Live time and live bulb: Is it really a free lunch?

Started Dec 3, 2012 | Discussions thread
ForumParentFirstPreviousNextNext unread
Flat view
Anders W
Forum ProPosts: 17,925Gear list
Like?
Live time and live bulb: Is it really a free lunch?
Dec 3, 2012

Ephotozine just voted the live-time features introduced with the E-M5 the best new technology of 2012:

http://www.ephotozine.com/article/olympus-livetime---best-new-technology-2012-20754

In spite of being an old darkroom junkie who certainly enjoyed seeing my images form in the developing dish back in the film era, I am not sure I'd like to go quite that far. But I have no difficulties agreeing with Ephotozine that this is a useful feature with a sprinkle of fun on top.

What is still not clear to me, however, is precisely when it's a good idea to use it. One purpose for which the feature is obviously useful, no matter what, is as a means of determining approximately the right exposure time for long exposures. Without live time/bulb, this is likely to be a more time-consuming process of trial and error.

But once you know the proper exposure, is it a good idea to use it for the actual shooting too? For scenes that aren't perfectly static, e.g., fireworks, it certainly offers potential advantages in this regard as well. You can stop exposure at precisely that point when the image looks just right.

The question, however, is if this advantage is for free or if there are hidden costs involved. In order to show the image as it gradually forms, the sensor must be read multiple times. If these read-outs are non-destructive, so that the charges on the sensor remain completely intact until the final read-out, then there should be no problem. If, however, the charges are destroyed by each read-out, and the final image formed by gradually aggregating the data colllected, then there is a cost in terms of read noise. For example, if the sensor is read 25 times, which is the maximum number allowed at base ISO (24 previews plus one final read-out), the accumulated read noise would be five times higher (the square root of 25) than if the sensor had been read only once, when the exposure was finished.

It is hardly perfectly clear which of these two scenarios we are actually dealing with. One reason is that there appears to be sensors with as well as without the capacity to be read in a non-destructive fashion. Another is that the camera imposes limits on the number of live time/bulb read-out it allows prior to the final one. At base ISO, it allows up to 24 but at ISO 1600 only 9. At still higher ISOs, the live time/bulb feature presumably doesn't work at all. Why would this be the case if the sensor could actually be read in a completely non-destructive fashion?

So I performed a little experiment with my E-M5 to find out. I shot two black frames at base ISO, both using live time, both with an exposure time of about 25 seconds, and both with black-frame subtraction turned on as you normally would for long exposures. For the first shot, I set live-time on with updates once a second so that 25 read-outs were perrformed. For the second, I used time but turned live-time off. I then checked the files by means of RawDigger.

Interestingly, the read-noise distribution, which you can normally inspect in full on the E-M5, was in this case clipped in both shots, possibly because of the black-frame subtraction. For the file shot without live time, the standard deviation of this clipped distribution was 1.0 for all channels alike. This is the value we would normally expect for the unclipped distribution in a black-frame shot using a short exposure time so the long exposure time itself adds a bit to the read noise as would be expected. The corresponding value for the shot with live time turned on varied from 1.4 to 1.9 depending on which channel we are talking about.

This clearly suggests that using live-time is not quite the free lunch one could ideally hope for. On the other hand, the results are noticeably better than would be expected if the image had actually been the results of 25 different destructive read-outs. Consequently, it seems that the actual process does not quite conform to either of the two extremes that one can theoretically outline in advance but falls somewhere inbetween. Luckily, however, it appears to be closer to the best-case than to the worst-case scenario.

Comments and questions are of course welcome.

 Anders W's gear list:Anders W's gear list
Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Olympus E-M1 Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS Panasonic Lumix G Vario 7-14mm F4 ASPH +21 more
ForumParentFirstPreviousNextNext unread
Flat view
ForumParentFirstPreviousNextNext unread
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow