Response time - a job killer

I just thought I would put something in the title to interest my friends on the political right.

A few years ago, our main camera was a Fujifilm S602 Z, a very nice machine with a generally-excellent electronic viewfinder. But a big problem with it was "effective response time". As I said in a report at the time, "By the time it fired, the councilman had already taken his hand off the mayor's knee".

The delay in part was direct, just from the camera's overall mode of operation, and was in part indirect, in that the image in the finder was delayed from real-time (partly a result of the modest refresh rate), thus the term "effective response time".

It was largely this issue that moved us into the dSLR world, with a Canon EOS 300D (EOS digital rebel).

We are right now in Las Vegas, attending Carla's 55th high school class reunion. As part of our interest in being able to travel with a smaller, lighter burden, we decided to leave behind our workhorse EOS 40D and EF-S 18-200 and the 40D kit bag, and to shoot everything with our new Powershot SX20 IS.

Last night was the opening reception, and I had horrifying flashbacks to the era of the S602. Every time I had a group of two couples in a nice grouping and fired, I got one person leaning behind another, or turning around to look at someone that just came in. Autofocus time, incidentally, was not part of the equation, as I almost always prefocused.

One part of the problem seemed to be the flash metering preflash cycle. (I was operating with a Canon Speedlite 270EX.) Of course, there is that delay component with the 40D as well, but somehow I have the feeling that it takes more time as practiced on the SX20 IS.

In any case, when we get home, I plan to do some tests to separate out the various components of the total effective shooting lag.

But it becomes clear that the SX20 is not really the weapon for last night's form of combat.

Just a thought as the various modern camera configurations are contemplated.

Best regards,

Doug

One thing I would like to see in the future is high frame rates and very large buffers. If the camera can run continuously with full resolution at 60 FPS with a 60-frame buffer (exposure duration permitting), then it could offer user-customized response time. If the viewfinder delay is 250ms, the camera's delay is 500ms, and your own personal delayed response time is 250ms, then you would set the variable to 1000ms. When the shutter release is pressed, it grabs the photo out of the buffer. Wouldn't work in your case, of course, because of flash.

I did some primitive response time tests this morning with the EOS 40D and the Powershot SX20 IS.

I used a test clock we made a while ago, with a large hand rotating at one rev/s. It can be read to 1 ms.

Of course, I have no way to reliably synchronize the shutter release full press with the clock hand crossing zero. My technique was to make several runs for each setup, then throw out the obvious "outliers" and average the remainder of the results. There is no guarantee, however, that my error would have been symmetrically disposed about "on the money", so there could be systematic bias in the results.

These test series' were all done with the camera at ISO 400 using P metering mode and AF. I each case, I would half press in advance to lock AF and AE.

Here is a summary of the findings. The value stated is the apparent time of the shot, in ms, with respect to when I sought to full press the shutter release, based on my observing the clock hand in the way described for each series:

Canon Powershot SX20 IS

No flash; observe clock hand directly (not through camera finder): 91 ms

No flash, observe clock hand through eyepiece finder (EVF): 188 ms

Onboard flash, observe clock through EVF: 376 ms

Speedlite 270EX flash, observe clock through EVF: 430 ms

Canon EOS 40D

No flash, observe clock through reflex viewfinder: 86 ms

Onboard flash, observe clock through reflex viewfinder: 115 ms

Speedlite 270EX flash, observe clock through reflex viewfinder: 115 ms

*********
Again I emphasize that there is great opportunity for substantial experimental error in this set of tests.

Still, the indications should be useful.

In summary:

• The basic shutter release delay (in the indicated mode of operation), not considering the effect of viewfinder lag where applicable, was 5 ms shorter for the 40D than the SX20 IS.

• The effective shutter release delay (in the indicated mode of operation), that is, considering the effect of viewfinder lag where applicable) was 102 ms shorter for the 40D than the SX20 IS; on the SX 20 IS, the increase in effective delay over basic delay (the apparent viewfinder latency in this situation) was 97 ms. Note that viewfinder latency may well depend on ambient illumination level or many other factors; no probe of such was part of this series of tests.

• On the SX20 IS, the use of the onboard flash increased the effective release delay by 188 ms; the use of the Speedlite 270EX increased the effective release delay by 242 ms (the total effective delay in that case being 430 ms).

• On the EOS 40D, the use of the onboard flash increased the effective release delay by 29 ms; the use of the Speedlite 270EX also increased the effective release delay by 29 ms (the total effective delay in either case being 115 ms).

• If we consider operation with the 270EX flash, the overall effective shutter release delay with the EOS 40D was 315 ms less than with the SX20 IS.

This seems to say that we will need to use our EOS camera to shoot serious indoor social activities. (Well, they're so much more impressive anyway!)

Best regards,

Doug

I have found, generally, that faster cameras make me happier when taking pictures, and particuarly when people are moving. Appreciate your tests are fraught with potentia experimental error, but the results are not surprising. when I used e-tt flash on my canons it always slowed them noticeably, particuary the 1Ds3

Mike