Canon EOS-10DCanon revamps their hugely popular D60 SLR, with ahost of improvements and a dramatic price cut!
<<Exposure & Flash :(Previous) | (Next): Operation & User Interface>>
Page 7:Shutter Lag & Cycle Time TestsReview First Posted: 02/27/2003
Shutter Lag/Cycle Times
When you press the shutter release on a camera, there's usually a delay or lag time before the shutter actually fires. This time allows the autofocus and autoexposure mechanisms time to do their work and can amount to a fairly long delay in some situations. Since this number is almost never reported on, and can significantly affect the picture taking experience, I now routinely measure it.
NOTE: My qualitative characterizations of camera performance below (that is, "reasonably
fast," "about average," etc.) are meant to be relative to
other cameras of similar price and general capabilities. Thus, the same
shutter lag that's "very fast" for a low-end consumer camera might
be characterized as "quite slow" if I encountered it on a professional
model. The comments are also intended as only a quick reference: If performance
specs are critical for you, rely on the absolute numbers to compare cameras,
rather than my purely qualitative comments.
|Power On -> First shot||
||Effectively zero, as there is no lens to retract. Camera may spend more or less time writing the contents of its buffer memory to the memory card, but that'll depend a great deal on what you were shooting, how full the buffer was, and how fast the memory card is that you're using.|
|Play to Record, first shot||
||Time from playback mode to ready to shoot. Very fast, as camera has "shooting priority," is always ready to shoot.|
|Record to play (max res)||
||Time to display large/fine file immediately after shot is captured. Fairly fast Camera apparently has to write image to card before it can display.|
|Shutter lag, full autofocus||
||AF speed will vary greatly depending on lens used. This time was with Canon 16-35mm zoom lens, focusing at fairly close range (about a meter). Very fast.|
|Shutter lag, manual focus||
||Time with same lens as above, but set to manual focus mode. Also very fast. (In the "full AF" time above, the lens was already focused on the target from the previous shot, hence the AF and manual focus times are the same.)|
|Shutter lag, prefocus||
|Delay with shutter button half-pressed and held before the exposure. Very fast. Second time is for Continuous AF with the camera in AI Servo mode, also quite fast. Not the very fastest times I've seen in an SLR (which are somewhere around 0.059 seconds), but much faster than most consumer cameras.|
|Cycle Time, large/fine JPEG||
||Very fast. Shoots at around 0.42-second minimum intervals for first nine shots, then slows to about one second per shot. Time is with Lexar 256MB CompactFlash card. With slower Mr. Flash card, times are similar. First nine shots at about 0.42-second minimum intervals, then about 1.22-second minimum intervals.|
|Cycle Time, small/basic JPEG||
||Very fast. Nearly identical performance to large/fine files, same buffer capacity, only buffer clearing time reduces with small files.|
|Cycle Time, RAW mode||
||RAW mode cycle times are the same while filling "first buffer" (first eight shots), drop to right about two seconds for "second buffer." Buffer clearing times ranged from 131.36 seconds with Mr. Flash card to 50.62 seconds with Lexar card. Note that RAW mode also embeds a JPEG file. Times are for RAW plus a small/basic JPEG.|
|Continuous Mode, large/fine JPEG||
|This fast for nine shots. Buffer clearing ranged from 30.25 seconds with Mr. Flash to 13.8 seconds with Lexar.|
|Continuous Mode, small/basic JPEG||
|Same speed and run length as large/fine files. Buffer clearing from 6.95 seconds with Mr. Flash to 5.37 seconds with Lexar.|
|Continuous Mode, RAW mode||
|Same speed and run length as large/fine. Buffer cleared in 37.13 seconds with Lexar card. (89.60 seconds with Mr. Flash.)|
Overall, the EOS-10D is a pretty fast camera. It's roughly three frames/second in continuous mode is pretty equivalent to the speeds of competing SLRs, and shutter lag is very good. The 9-frame buffer memory is great for capturing rapid action, and in all but the most intensive sports photography applications, should prove more than adequate.
Two Buffer Memories?
As with the D60 before it, though I was puzzled by how quickly the 10D seemed to process each shot after its buffer filled (e.g., after the first nine shots), but how long it took to completely clear the buffer. If it was snapping additional images every second or so, how could it take 28 to 85 seconds for the buffer to empty completely? Out of curiosity, I experimented with much longer run lengths, and discovered that the 10D apparently has two buffers (just like the D60): After the first buffer fills, shooting proceeds at a somewhat slower pace for some number of shots (until the second buffer fills?), after which cycle times increase pretty dramatically.
Shooting in continuous mode, saving files in the large/fine JPEG format, the first nine shots were captured at intervals of about 0.34 seconds. The interval between shots eight and nine was 1.14 seconds, and then the intervals between shots nine through 27-35 were about 0.84 seconds. (The point at which the "second buffer" filled varied quite a bit depending on the subject matter I was shooting.) After that point, the interval jumped to about 2.4 seconds, and remained there. Shooting in continuous mode and saving RAW format files, the behavior is similar, but the number of shots the "second buffer" holds is less, and the eventual slowdown much more severe. In RAW mode, performance is identical for the first 9 shots, then the camera pauses for 2.04 seconds before the 10th shot, and then alternates between 1.67 and 1.88 seconds for the next 6 shots. Finally, it slows to about 7 seconds between shots for shots number 17 and above.
Unlike the D60 though, I didn't observe this behavior in single-shot mode. After the first nine shots that filled the main buffer memory, the cycle time between shots fluctuated wildly, ranging from 0.61 to 4.03 seconds.
I'm not sure how to interpret these results, but the bottom line is a bit more graceful degradation of cycle time once you get past the limits of the primary buffer memory than is usually the case.