D300s, Long exposure noise reduction

[Seapy]

I am going out tonight to try to get some star tail images.  I had been holding out to use the D300s on a 4 or 5 hour single exposure but am concerned about heating the sensor and generating much internal sensor noise in one image.  I am visualising a featureless brown image...

So, I plan to take a series of 10 or 20 second exposures using the built in intervalometer with as little delay between the exposure as possible then stacking the images and blending them, either in Ps or StarTraX blending software.  My previous experience of multiple long exposures is a long delay while the camera processes the long exposure noise reduction process.  I understand this can take the same amount of time  as the actual exposure, hence another reason for not making a single multi hour exposure.

My question is can I turn off the Long Exposure NR and do something similar in Ps.  I understand I need to make a long exposure with the lens cap on to act as a reference by revealing the 'hot pixels'.

Or, can the camera process and save the previous image at the same time as exposing the next?

I want each exposure to follow as quickly as possible after the previous one so there are no breaks in the star circles, otherwise if it has to be an interval of 15 seconds there will be more gaps than trail! I feel the maximum gap between exposures would be 2 seconds, I will be using my Nikkor 20mm f2.8 @~f4.

[Øivind Tøien]

If you use DeepSkyStacker (free) to combine the images (without alignment) you should be able to subtract one or more dark frames and turn off Long Exposure Noise Reduction in the shooting menu. However I found that with the bodies below it worked well without dark frames, but then it was pretty cold when I tested. Example down the page here: http://nikongear.net/revival/index.php/topic,2411.45.html

Note that on both my D5100 and D7100 bodies a 30 sec. exposure is not 30 sec. but closer to 32 sec. Thus on my 7100 I have to set the interval to 33 sec for the timer to work as intended with a minimal gap between exposures. I have found it very useful to do a dry test before trying the real exposures...

If one does not live in a very dark place without light pollution, the multiple exposure approach may be the only way, as a single several hour long exposure would completely blow out the frame.

[Seapy]

Thank you Ølvind, that's very helpful, I am only experimenting, I have been waiting to find my remote release which would have allowed me to take a long 'bulb' exposure but now I am realising it's probably not the best way.

I will go out to my local viewing point now and try some exposures.    Not much risk of the camera overheating more perhaps freezing!

[Bjørn Rørslett]

If the ambient temperature is low you should be able to make shorter exposures, say up to 10 sec, without the need for LNR. There might be the occasional "hot" spot caused by noise,  but as you are adding many images and noise typically is randomly distributed, most of this should disappear or be reduced to insignificance during the final processing.

To be able to handle any stationary hot pixels, run a separate basic exposure of the same length with LNR on, then use that as a template to further mitigate hot points.

Edit: Øivind made basically the same advice while I was typing

[Asle F]

Better than guessing how long the intervals have to be and miss some seconds between every exposure, it is an easy way to just set the camera at 30s, continous and lock the remote release. Then the camera will start the next exposure right after the previous is finished.

[Øivind Tøien]

 

To be able to handle any stationary hot pixels, run a separate basic exposure of the same length with LNR on, then use that as a template to further mitigate hot points.

Edit: Øivind made basically the same advice while I was typing

I think you probably mean to say that the dark frame has to be taken with LNR OFF like the other exposures?   


Asle, does not some bodies have a limit to the number of exposures that can be captured in continuous mode?

[Bjørn Rørslett]

Øivind: I stand corrected. In my suggested setup, LNR should be off.

As to the number of consequential images, there probably are limits set by either camera or the release device. Perhaps 999 frames or something like that? Buffering ought not to be a problem if the buffer is read-out and emptied during the next frame. Well worth trying out in practice.

[Seapy]

Thank you for the extra info and discussion, very helpful, in a  way confirmed my thoughts and help steer my technique.

I made about 300 exposures at 30 seconds, f3.5 @ ISO800 taken at 33 second intervals (to allow for possible longer than expected exposures), they don't look too bad at first run through.  I made three 30 second exposures with the lens cap on, so I have three reference files, although I have roughly compared the images and while there are some constant bright hot pixels, there are also some random ones too, guessing 50/50?  Won't have time to process them fully today, maybe  this evening?  There was a little light passing cloud which may interfere with the final stacked image but we will see.

There is another clear night forecast for tonight, possible snow tomorrow... So I would rather get out tonight and capture rather than sit in front of computer!  Yet I need to see if my technique is good...

I have been trying to think of easy way to check the shutter time, I accept that it may interfere with the interval exposing and getting it thighter with less interval between exposures should make for a better final image.  I plan to make an exposure of my clock with second hand, make a 30 second exposure of the clock, covering the lens just after the start of the exposure then removing the cover at about 25 seconds which should leave a trace of the second hand at the start and at the close of the shutter.

[Seapy]

It seems I have a similar overlong 30 second shutter time. within the equipment to hand it seems to be greater than 32 seconds, just. I would need a proper stop watch with split seconds to get better timing. My clock second hand jumps every second, it isn't a smooth movement.

I lit the clock with an LED torch to give a harsher light and covered the lens for most of the exposure, except the first second and the last couple of seconds.

[Bjørn Rørslett]

Have you read out the EXIF data on shutter time? It tends to be accurate within 1/10 of a second for long exposures (until the counter overflows ...).

[Øivind Tøien]

Have you read out the EXIF data on shutter time? It tends to be accurate within 1/10 of a second for long exposures (until the counter overflows ...).

That probably applies to B setting, but if set to nominal 30 sec. it will report that even if the actual time is slightly over 32 sec. Perhaps you could test on a few of your bodies? I suspect this is by design. A nominal 15 sec exposure is 16 sec or so, while a 20 sec. exposure is as I recall closer to 20 sec. It might be the way shutter speed divides/multiplies, 1, 2, 4, 8, 16, 32 sec would give full stop increments. It could also be due to how the processor controlling shutter speed needs to count.

Edit: My approach to recording the actual shutter speed reported here was simply to use the chronometer function on my watch.

[Asle F]

Taking 2 serial exposures at 30s, the difference in clock time in the two exif will tell the absolute minimum interval between two exposure.

[Seapy]

Have you read out the EXIF data on shutter time? It tends to be accurate within 1/10 of a second for long exposures (until the counter overflows ...).

EXIF says: 4906891/1000000

That does not make sense to me, I can't get an answer anywhere near 30 seconds (or 32 seconds) no matter how I calculate it.

Do we have any photography mathematicians who can explain please?

[Øivind Tøien]

I see no such field as ShutterSpeedValue (where you read that value from above) reported by ExifTOolsGUI for my D7100 files (perhaps that is a field used by other brands indicated by the negative value?). Look at ExposureTime, it reports 30/1 seconds.
 

[Seapy]

Cant make head nor tail of it, I will do further experiments when time permits, for now my first attempt at a star trail photograph:

Made from about 55, 32second exposures from Nikon D300s, Nikkor 18-105 VR lens @ 18mm, f5.6, ISO 1600, No 'in camera' noise reduction. Exposure interval 33 seconds. Manual focus and Manual exposure.  Images processed in Lightroom CC, stacked in StarStaX and cleaned up in Photoshop CC.



Close examination may reveal a series of lights travelling down mid right hand side which I am assuming are a satellite, perhaps the ISS?  The series of visibility covers about 40 minutes of time, which is MUCH slower than any aircraft or meteor. Also I don't know why it appears to be visible, then not visible. It also appears to be describing an ark.