Digital for long star trail exposures?

[Zzyzx]

Am used to shooting long exposures for star trails with Velvia and B&W in formats from 35mm to 8x20. Being able to keep the shutter open for a number of hours while I sleep and get up to change to the next exposure/sheet for a second exposure is a nice luxury I have enjoyed for years.

Are there any digital cameras that allow the long and uninterrupted exposures I am used to? Especially in cold weather where being battery dependant is a problem. By cold I am talking temps around zero to 30 below.(farenheit) where most battery models die quickly.

[ternst]

I did exposures of up to eight hours last winter with a Nikon D3s using an external battery. Worked great, in fact I was stunned at the lack of noise. These were done at 20 degrees F or below.

[Zzyzx]

How did you do this? Nikon tech folks tell me the camera is limited to a 30 second exposure.

[ternst]

That is incorrect and I'm sure they were just talking about the longest actual shutter speed. The camera has a bulb mode just like most, and the shutter will remain open as long as the shutter/release is depressed. The limiting factor is the battery, which is why I used an external battery via AC adapter.

[Zzyzx]

ternst, thank you for clarifying that. Will check it out before committing to a new digicam slr type.

Have gotten used to using the view camera and the Nikons for this type of work and not tried the digital for it yet beyond short exposures of a minute or less for Northern Lights displays.

[Scott Hargis]

With regard to using digital for recording star trails -- you can avoid the buildup of noise and gain other benefits by shooting a long series of relatively short exposures with no intervals between exposures. There's a nifty Photoshop action that layers them all together to produce the final image showing the full arc of the stars. Works splendidly, and allows you to fine-tune the length of the star trails in your final image by simply adding or subtracting shots. Here's a link to the action:
http://www.schursastrophotography.com/software/photoshop/startrails.html

And to a shot I made using the technique:
http://www.flickr.com/photos/scotthargis/4207182025/

[elf]

Max Lyons also has several programs to assist creating star trail shots:
http://www.tawbaware.com/forum2/viewforum.php?f=8

[Sussex Landscapes]

just a quick note, and ive said this before some time ago now, that using anyform of stcking in PS using the standard `lighten` blend mode will leave gaps in your trails. there are work arounds, but its time consuming until you understand the process, or you could use teh clone tool

[Bart_van_der_Wolf]

just a quick note, and ive said this before some time ago now, that using anyform of stcking in PS using the standard `lighten`
blend mode will leave gaps in your trails. there are work arounds, but its time consuming until you understand the process, or
you could use teh clone tool

Wouldn't that require the fastest moving stars to move more than, say, half a pixel between exposures?

Let's see what that means in theory.
So, stars move at their maximum angular speed: 86400 seconds for a 360 degree rotation is 0.0041666... degrees per second.
1/2 pixel on a 5616 pixel sensor of 36mm, 1/11232*36mm = 0.003205 mm (or half the sensel pitch).
Thus,  with a 24mm lens, rotation per 1/2 pixel is 2*atan((0.003205/2)/24) = 0.007652 degrees, which would therefore take a star at least 1.84 seconds.

So if I didn't make a mistake, any pause longer than 1.84 seconds would run the risk of gaps showing up, and combined with shorter exposure times than, say, 20 seconds would run the risk of producing a slightly darker pixel in the trail . Therefore, darkframe subtraction during the shooting is not possible, the interval between images would become too long. One could stack some darkframes taken separately and subtract them with appropriate software afterwards. Exposure intervals of no more than 1 second seem safe with such a setup, and shooting stars closer to the polar axis is even safer.

Does that correspond to your findings, or do you use a much longer focal length or more time between exposures?

Cheers,
Bart

[Sussex Landscapes]

Wouldn't that require the fastest moving stars to move more than, say, half a pixel between exposures?

Let's see what that means in theory.
So, stars move at their maximum angular speed: 86400 seconds for a 360 degree rotation is 0.0041666... degrees per second.
1/2 pixel on a 5616 pixel sensor of 36mm, 1/11232*36mm = 0.003205 mm (or half the sensel pitch).
Thus,  with a 24mm lens, rotation per 1/2 pixel is 2*atan((0.003205/2)/24) = 0.007652 degrees, which would therefore take a star at least 1.84 seconds.

So if I didn't make a mistake, any pause longer than 1.84 seconds would run the risk of gaps showing up, and combined with shorter exposure times than, say, 20 seconds would run the risk of producing a slightly darker pixel in the trail . Therefore, darkframe subtraction during the shooting is not possible, the interval between images would become too long. One could stack some darkframes taken separately and subtract them with appropriate software afterwards. Exposure intervals of no more than 1 second seem safe with such a setup, and shooting stars closer to the polar axis is even safer.

Does that correspond to your findings, or do you use a much longer focal length or more time between exposures?

Cheers,
Bart

i wont quibble on the math, but as i said above its the PS lighten blend mode thats the issue as it fades the end of the star trails regardless of the interval time.
 on another note and if wanting static points of star light then you need the formula as originally devised by astrophotography writer Barry Gordon was 700/F (focal length) for stars at the equator. This was designed to achieve a trail length of 0.05mm which would be indistinguishable from a point of light on film. For the hypercritical the 500/F "rule" came into use, yielding a trail length of only 0.036mm. As you get farther away from the equator the star movement is slower and the "rule" can be revised. At declination 50 degrees you can use 1000/F to calculate the maximum exposure before star trails become apparent.

[Bart_van_der_Wolf]

i wont quibble on the math, but as i said above its the PS lighten blend mode thats the issue as it fades the end of the star trails regardless of the interval time.

That's not how the Lighten color Blending mode works. It just takes the brighter pixel of the two or more layers at that position, there's no fading caused by that blending mode itself. Only if the end and start pixel for two subsequent arc segments are relatively underexposed, then a dip in brightness relative to the segment average exposure can occur.

on another note and if wanting static points of star light then you need the formula as originally devised by astrophotography writer Barry Gordon was 700/F (focal length) for stars at the equator. This was designed to achieve a trail length of 0.05mm which would be indistinguishable from a point of light on film. For the hypercritical the 500/F "rule" came into use, yielding a trail length of only 0.036mm. As you get farther away from the equator the star movement is slower and the "rule" can be revised. At declination 50 degrees you can use 1000/F to calculate the maximum exposure before star trails become apparent.

While I applaude simple rules of thumb, it would be nice to have a reference to its applicability for digital imaging (it's predicted to be cloudy here, so not verification possible). A trail length of 0.036mm will, while roughly the same as the COC criterion traditionally used for 35mm film, span a distance of 5.6 pixels (assuming a 6.4 micron sensel pitch). When translating my calculation (assuming I made no errors) above to a similar rule of thumb, it would be 88/f for a single (6.4 micron pitch) sensel travel distance. Anything longer will no longer resolve to a single pixel, unless the output size is reduced.

Cheers,
Bart

[Scott Hargis]

Haha! I was not nearly so scientific with my shot...I ran a few tests and settled on exposures of about 4 seconds each, with no intervals between (I was using my Canon remote timer to run the camera). Twice during the night I went out to check on the camera and laptop, wearing a headlamp. That resulted in two frames with my headlamp traveling through the scene, which I deleted -- causing two small gaps in all the star trails. The final image uses about 200 exposures, give or take.

[Sussex Landscapes]

given that all exposures are of the same lenght and processed together and that what ever program you use for the stacking(ps or startrails.de or whatever other software is available) they all use the same lighten blend method. for whatever the reason its is this that causes the gaps as illustrated.

i cant say anymore than that. the OP wanted to know about star trails from either one long exp or stacking, both have there pro`s and con`s which up until my post no one had brought up the cons of stacking.

[Bart_van_der_Wolf]

given that all exposures are of the same lenght and processed together and that what ever program you use for the stacking(ps or startrails.de or whatever other software is available) they all use the same lighten blend method. for whatever the reason its is this that causes the gaps as illustrated.

That's what I'm trying to figure out. Is it the blending (I think not) or is it the interval between the indivdual exposures. Hopefully someone else with experience in this technique will chime in. We can only learn from that.

i cant say anymore than that. the OP wanted to know about star trails from either one long exp or stacking, both have there pro`s and con`s which up until my post no one had brought up the cons of stacking.

That's fine. If you have some background info about the examples you posted, e.g. focal length + interval between exposures + exposure time, them it might help to pinpoint the issue (and avoid it).

Cheers,
Bart

[Sussex Landscapes]

32x 5-minute exposures taken at ISO 400, f/5.6, with the Canon 5D and Sigma 15mm f/2.8 fisheye lens, using the TC-80N3 with the time between exposures to set 1 second (it doesn’t shoot multiple exposures if you leave it at zero).

at this point in the discussion, i will point you in the direction of the article my info is taken from. this includes the two pictures posted as mine are in the depths of my system somewhere. though through my own tests (and discussions with Floris and Alister)when we have clear skys in the UK, it does casue the probelms noted and the `fix` of LSB method discribed also works.
The lik will take you to Floris`s website where you can follow the tutorial link: http://www.artinnaturephotography.com/gallery.php?gallery=twilight
and Alisters article: http://www.naturescapes.net/Joomla1.5/index.php/category-photographic-technique/41-photographic-technique/390-alister-benn

[Bart_van_der_Wolf]

So if I didn't make a mistake, any pause longer than 1.84 seconds would run the risk of gaps showing up, and combined with shorter exposure times than, say, 20 seconds would run the risk of producing a slightly darker pixel in the trail .

Well, last night we had a clear sky, so I did some testing just after midnight with my 24mm lens on a 1Ds3 (6.4 micron sensel pitch), and apparently I made a mistake in the above calculation. The stars travel at approx. 1/1.84 for half a pixel = 0.28 pixels/second near the celestial equator, as seen from my 52 degrees latitude position with the 24mm. Interestingly, the momentary (1 second) lapse of star exposure between shots is enough to slightly underexpose each pixel that is being exposed by a star. That is the cause of the dips in brightness in the star trails, and it has nothing to do with the Lighten blending mode which just takes the brightest pixel of layers. The brightest pixel just happens to be underexposed by up to 28% (depends also on the exact position of the star and the AA-filter).

The Screen blending mode uses a different function to add layer brightnesses, so therefore it can reduce the differences between the relatively underexposed pixel at the ends of a track segment and the brighter fully exposed track parts. Unfortunately it also changes the color balance when used on gamma adjusted images, so special precautions would be required to circumvent that. It's better to use a method that doesn't require fixing gaps, although that may result in many more images that need to be taken for a desired trail length.

Cheers,
Bart