I gave mine the high speed track test yesterday. It really did a nice job. The hockey photos were taken in a rink wiht half the lights on. The camera had no trouble tracking the fast pace of hockey. These were taken @ ISO 6400, I used DPP to convert them and used with the noise filter on. The auto focus also did a nice job with the dog shots. I was able to grab 3-4 shots in mid air for each jump and all are sharp. I had a MkII and sold it for a 5D, I shoot mostly weddings. I forgot how quick the shutter is on the 1 series camera!!!
gmaxwell wrote:
The only sigma lense I own, the 15mm full frame fisheye, works fine.
.........
...
Does the sigma work with the AF micro adjustment AF custom function?
There is one thing that I think would be good to address in a firmware update. Page 184 "registering Basic Camera Settings" is all well and good, but if I am shooting to multiple cards, it has no way to set this as part of the procedure. Therefore, whenever I press apply basic settings, it also wipes out my recording to two cards, setting it back to the default of recording to CF only ("standard mode"). I then have to reset to record to multiple cards, both CF AND SD.
I thought I would give some rough timings on cards, they differ from what Rob Galbraith was getting. I used Sandisk Extreme III CF and Sandisk Extreme III SD cards, both 4 GB. Please also note that I have my low speed shooting rate jacked up to 5 fps. The high rate is still at 10 fps. I held down the shutter until the camera filled it's buffer. From that point I timed how long it was until the red light went out- a fully cleared buffer. I shot in RAW + Small jpeg mode. At 5 fps it got in about 30 shots or so before buffer filled, and of course at 10 fps, it was probably about 22.
I tried different card combos, either writing "standard" (one card) or "record to multiple" which duplicated the files onto both types cards at once. In all cases, 1/500 sec shutter speed was selected, and the resulting images were grossly underexposed, very dark. So the image size may be smaller than normally exposed shots. I didn't care about exposure at the time, just timing.
10 fps----
28 sec CF only
54 sec CF +SD
52 sec SD only
5 fps----
30 sec CF only
56 sec CF + SD
55 sec SD only
please note that at 5 fps, slightly more images are recorded; buffer can recover a bit more between shots. These are not timings for being able to take an additional shot, these are timings to completely clear buffer of all shots just taken and red light to go off.
I've found the ISO safety shift handy. If I combine it with limiting the slowest shutter speed in custom functions, I can set up three sets of custom functions to allow me to
set 1: don't let me shoot below 1/8 sec, no mirror lockup. For use handholding IS lens or on monopod. ISO shifts up if light isn't available.
set 2: "normal" 30 sec max shutter but always has mirror lockup enabled.
set 3: don't let me shoot below 1/60 sec, no mirror lockup. For walking around, non IS lens, and for flash sync without background blur.
I may not want to waste my three custom settings just based on shutter speed, but ISO shift does open up interesting opportunites. I'm still experimenting.
jb987 wrote:
I have performed some high ISO shots and comparison to the 1 Ds Mk II. I found this interesting cause the pixel size of the 1D Mk III and the 1Ds Mk II is nearly the same.
My conclusion: the 1D Mk III has a one stop advantage with respect to high ISO performance. Not more and not less. All comparison was done without noise reduction applied.
They did keep the pixel-size the same. However, they made some changes to the microlenses that helped bring in more light, IIRC.
Wow, the ISO shift sounds great, because it will automate (and idiot-proof) the way I shoot late afternoon/evening outdoor sports, which is to shoot Av, ratcheting the ISO up as it gets darker, then switching to M at 1/400 once ISO maxes out at 1/400. If I'm not careful and try to push it too far, I inevitably blow a few shots when I let the shutter drop to 1/320 or less. Cool.
Jim Hayes wrote:
I thought I would give some rough timings on cards, they differ from what Rob Galbraith was getting. I used Sandisk Extreme III CF and Sandisk Extreme III SD cards, both 4 GB. -----
I shot in RAW + Small jpeg mode. At 5 fps it got in about 30 shots or so before buffer filled, and of course at 10 fps, it was probably about 22.
I tried different card combos, either writing "standard" (one card) or "record to multiple" which duplicated the files onto both types cards at once. In all cases, 1/500 sec shutter speed was selected, and the resulting images were grossly underexposed, very dark. So the image size may be smaller than normally exposed shots. I didn't care about exposure at the time, just timing.
10 fps----
28 sec CF only
54 sec CF +SD
52 sec SD only
5 fps----
30 sec CF only
56 sec CF + SD
55 sec SD only...Show more →
Thanks for posting this Jim. It's interesting to me because I use the Extreme III 4GB cards with the Mark II N and was wondering how they will perform with the Mark III. I'm sure many like me with Mark IIIs on order will be interested in similar feedback from those who have already received their cameras since the type of memory card will definitely impact camera performance.
Your results unfortunately give some concrete numbers to what I suspected might be the case, that the increased size of the 14 bit RAW files will significantly impact buffer timing. It appears RAW performance will not be a substantial improvement over the Mark II N, both in the time to clear the buffer and in the total number of shots stored in the buffer. What ISO was set for your initial tests?
Are you able to repeat the tests for RAW only? I am curious whether you'll get the 30 frames claimed by Canon. I would also be interested in the actual number of shots the buffer will store at low and high ISOs (100 vs. 1600 would be good).
It makes me wonder what the issue is here for Canon? Maybe it's not a simple matter of just upping the amount of buffer RAM, but whatever the case, there is still a buffer issue for those of us who opt to shoot RAW or RAW+Jpeg for high paced action.
Lastly, can anyone comment whether there is a significant write speed penalty when using 8GB cards vs. 2 or 4GB? Based on the Mark II N timings at Rob Galbraith's CF card database, there was a significant slow down with most 8GB cards, yet with the Mark III's larger RAW files, 8GB would seem like an ideal choice.
rscheffler wrote:
...Lastly, can anyone comment whether there is a significant write speed penalty when using 8GB cards vs. 2 or 4GB? Based on the Mark II N timings at Rob Galbraith's CF card database, there was a significant slow down with most 8GB cards, yet with the Mark III's larger RAW files, 8GB would seem like an ideal choice.
The Ultra II is an anomaly in this regard — the 8GB is *really* slow, relatively speaking. But look at the Extreme III — the 4GB and 8GB are in a dead heat, with the 8GB actually edging out the 4GB on jpg writes (by an amount well within the timing margin of error). Yes they're slower than 2GB cards, but that's likely because you're making the jump from FAT to FAT32 there.
I like using 8GB cards in my Mark II's. They do slow down as they get full, but DavidP's trick of formatting them with larger cluster sizes seems to ameliorate this to a significant degree.
I'm surprised that the SD slot is so much slower than CF in your timings.
Dan Martin wrote:
Those sets sound interesting Jim. Is it possible to override the ISO shift if you have it enabled?
A number of ways come to mind. I've been using AV mode mostly. If I spin the back dial it sets exp compensation, no other buttons to press first. So I can send the ISO back down, and risk a darker shot.
I can spin the main dial on top and change aperture and send the ISO back down.
I can limit max AND min ISO speed in custom functions. So If I didn't want to shoot higher than ISO 800 say, or below ISO 200, I could do that. I don't recall if the restriction goes by full ISO stops or 1/3. I know the shutter limit goes by full stops.
I can also limit max and min aperture just like shutter and ISO.
I could add the safety shift custom function to "my menu" so it's a little more able to be acessed.
I could create two custom function sets, one with and one without ISO shift. Then put the CF set selection in "my menu".
If I shoot in Manual mode, the ISO doesn't shift, but the shutter speed limits are still in effect...I assume this goes for aperture and ISO limit setting as well.
Edited by Jim Hayes on May 28, 2007 at 06:48 PM GMT
rscheffler wrote:
Thanks for posting this Jim. It's interesting to me because I use the Extreme III 4GB cards with the Mark II N and was wondering how they will perform with the Mark III. I'm sure many like me with Mark IIIs on order will be interested in similar feedback from those who have already received their cameras since the type of memory card will definitely impact camera performance.
Your results unfortunately give some concrete numbers to what I suspected might be the case, that the increased size of the 14 bit RAW files will significantly impact buffer timing. It appears RAW performance will not be a substantial improvement over the Mark II N, both in the time to clear the buffer and in the total number of shots stored in the buffer. What ISO was set for your initial tests?
Are you able to repeat the tests for RAW only? I am curious whether you'll get the 30 frames claimed by Canon. I would also be interested in the actual number of shots the buffer will store at low and high ISOs (100 vs. 1600 would be good).
It makes me wonder what the issue is here for Canon? Maybe it's not a simple matter of just upping the amount of buffer RAM, but whatever the case, there is still a buffer issue for those of us who opt to shoot RAW or RAW+Jpeg for high paced action.
Lastly, can anyone comment whether there is a significant write speed penalty when using 8GB cards vs. 2 or 4GB? Based on the Mark II N timings at Rob Galbraith's CF card database, there was a significant slow down with most 8GB cards, yet with the Mark III's larger RAW files, 8GB would seem like an ideal choice.
ISO was set to I believe 3200. And remember it's hard to count frames going by at 10 fps.
One thing I will state because no reviewer has said it yet is that the jpeg compresion for all sizes (L, M1, M2, S) can be set indivually from 1 to 10. So you could set a large jpeg to have a compresion of 1 or 2 and a small jpeg to 10, and the large jpeg might(?) not be much larger in file size than the small jpeg...but I haven't tried this. Default for all jpeg sizes is "8".
Okay, I did some more tests. It's hopeless for me to try to count how many shots before buffer fill, even at 5 fps. I decided to rather express it as length of time until buffer filled, then total time to clear buffer (red light goes off after last shot). Since in many cases it's 2-4 seconds to fire off a buffer full of shots even at 5 fps, I may not be accurate with this number. I'm timing these with my wrist watch!!
Conditions: RAW only, Manual mode, 1/500 sec, f4 70-200 F4L IS (on) lens. ISO either 100 or 1600, speed either 5 fps or 10 fps. ISO 100 shots all black (grossly underexposed), 1600 ISO shots properly exposed but defocused to a blur with no detail. Extreme III 4 GB... using CF and SD version of card. Writing either "standard" mode when to one card, "multiple" when writing to both cards identical files.
I only did each test once, my battery needs recharging and the shutter actuations really rack up.
The first time before the "/" gives an indication of # of shots fired; so a 5 sec burst is better than a 3 sec burst. The second number gives an indication of how quickly the images can be completely cleared and writing stops. As stated, there may be up to a sec error on first number and maybe 2-3 seconds on second number as I'm eyeballing my watch not a prescion timer.
Writing to CF only:
ISO 100: 10fps: 3 sec burst/followed by 26 sec buffer clear
Jim Hayes wrote:
Okay, I did some more tests. It's hopeless for me to try to count how many shots before buffer fill, even at 5 fps. I decided to rather express it as length of time until buffer filled, then total time to clear buffer (red light goes off after last shot). Since in many cases it's 2-4 seconds to fire off a buffer full of shots even at 5 fps, I may not be accurate with this number. I'm timing these with my wrist watch!!
The way I did it when testing different cards on my Mark II was, as I recall, something like this:
I'd reformat the card. Then I'd simultaneously hit the button on the stop watch and the shutter button. I'd hold down the shutter button until the second "slow" frame. This is the key to what you're trying to do, and it's not hard once you figure it out. The camera goes clickclickclickclick...clickclickclick 20 times, then when the buffer fills it slows down to clickclickclick...click...click. You can't catch the first slow one, but it's easy to stop after the second. On the Mark II this gave me a consistent 22 RAW-only frames each time.
I did it in manual mode at 1/500 with the lens cap on so the frames would be as identical as possible.
Now to find those timings I did so we can compare them to yours....
Actually the pixel size is smaller in the 1D3 vs the 1D2N. The size in the mark III is 7.2 um vs 8.2um for the 1D2N. So the 1D2 had a larger pixel size.
From the white paper:
Pixels Size and ISO Speed
Camera
EOS-1D Mark III 7.2x7.2
EOS-1D Mark II N 8.2 x 8.2
EOS-1Ds Mark II 7.2 x 7.2
The pixel-pitch of the 1D-III is definitely smaller than that of the 1D-II . . it has to be, as it has more pixels in it for the same sensor size.
However, because Canon reduced the "wasted space" between the acutal light-collecting photosites, they were able to utilize the same size photosites in both cameras.
Additionally, Canon worked some magic on the microlenses that collect the light for the photosites, resulting in better light collection.
Thus, the dynamic range (signal to noise) of the 1D-III is better than that of the 1D-II, even though the pixel density has increased.
Thanks for taking the time to do that. One way to determine the total number of shots is to clear the card between bursts so you only have images from the last set, which are easy to count, or make a new folder for each burst and in folder view it will indicate how many files are in each. Take the total MB of a folder and divide by the time taken to capture & clear the files to get the MB/sec value.
Nill:
Regarding the 8GB cards, I also noticed that in the Galbraith database, but you never know, performance could be different in the Mark III. With the larger Mark III RAW files and a likely move to larger CF cards, I'm particularly interested in performance with UDMA cards, like the Sandisk Extreme IV, Lexar 300x, Transcend 266x, etc and especially between 4 and 8GB capacity. It appears UDMA cards will make transferring to computer the nearly double sized RAW files slightly faster than Mark II files stored on non UDMA cards (based on an equal number of total files).
I'm also somewhat perplexed by Jim's SD results as it would appear based on the Galbraith Mark III preview that SD could be faster. I wonder which SD cards he used?
Again, if anyone else cares to comment with timing from their camera and cards, it would be appreciated.
rscheffler wrote:
...Regarding the 8GB cards, I also noticed that in the Galbraith database, but you never know, performance could be different in the Mark III.
Could be, but the 4GB and 8GB Extreme III are also nearly identical on the Mark II and the 1Ds Mark II, FWIW.
I'm also somewhat perplexed by Jim's SD results as it would appear based on the Galbraith Mark III preview that SD could be faster. I wonder which SD cards he used?
He said he was using 4GB Extreme III's in both cases, so I was even more surprised to see his results.
DavidP wrote:
The pixel-pitch of the 1D-III is definitely smaller than that of the 1D-II . . it has to be, as it has more pixels in it for the same sensor size.
Yes, but in addition the 1D MK III has a smaller (more cropped) sensor than the 1D MK II.
is that the jpeg compresion for all sizes (L, M1, M2, S) can be set indivually from 1 to 10. So you could set a large jpeg to have a compresion of 1 or 2 and a small jpeg to 10, and the large jpeg might(?) not be much larger in file size than the small jpeg...but I haven't tried this. Default for all jpeg sizes is "8".