Star trails - confused

[MartinSykes]

I've managed some OK shots of stars in the past using 5-15s exposures but last night I tried to use a longer exposure for star trails and to be honest I'm confused by what I got. a 100% crop below.

Picture has lost EXIF in editing but the details are:
Sony a580+35/f1.8SAM
Manual mode /22, 1476s, ISO-100
Tripod mounted, cable release, steadyshot turned off
Shot in fine JPEG - RAW would have been better but I forgot.

Now, I also had long-exposure NR turned off so I was expecting a few hot pixels, but this looks extreme so what's going on there?

The few trails I can see are also not proper arcs. They look like they start curved and then straighten out (and maybe even start to curve the other way). I've added a straight line for reference. I'm not sure if that's expected or not. If the camera is pointing in an odd direction can you get s-shaped trails?

[TCav]

I don't see any "trails" at all. You left the shutter open for almost 25 minutes and got no motion blur?

[corkpix]

Originally Posted by MartinSykes

I've managed some OK shots of stars in the past using 5-15s exposures but last night I tried to use a longer exposure for star trails and to be honest I'm confused by what I got. a 100% crop below.

5-15s is a very short exposure for star trails unless you are
merging a sequence of many images. The shot below shows
a 20s exposure of Orion. This was taken at 17mm. A longer
lens would show more motion.

Manual mode /22, 1476s, ISO-100

25 minutes is a very long time. It seems there is a fairly
large accumulation of noise over this period. I have never
tried such a long exposure. I found a really nice star trail
image on pix.ie recently. It was taken with a Canon 40D
using 16m exposure. The EXIF is intact so it might be
of interest: http://pix.ie/necronomicon/2114620/in/album/398851

Now, I also had long-exposure NR turned off so I was expecting a few hot pixels, but this looks extreme so what's going on there?

Lens cap left on? If the Sony uses dark-frame-subtraction to reduce
noise. I would enable it. The Canon implementation certainly works well.
Unlike other forms of NR, it doesn't destroy image detail.

The few trails I can see are also not proper arcs. They look like they start curved and then straighten out (and maybe even start to curve the other way). I've added a straight line for reference. I'm not sure if that's expected or not. If the camera is pointing in an odd direction can you get s-shaped trails?

I'm no sure what to make of that. Are you sure it is a star trail and not
a meteor? Could it be a reflection in the lens caused by car headlights
or some other bright light? Could the camera have moved slightly?
Even if the tripod is steady, the legs might tend to sink in soft ground.


Orion through the trees. January 2012.
Canon 50D, Tamron 17-50mm.
f/4 20s.

[MartinSykes]

Originally Posted by TCav

I don't see any "trails" at all. You left the shutter open for almost 25 minutes and got no motion blur?

This is a 100% crop showing 1 trail (just above the straight white line which I added for reference). There is also a very faint second trail just above and to the left of that as well if you look closely. I think there are so few trails because of the low ISO but there are a dozen or so over the entire image that all have the same shape. I'm assuming that all of the static dots are noise.

cork - the 5-15s shots were intended for static stars and not star trails. I don't think the tripod would have moved - the ground is hard at the moment due to the temperature here but it's possible the head could have dipped during the exposure under the weight of the camera if it wasn't tight enough.

The dark frame subtraction does seem to work well on the Sony but from what I can tell, a 15m exposure would be followed be a 15m dark frame so it's halving the number of pictures you can take. If you want to stack the frames it's hopeless because you get a 5m trail, then a 5m gap then a trail,gap etc. so you get dotted trails.

[VTphotog]

Is that f/22 at ISO 100? Your exposure allowed lots of time for pixel noise to accumulate, but didn't allow enough light from the stars to give you good trails. Since the stars are moving relative to the camera, f/22 wouldn't let in enough light for them to show up well.
The tracks of the stars will only be true arcs if you have Polaris centered. Pointing the camera anywhere else will give different results. The tracks will vary depending on the angle of the stars relative to the elevation of the camera.

brian

[MartinSykes]

Originally Posted by VTphotog

Is that f/22 at ISO 100? Your exposure allowed lots of time for pixel noise to accumulate, but didn't allow enough light from the stars to give you good trails. Since the stars are moving relative to the camera, f/22 wouldn't let in enough light for them to show up well.
The tracks of the stars will only be true arcs if you have Polaris centered. Pointing the camera anywhere else will give different results. The tracks will vary depending on the angle of the stars relative to the elevation of the camera.

brian

I think I see what I did wrong. I started with something like 5s at f/5.6 and 1600 which gave me some nice pictures of static stars against a reasonably black background. I then decided I wanted about a 30 minute exposure to get trails so I started doubling the time and dropping either the ISO or the aperture by 1-stop to give me the same overall exposure:

5s at f/5.6 and 1600, 10s at f/5.6 and 800, 20s at f/5.6 and 400, 40s at f/5.6 and 200, 80s at f/5.6 and 100, 160s at f8 and 100, 320s at f11 and 100, 750s at f16 and 100, 1500s at f22 and 100

What I guess that gave me though was the same total light from each star but spread over the length of the trail which obviously made them very dim. The problem I have though is that if I keep the ISO high and the aperture wide for a long exposure then the light pollution will kill the picture.

Is that basically what I'm facing? From my back garden with typical suburban light pollution I can get away with static shots of the starts but for light trails I really need to be somewhere much darker?

[Ozzie_Traveller]

G'day Martin

Reading the multiple posts gives some interesting insights into your 'fun & games'

I love star trails too - so may I offer my 2-bob's worth here using experience from 2 cameras and techniques ... Pentax and Panasonic

Looking at your original image and your OP question - in one sense I cannot answer 'why' you have curved streaks - however I have myself got funny lines when using the Pentax on Bulb with a 15-minute exposure + Sigma 18-125mm lens. My only presumption was they were 'possibly' caused by a stray street light at 90-degrees to the camera hitting the lens so obliquely that some of the glass curvature was transmitted onto the image. But it's only a guess

I now use a home-made lens hood at night to exclude any possible sideways light - and my feeble torch when I'm checking that the camera's still clicking away merrily without dew on the lens [which gets wiped off carefully every 5 minutes if it is there]

As to exposures - I found with the Pentax that when I used Bulb + 5-minutes to 30-minutes + NR='off', the sensor soon was overcooked and the image washed out. It was a complete waste of time

The only successful star trails I got then were via shorter exposures at higher ISOs. Using both the Pentax & the Panny, if NR = 'on' then there is a 15-sec 'on' & 15-sec 'off' exposure series which gives me a) breaks in the image and b) no significant effect with noise reduction ~ sometimes I actually got more noise when NR was turned 'on'. So these days all NR is 'off'

Nowadays I use ISO-1600 x 15-secs x f5,6 x lens at 18mm to 35mm and shoot for 45 minutes to 180 minutes and stack the images using the freebie "startrails.exe" from www-startrails-de

I can post some samples if you like
Regards, Phil

[MartinSykes]

Thanks All.

I guess I'll have to try stacking them. I just thought I could get a single 15min exposure but it looks like the noise is a killer. It's cloudy here tonight so I'll try at the weekend if it clears up.

[TCav]

This is a 30 second exposure of Orion at f/8, into which I inserted the moon.



The only source of light is the stars, and they're moving, so they won't ever cause any one portion of the image to overexpose. If you use a larger aperture, the brighter stars will get brighter, and so will the dimmer stars you missed by using a smaller aperture.

Since you don't have to pay for developing the film, and all you lose is time, try lots of different exposures, and see which gives you the results you want.

[BillDrew]

You got the moon in proper scale. The moon is 0.5182 deg across when viewed from earth. The stars track at .25 degrees per minute, or 0.0625 degrees in 15 seconds, a ratio of eight (0.5185/.0625). The moon in the photo is seven or eight times wider than the length of the trails.

Try stuff and see what works. Something that is not going to work is to getting a photo of the moon stationary while the stars move.