This seems like a trick question akin to “How long did the Hundred Years War last?” (116 years, it turns out) or “Who is buried in Grant’s Tomb?”. Ulysses Grant is not buried, but entombed (above ground) and he shares that space with his wife, Julia. So while the question seemed to provide the answer – like these two seemingly obvious cases, the 30 second exposure, it turns out does not take 30 seconds.
I have 3 DSLR cameras: a Canon 40D, a 50D and a 5D Mark II. In “high speed continuous exposure mode” the 40D can shoot 6.5 frames per SECOND – that’s faster than anyone can say “click” six times. The 50D tops out a 6.3 fps (frames per second), and the 5DII is a still-speedy 3.9 fps.
But I take long exposures in night environments. A 30, 60, or even 480 second exposure is not unusual in my nocturnal world. Surprisingly my cameras that can take 3.9 to 6.5 frames per second under good light, actually take an average of 32.6 seconds for each 30 second exposure when in high-speed continuous mode and it seems to not matter at all whether the exposures are RAW or JPG, whether the card is a speedy one – a 30Mb/second or a paltry 8 Mb/second.
How is it that a 30 second exposure takes 32.6 seconds? I asked Canon about that and didn’t get a very good answer. Something about “longer exposures take longer to process.” Indeed, some sleuthing and testing reveals that on average as exposures go longer than 1 second, they incur a penalty between shots of about 10%. Two 20 second exposures take about 21.8 seconds. Two 30 second exposures take 65.8 seconds (not 60). Fortunately after 30 seconds, the penalty remains at about 2.8 seconds and does not increase farther.
Why does that matter you might ask?
It may matter a lot. A meteor streaks by every minute or so under a great meteor shower like the Perseids which peaks on August 12/13. A typical Perseid shower tosses out a meteor once a minute on average. Taking continuous 30 second exposures means about 10% of the possible streaks will be missed. We *could* take longer exposures, but then two things happen that are not so much fun.
- The stars begin to noticeably streak.
- Sky glow and light pollution begin to decrease the contrast between the stars and meteors and the dark sky.
What’s more, the delay between shots also means that a plane flying across the sky will have gaps in successive images. When joining the images together (stacking) the star trails formed by their motion may have gaps as well even though those gaps may not be readily obvious. In the image below, the gaps are not very noticeable until you see the larger sizes.
Granite Park Star Trail – 59 minutes with “gaps”
As a result of the “processing delay” it is necessary to change the way images are captured.
- A delay of 3 seconds may need to be be included when attempting to take successive exposures using a timer on a Canon body. UPDATE: The Canon 5D Mk II apparently is able to buffer the writing to the card so that a second shot can be started with a delay of only one second – however if you use the continuous drive mode a 2.8 second delay is introduced.
- We will want to shoot a long as we reasonably can so there will be fewer gaps.
Considerations like these are just one of the subjects we endeavor to teach participants in our Star Circle Academy Workshops. I hope you’ll join us on one soon!
Pingback: Trouble with Long Exposures – Part 2 of 2 | Star Circle Academy
Its the dark current leakage noise cancellation technique that canon uses. After the exposure ends, camera closes the shutter, and takes a reading from the sensor, ideally there should not be any reading at all, but few pixels will report some voltage. That pattern is then used to to determine what is actual light generated voltage and what is noise voltage. If you have Long Exposure noise reduction turned on (one of the C-Functions) it may help.
I mean If you have Long Exposure noise reduction turned off(one of the C-Functions) it may help.
Adarsha, as I may not have fully noted in the article “LENR” and High ISO NR are both off. Indeed, if LENR (Long Exposure Noise Reduction) is on, 30 second exposures will now take 60 seconds – 30 for the exposure and then 30 for the “dark frame” exposure.
The behavior remains, however – even on the 5D Mark II which is able to physically start a second exposure if you do it manually. However in “continuous exposure” mode, the 2.8 second delay between shots is an unavoidable penalty. I have also found that setting an intervalometer to allow a 1 second delay between 30 second exposures usually works – but not always, even for the 40D and 50D. It’s the “not always” part that is problematic!
It it makes you feel any better, it’s pretty much the same with Nikons too.
I own a Canon 60D and tested with a 30-second exposure, and indeed it took around 32.4 seconds to complete. However I fired a flash after 31 seconds had passed, and the flash light is clearly visible. This means there is no ‘dark time’ and that the shutter is open and the sensor is recording for the entire duration of 32.5 seconds. Also, I usually shoot stars with Canon’s EOS Utility (so I have the camera connected to the laptop), and I did not notice any problems with interval-shooting, but I’ll have to check in more depth before being able to say for sure.
Pingback: Interval…what? | Star Circle Academy
John’s observation is an interesting one. It is possible that 30″ is really 32.8 seconds (or very close to that). It’s also possible that the actual intershot delay is not the 2.8 seconds I’ve been asserting… but it’s clearly not zero, and it’s clearly MORE time than the inter-shot delay when shooting in high speed continuous exposures in daylight settings. So the next step is to actually measure the inter-shot delay. One way is to do so using the stars since the earth spins at a constant rate.
Meanwhile I now have a Nikon D600 and I have measured the 30″ shot interval at 32″. I haven’t repeated the experiment enough to come up with sub-second decimal point accuracy, but it does seem to be slightly faster than any of my Canons. Of course that doesn’t mean that the delay between shots is actually shorter as John noted.
OK, so I’m a bit new to the star trail pics but I’m planning on making a go of it in a week or so during the next new moon. I own a 5D Mark II and I’m borrowing a friends 5D mark III. I’ve got two tripods and two of the Canon remote Intervalometers and plan on shooting RAW on both cameras. I’ve done a bit of testing in the house here and noticed if I set the remote to an interval of 1 second and the exposure to 20 or 30 seconds it seems to be working OK. I’ve run a short test and haven’t noticed the buffer filling up.
Is this the best way to set it up?
I tried setting the interval to 21 or 22 seconds and exposure to 20 but it seemed to trip up every once in a while and miss, waiting another 21 seconds before firing off the next shot.
Also, I’ve rented a 24mm 1.4 Rokinon for one of the cameras and I’ll be using my 24-70 2.8 on the other camera. Any suggestions which camera should get which lens?
If you are using an intervalometer to drive your camera, make sure your camera is in BULB mode. Otherwise your camera and your intervalometer may not agree on exactly how the exposure actually is. I’ve never seen my 5DII miss a shot when set to a one second interval when in bulb mode. But if you’re seeing it, unfortunately you only have two options: set a 2 second interval between shots, or switch to a device like the TriggerTrap which will allow you to set a fractional interval like 1.3 seconds.
Your other questions are more appropriate for the http://starcircleacademy.com/startrails page so I’ll answer them there.
I feel like I am getting the bet results so far with my 1D3 by setting to high speed continuous mode in live view mode and locking my remote ‘on’ (using between 2s on 200mm to 25 for 14mm). Now that I found your article I will run some ‘tests’ to see what I am really getting in these different modes. I also have a 50D I can try the same on.
Thanks for posting these!
You didn’t state how you’re taking those shots. If in high speed continuous mode with the shutter latched down, we’re pretty confident you’ll find the same strange “lost seconds” behavior – even with a 2 second exposure where the loss is likely to be about 0.2 seconds. If you are driving the camera with an intervalometer you are not likely to see the discrepancy.
Looks like I will be using the intervalometer a LOT more – no matter how I did it a 30s exposure took 32s of camera time… even the shutter would just quick click to the next frame, but it did so every 32 second! freaky and THANK YOU for sharing this.
Now to tame the mirror slap on this 1D3 – OOF! (good tripod and live-view do help)