I did it. I went out and got my 600D last night with a decent tripod and stuff but didn't get a new lens. Just got a single lens kit. Anyway, I was readying a blog post just before about taking pictures of the Milky Way (I wanted to learn the technical side not just 'Expose the ass off it!") and I found this post about the 600 Rule
Basically it states that the maximum length of an exposure with stars that doesn’t result in star streaks is achieved by dividing the effectivefocal length of the lens into the number 600.
Apparently the maths doesn't work out 100% when you get up the full frame sensors though.
Hi Mate, I shot the Milky way from time to time and I use the rule of 600 on my D7000. (Small sensor) but as you can tell from my avatar it works pretty well but is only a rough guide so best to play around with it a little. But in saying that...600 is what I work off. :-). (If you do zoom in on your final image there is small trailing but normaly its fine unless you really go looking for it!!!
Hi Mate, I shot the Milky way from time to time and I use the rule of 600 on my D7000. (Small sensor) but as you can tell from my avatar it works pretty well but is only a rough guide so best to play around with it a little. But in saying that...600 is what I work off. :-). (If you do zoom in on your final image there is small trailing but normaly its fine unless you really go looking for it!!!
Going to use the Rule for the first time tonight. Just started looking at the details for my camera
14mm widefield lens on a full frame goes about 30 seconds usually with no trouble. But you will find those stars further away from the suth celestial pole move faster (more angle to cover) than those closer to the celestial pole so you can get in a widefield image stars streaking in some parts of the image and tight in other parts.
The solution is use a widefield lens like 10mm on APS sensors or 14 on full frame, use your 30 seconds, use maxmium ISO before noise becomes too bad (probably around ISO3200) and long exposure noise reduction on. White balance may vary with cameras with Nikon being a little green/yellow and Canon being a little bluish but around 4200K should be close. Save as a RAW that way you can fiddle with the white balance mostly afterwards.
If you want longer than that you need a Vixen Polarie or similar product. Then you can get 2 or 4 minutes if you set it up just right and are within the weight contraints.
50mm in my experience is way too long a focal length to use without tracking then it is awesome. 24mm on full frame I would consider the cutoff point and is pushing it a bit.
The rule should really state - the longest focal length you can use at 30 seconds (anything less than 30 secs ISO3200 is going to be underexposed) is 24mm on full frame or 15 seconds on APS sensors without tracking.
I don't know why camera manufacturers of DSLRs have maximum time of 30 seconds before the stupid bulb setting has to be used. Its like they think noone in their right mind will need more than 30 seconds. Hey - we're being discriminated against!!
Location: '34 South' Young Hilltops LGA, Australia
Posts: 1,323
Revised rule for DSLR's
These rules or formulae are based on a set of starting assumptions and have been kicking around for a long time.
In the old days of 35mm film where very fast films used for tripod astrophotography had very course grains and so a bigger movement could pass off as no drift. It was based on someone looking at a 8x10inch enlargement at normal reading distance perceiving trails as almost stationary. The eye can resolve about 1 arc minute so it can in fact resolve approximately 200 microns on a print which scales back to 25 microns on the film. However the number 700 corresponds to a trail length of about 50 microns. This is about 9 pixels on a modern DSLR. But today, we mostly look at screen based images and images that are highly reduced from full size.
The number is related to how much drift you find acceptable. So I've derived a revised formula, modernized for DSLRs.
max exposure time(s) = [14*N*P] / [FL*cos(d)]
where
N..........Number of pixels of drift
to work this out load any full frame image from your camera and scale it to the size you want to view on screen. Look at the scale percentage. Number of pixels drift = 100 / percentage
P..........Pixel size (microns) Most DSLR's are between 5 and 8 microns. You can look it up in your manual or just split the difference and make it 6.5
FL.........Focal length(mm)
cos(d)...cos of the declination. Use the declination of the stars in the field that are closest to the celestial equator in the camera field
Example : Using a 14mm lens, 5 pixels drift, on an APS-C sensor with 5.4µm pixels pointed at the celestial equator can take a 27 sec exposure. Pointed at the SCP the lens will see 45 degrees each side of the pole extends the time to 38 seconds.
(14 * N * P / FL) * cos(d)
or
14 * N * P / (FL * cos(d))
Associative rules for multiplication say the first but it would be good to be sure. I would write that as 14 * N * P * cos(d) / FL which is less likely to be misinterpreted.
Location: '34 South' Young Hilltops LGA, Australia
Posts: 1,323
Quote:
Originally Posted by mithrandir
Joe, is that supposed to be read as:
(14 * N * P / FL) * cos(d)
or
14 * N * P / (FL * cos(d))
Associative rules for multiplication say the first but it would be good to be sure. I would write that as 14 * N * P * cos(d) / FL which is less likely to be misinterpreted.
Andrew,
Sorry about that :
It is:-
max exposure time(s) = [14*N*P] / [FL*cos(d)]
I've edited the original post to reflect this.
The cos(d) term is 1 for a declination of 0 so you can leave it out all together if you want to calculate for stars on the celestial equator.
Andrew,
Sorry about that :
It is:-
max exposure time(s) = [14*N*P] / [FL*cos(d)]
I've edited the original post to reflect this.
It seemed to make more sense that way than the other so I thought I check it.
Quote:
The cos(d) term is 1 for a declination of 0 so you can leave it out all together if you want to calculate for stars on the celestial equator.
To make use of that you need to know the image center and FOV to work out "d".
eg APS-C long axis 23.3mm. At 11mm -> 93° FOV so pointed at dec -45 with that axis aligned n/s some stars are on the equator.
Location: '34 South' Young Hilltops LGA, Australia
Posts: 1,323
Quote:
Originally Posted by mithrandir
It seemed to make more sense that way than the other so I thought I check it.
To make use of that you need to know the image center and FOV to work out "d".
eg APS-C long axis 23.3mm. At 11mm -> 93° FOV so pointed at dec -45 with that axis aligned n/s some stars are on the equator.
Andrew,
Yes that's right. That's why I say declination of the stars in the field closest to the equator rather than dec of the stars at the center of the field of view.
I'm still pretty new at this having similarly just bought a 650d, but I have already come up with my own simplified equation for exposure without star trails. Its easier to use that the others and much along with Gregs comments. Its:
30 seconds or less / conditions on the night.
The attached photos are using an 8mm lens and with a 30 second exposure. Everything looks great zoomed out, but I chose an area with the LEAST trailing to enlarge.
I have been reading the Nightscapes section avidly, and that seems to be the general concensus of exposure before having to resort to making a composite image with the stars tracked.
The 600 rule gives 8mm @ f4.5 = 16 seconds which is better but will require higher ISO and corresponding noise issues.
I haven't done Joes' maths, but of course the key factors are fl, pixel size (chip size is sort of irrelevant) and rate of movement of the field. I'll stick to my equation for the timebeing
Location: '34 South' Young Hilltops LGA, Australia
Posts: 1,323
Quote:
Originally Posted by AstroJunk
I'm still pretty new at this having similarly just bought a 650d, but I have already come up with my own simplified equation for exposure without star trails. Its easier to use that the others and much along with Gregs comments. Its:
30 seconds or less / conditions on the night.
The 600 rule gives 8mm @ f4.5 = 16 seconds which is better but will require higher ISO and corresponding noise issues.
I haven't done Joes' maths, but of course the key factors are fl, pixel size (chip size is sort of irrelevant) and rate of movement of the field. I'll stick to my equation for the timebeing
Hi Jonathan,
In which universe does 600/8 = 16 ? Is this some sort of corollary to Douglas Adams 6x9=42?
None of these things are perfect, they are just a guide. My formula takes no account of the light spread of brighter stars which makes a trail longer but also wider nor does it give you trail free images. Of course if you zoom right in then N, the number of pixels drift has to be made smaller and your time becomes much shorter with commensurate increasing noise problems.
Tripod photography is always best expressed using a small print or small screen image to minimize visibility of trailing.
All these rules fall over when you go to these ultra wide and fisheye lenses. The focal lengths stated are obviously not the real focal lengths. Your lens is trailing more like a 12mm focal length lens than an 8mm.
In which universe does 600/8 = 16 ? Is this some sort of corollary to Douglas Adams 6x9=42?
I blame the PangalacticGargleblasters myself.
I had rather telescopically taken the aperture multiplied by the focal ratio to be the fl. As I said, real cameras and me are a very new thing! I'll google what those numbers actually mean.
To be fair though, I had done 600/8 = 75 and thought that was such an absurd number that it couldn't be true...
Ah yes the good old "600 Rule". I have a blog post here http://www.capturingthenight.com/blog/?p=12 on how I use it for my fixed tripod night sky scapes. Personally I use 500 or less when imaging. It's not perfect as stated, but it's a pretty good guide.
Quote:
Originally Posted by AstroJunk
To be fair though, I had done 600/8 = 75 and thought that was such an absurd number that it couldn't be true...
With your new 650D being a non frame camera then you need to factor that into the equation as well Jonathon, but yeah you don't need to worry about the aperture. Canon's crop factor is 1.6, so with your 8mm it is effectively (for the purpose of the 600 Rule) 12.8mm (8 times 1.6). 600/12.8= 46.8 seconds maximum shutter speed before the stars trail too much to be really noticeable. As I said I tend to do the 500 rule or less. 500/12.8= 39 seconds but for that lens I would just stick it on 30 seconds.
These rules or formulae are based on a set of starting assumptions and have been kicking around for a long time.
In the old days of 35mm film where very fast films used for tripod astrophotography had very course grains and so a bigger movement could pass off as no drift. It was based on someone looking at a 8x10inch enlargement at normal reading distance perceiving trails as almost stationary. The eye can resolve about 1 arc minute so it can in fact resolve approximately 200 microns on a print which scales back to 25 microns on the film. However the number 700 corresponds to a trail length of about 50 microns. This is about 9 pixels on a modern DSLR. But today, we mostly look at screen based images and images that are highly reduced from full size.
The number is related to how much drift you find acceptable. So I've derived a revised formula, modernized for DSLRs.
max exposure time(s) = [14*N*P] / [FL*cos(d)]
where
N..........Number of pixels of drift
to work this out load any full frame image from your camera and scale it to the size you want to view on screen. Look at the scale percentage. Number of pixels drift = 100 / percentage
P..........Pixel size (microns) Most DSLR's are between 5 and 8 microns. You can look it up in your manual or just split the difference and make it 6.5
FL.........Focal length(mm)
cos(d)...cos of the declination. Use the declination of the stars in the field that are closest to the celestial equator in the camera field
Example : Using a 14mm lens, 5 pixels drift, on an APS-C sensor with 5.4µm pixels pointed at the celestial equator can take a 27 sec exposure. Pointed at the SCP the lens will see 45 degrees each side of the pole extends the time to 38 seconds.
cheers
Joe
Very interesting read Joe. Thanks for posting. Might have to have a play around with your new formula one night....
Cheers
Greg
Ah yes the good old "600 Rule". I have a blog post here http://www.capturingthenight.com/blog/?p=12 on how I use it for my fixed tripod night sky scapes. Personally I use 500 or less when imaging. It's not perfect as stated, but it's a pretty good guide.
With your new 650D being a non frame camera then you need to factor that into the equation as well Jonathon, but yeah you don't need to worry about the aperture. Canon's crop factor is 1.6, so with your 8mm it is effectively (for the purpose of the 600 Rule) 12.8mm (8 times 1.6). 600/12.8= 46.8 seconds maximum shutter speed before the stars trail too much to be really noticeable. As I said I tend to do the 500 rule or less. 500/12.8= 39 seconds but for that lens I would just stick it on 30 seconds.
Great explanation on your blog Greg - thanks for taking the time to put it together. I feel that the resolution of the chips is beginning to give some of these equations a hard time as even 30 seconds on the 18mp aps-c chip is pushing it let alone 45+ which would provide a clear streak.
However, I have a way to go yet before streaky images are my primary concern
Great explanation on your blog Greg - thanks for taking the time to put it together. I feel that the resolution of the chips is beginning to give some of these equations a hard time as even 30 seconds on the 18mp aps-c chip is pushing it let alone 45+ which would provide a clear streak.
Try an APS-C with 24m 3.89um pixels.
With Joe's formula if I want less than 4px of trailing, at 14mm (effective 21.6mm) I get 90° FOV and can't go beyond 16 sec.
Location: '34 South' Young Hilltops LGA, Australia
Posts: 1,323
Quote:
Originally Posted by mithrandir
Try an APS-C with 24m 3.89um pixels.
With Joe's formula if I want less than 4px of trailing, at 14mm (effective 21.6mm) I get 90° FOV and can't go beyond 16 sec.
The crop factor changes the 600 rule to the 400 rule when going from a full frame to APS-C.
However, the crop factor does not change the focal length, it changes the FOV. When using my formula, do not multiply FL x crop factor in my formula.
Also with 24MPx, you can get away with a lot more than 4 Px of trailing. It goes without saying that these fixed tripod images usually get shown a bit smaller. Let's say you can accept 4 Px trailing at the final displayed image and you are going to scale the image to 800x1200 Px(20% scale of the 4000x6000px original).
Then you can have N, the number of pixels as
4 x 100% / 20% = 20 Px
so your max exposure is
20Px * 4µm * 14 ÷ 8 = 140s
When I made the comment about Jonathan's lens working at 12mm, I meant that the pixel smear indicated that the lens was acting more like a 12mm indicating that the manufacturer may have deliberately understated the focal length for marketing purposes. They do this all the time.
If you think about it, there is 45-55mm between the back of the lens mount and the focal plane in most brands of DSLR so any focal length shorter than about 70mm is a marketing label and not a true statement of focal length.
The crop factor changes the 600 rule to the 400 rule when going from a full frame to APS-C.
Joe, that must depend on the camera. Canon have a 1.6 crop. Sony have a 1.5 crop. I forget what Nikon is but it is a bit different again. So does 400 apply to Sony and should it be 375 for Canon?
Quote:
Originally Posted by OzEclipse
However, the crop factor does not change the focal length, it changes the FOV. When using my formula, do not multiply FL x crop factor in my formula.
I added the chip size to my spreadsheet so it can calculate the "effective" focal length so I can use that in the FOV.
Quote:
Originally Posted by OzEclipse
Also with 24MPx, you can get away with a lot more than 4 Px of trailing. It goes without saying that these fixed tripod images usually get shown a bit smaller. Let's say you can accept 4 Px trailing at the final displayed image and you are going to scale the image to 800x1200 Px(20% scale of the 4000x6000px original).
That's certainly true on IIS. 800x1200 is about as big as you can upload at any reasonable quality setting.
If today's predicted thunderstorms clear tonight I'll see what 80sec looks like at 14mm. (Jonathan is the one with the 8mm lens.)
The numbers show why the 150-500mm @500 has to go on my StarLapse. You don't get much light in 2.3sec.