little motion blur this is as good as it gets. My old MkII as good as it was would never nail an entire series of images when travelling directly at or away from you. I also feel that because Canon did not acknowledge that there was an issue with the AF system to start with everything they do is now viewed with suspicion and results are scrutinised with a far more critical eye than would normally have been the case. They only have themselves to blame for this!

David

Hi Tom I would say 3 at a push look okay(I would definitely be upset if my d2x didn't nail at least five of those with a large subject like that and my 300f2.8), on a side note I was with a friend with his 1 D-mark 3 latest firmware 113 I was using my d300 & 800 he was using his 500f4 IS+1.4 I was also using it as well just to see the difference the latest firmware made, our conclusion was not a lot it wouldn't even focus consistently on stationery cormorant he is definitely hoping that the sub mirror fix will improve it even though he doesn't think it will.

Phillip.

Juza's review inc autofocus with a 600mm F4

Got around to doing the second set. Obviously I shot one coming at me and one going away.

First the full image

Now the sequence of which there are 10 which is the buffer limit on raw ( and I have given myself a slap on the hand for the pixel dimessions of the image)

Same as above hand held, 300mm F2.8 but this time the image as a whole (all plates) have been subjected to a (150, 0.2, 1) USM treatment once.

Given this would be about 2" wide on a native print of about 18" wide I don't think I have a problem for this is hand held, you cannot even see this detail to manually focus on it, the vehicle is accelerating down a hill about 40mph and this is the straight 10 buffer in sequence. I will try calibrating the AF next (sometime).

Tom,
Are you going to try your camera on a bird?
Frederic

When I can get a break from work... yes

I thought you were always in holidays...
Frederic

Wow!
Just printed, at 240dpi, a 24"x16" print from the RAW file that Juza had as a test shot.
No interpolation and the detail in the fur is incredible.

I'd better stop looking at the print, or I might do something stupid, like go and buy one

Have fun Tom

John