5D3 max iso for Milky Way

[Nate]

Hi,

I have read quite a lot on this topic. I want to shoot some milky way shots on my next vacation.

Now I am shooting with a 17-40 and was wondering what the 16-35 could do.
I mean if I go f4 6400iso and 30 sec, would I see more stars than with f2.8 640iso 30 sec after post processing?
I ask this question because I see many photographers on 5D2 using 3200iso or even 1600iso with 2.8.

Also 6400 iso on 5D3 with long exp noise red, in post processing the noise will be almost completely gone?
Because if I could use f4 for stars and dont need extra stop, I dont want to invest money in a better glass.

Thanks!

 

[scarey83]

If I've understood what you're saying, f4 @ 30s with ISO6400 will likely show more stars than f2.8 @ 30s with ISO640. f2.8 is double the light of f4 but ISO640 is one tenth the light of ISO6400.

More generally, I've found that many astrophotographers use as high an ISO as possible to maximise sensitivity (I suspect ISO 6400 on the 5D3 is fairly usable anyway)
To get rid of noise, rather than us the in-camera noise reduction, many turn this feature off and instead, take a dark frame (put the lens cap on) with the same settings. This 'dark' image can then be subtracted from main image.
You can also add images together to increase the detail (and then add dark frames to improve the noise subtraction). There are a number of other frames you can stack to remove different sorts of noise.

I've used DeepSkyStacker with some success. The processed output usually looks like it's not worked (it comes out light grey or white) but with some extra processing to change the levels in photoshop similar, you can get quite good detail from it.

I've not taken photos of the milky way as you can't see it where I am (light pollution!) but if the place is dark enough, I think 30 seconds at f4 would be adequate to capture it.

scarey

 

[noisejammer]

I strongly suggest you get a copy of Jerry Lodrigus' Beginner's Guide - here's the link http://astropix.com/BGDA/BGDA.HTM . Jerry has written four ebooks and almost everything you need to know is there.

Now on your specific question, remember some things

1. The 17-40/4L has soft corners. This is generally bad for astrophotography. It sharpens up at f/8 to f/11 but this doesn't work either because every stop means you double your exposure (or since you're not on a tracking mount, the available light is halved.) Half the light means roughly half the stars - something that's not really helping.

Rather than the expensive 16-35L, you would do far better if you purchased a 35/1.4 prime lens and stopped it down to f/2 - f/2.8. This allows a much shorter exposure which will suppress star trailing. A 35 mm lens would allow around 7-14 seconds exposure. 7 seconds will give you a blur of no more that 3 pixels, irrespective of where you point the lens in the sky. You can always take lots of images and stack them into a very deep image.

This leads me to an often overlooked point - the sky does not appear to move at the same angular rate for all places. If you point a camera at the celestial equator, stars appear to move 15 arc-seconds per second. If you point it at the celestial pole, you will record movement in the corners and nothing at the centre of the field. This makes deconvolution quite difficult and it's worst for large fields of view.

2. Increasing the ISO doesn't really change the sensitivity of the CCD sensor - all that happens is you get more electrons out per photon in. This degrades the signal noise to photon noise ratio. There's a sweet spot at which the camera noise matches the photon noise and the overall noise is minimised. For the 5D MkII this is one third of a stop above ISO 800 or ISO 1600 (subject dependent.) That makes is ISO 1000 or ISO 2000. The third of a stop comes from the weird way Canon configured some of the signal amplifiers. I can't tell you the optimum for the Mk III but it's probably 1600 ISO or thereabouts - you can easily check this.

 

[Olivier11986]

I may be a bit late here, but you can get great pictures all the way up to ISO 102,400 : http://www.extremeinstability.com/2013-5-4.htm

(well, that is on a 6D, but the 5Diii won't be much different)