Milky Way D800E

[gregbradley]

I took this shot as one of many on Thursday night. I took some panoramas, some lake shots. This is a single image out of one of the panoramas.

I was supposed to shoot a time lapse but I didn't hit the on menu item when I thought I had so next morning no time lapse and the next night was extremely windy and overcast. Perhaps tonight!

Nikon D800E, Nikon 14-24mm ED F2.8G at 14mm. A single 30 second exposure at F2.8 and 21mm and ISO6400. ISO 12800 is still useable with this camera and I took a test image to see how well it holds up. But for now ISO6400 is working so I'll keep it at that.

Processed in Nikon Capture NX2 and some minor tweaking in Photoshop CS3.

http://upload.pbase.com/gregbradley/image/145902821/large

Greg.

[iceman]

Looks great Greg! Nice colours.

[gregbradley]

Thanks Mike.

It was a fabulous night. There is a bit of red down low as it was super windy that day and the night before and there was a lot of dust in the atmosphere from it.

I tried to keep it more subtle with the colours as I feel I can sometimes get too ham fisted with colour saturation and I try to rein that in.

Greg.

[Phil Hart]

Greg.. can you do me a favour and take one RAW image at ISO3200 and one at ISO12800 (same aperture and exposure time) and then balance the brightness of the two images (increase exposure of first image 4x) in software and share the result?

cheers
Phil

[RickS]

Nice image, Greg, and great high ISO performance!

Cheers,
Rick.

[colinmlegg]

Nice Greg, I like the natural looking Milky Way colours. :)

[gregbradley]

Certainly. Tonight is currently cloudy but supposed to be sunny tomorrow so hopefully it will clear at some point.

That would be an interesting exercise.

One thing that is common with DSLRs is the dynamic range drops steadily with increased ISO.



Thanks Rick. The latest crop of cameras all seem to have excellent high ISO performance.

Greg.

[CDKPhil]

Nice image Greg,

It is amazing the amount of detail these new cameras can capture in low light.

I would like to do a comparison with the D800E and the 5D3.
From what I have read the Canon holds up better at higher ISO's, but the Nikon has a higher resolving power due to the higher pixel count. It would be good to do a side by side comparison under dark skies.


Cheers

[strongmanmike]

Nice work Greg, reminds me of Close Encounters of the Third Kind :thumbsup:

Mike

[gregbradley]

Yes they are both great cameras. D800 and 5D3 are virtually ideantical in high ISO performance (D800 very slightly better) in RAW. 5D3 is better by a stop or so in jpeg. I think Canon has used the Digic 5 to do a strongish noise reduction routine in jpeg. You could achieve similar in post processing with the D800 doing noise reduction, you can also set various levels of noise reduction in the menus but I haven't played with that yet. Either way high ISO jpeg performance would be handy for some. I find ISO6400 is useable even 12800. I tried 25600 but it starts to break down too much. 6400 is my default for now.
You can up an ISO6400 shot with curves to match a 12800 shot.

The main problem with high ISO is dynamic range falls steadily on DSLRs with higher ISO. Not that there is tremendously high dynamic range in these astro shots so it may be a better strategy to pick a highish ISO where the camera seems comfortable and then up it in post processing rather than use a very high ISO (25600) and then try to clean it up with its reduced dynamic range which I believe starts to become a bit more noticeable.

One reason D800 and 5D3 can do these higher ISOs more cleanly is they have higher QE (D800 is a massive 57% QE and 5D3 is 39% also impressive for a DSLR). As noise is the square root of the signal, a stronger signal means the noise is a smaller component of the image allowing higher ISOs (amplification).




Hehehe, thanks Mike. I have been spooked early on in astroimaging by low lying stars that appear to move and if you look at them long enough they start to appear to be moving towards you!

Greg.

[alocky]

Beautiful widefield - I'd almost convinced myself I need the 14-24mm f2.8 lens, but the 24-70mm is a much more practical thing to own, and that's really sharp for such a lens at full throttle.



Hi Greg - can you elaborate on this a little? Other than the root N improvement in signal to noise you obtain by stacking N frames, I had never heard of this relationship.
Noise is going to be a sum of the thermal noise, skyglow and light pollution, as well as the abberations and diffraction peculiar to the optics, and completely independent of what you're trying to photograph. Amplification (higher ISO) will boost noise and signal identically, whereas the time variant nature of the noise means a longer integration will average out some of the noise (with a limit imposed by the average noise power) while increasing the signal.





I've seen exactly this effect late on an all-nighter and drawn the attention of my observing partners to it. Only when I grabbed my binoculars did I realise those stars were completely staionary and my mind was playing tricks. I wonder how many UFO sightings are due to this?
Cheers,
Andrew.

[gregbradley]

The main advantage of the 14-24mm is the widefield. I shot this at 21mm (it was taped from an earlier session and I thought it was at 14mm).

24-70 is fine, especially if you do panoramas. I intend to do some panoramas with it next as I think it will show less distortion.

As far as noise goes I was simply referring to the mathematical relationship.

If the higher QE sensor gives a signal of say 100 then square root is 10 noise = 10%.

But if poorer QE gave a signal of 49 then square root is 7 and the percentage of noise then would be 14.3%. So %-wise you gained an improvement in signal to noise ratio.

D800E QE is 57% and a lot of DSLRs are more like 28% (approx half).

I believe this is the relationship. QE is king, so is low read noise.

Having said that I don't see much difference between these images and posted Canon 5D2 images or 5D3 images.

Greg.

[Phil Hart]

Interested in where you got those figures? I doubt that's an apples-to-apples comparison as I don't believe the technology difference between the two cameras is as great as that (i recall something about geometric scaling of the figures because of different pixel size.. which is not really what we mean by QE?). Canon and Nikon are both pushing hard within the same envelope of what's possible.. I find it very hard to believe they would be that far apart on a truly equivalent measure.

Phil

[FlashDrive]

Another nice photo ....love seeing your shots Greg :thumbsup:
Bit ' envious ' of your dark site there ... I'll have to see if I can come down one day ( holidays )

Flash .. !! :hi:

[gregbradley]

I'd have to find the post again. It was on DPreview and listed various Canon and Nikon models for some time. Most were under 30%, a couple of 1Ds were up over 50%. I don't think they were using Sony QE figures which as you mention seem to be worked out differently to what we normally consider QE to be. As I say some older Canon 1D models had high QE as well.

I totally believe the high QE though. The D800 has very high QE, you can see it in the performance. Its a Sony Exmor sensor. Its the most advanced sensor I have used. Its the reason I bought the camera - that amazing sensor.

Sensor tech has moved forward by Sony. They are ahead of everyone else. The D800E sensor has the highest ratings by DXO ever.

Having said that it is possible these figures are produced by advanced amateurs and are not official figures from Nikon or Canon (who do not give out QE numbers it seems).

For nightscapes both camera makers make exceptional instruments and its more about the photographer than the tool usually.

Greg.



Thanks for that. I have several others I have processed and yet to post.
I think I have settled on a processing routine.

Greg.

[h0ughy]

thats the first thing i was thinking of looking at the shot:rofl:

[strongmanmike]

Ooooh yeh, I see aaaaall sorts of strange things (http://www.pbase.com/strongmanmike2002/image/140710554/original)in astroimages :nerd:

:lol:

Great shot again Greggles :thumbsup:

[gregbradley]

I can see the resemblance.

I have a series of these. I am getting better at them with each one as I learn what works best.

I am also shooting more and doing some time lapses.

Greg

[Poita]

Lovely, hopefully the just announced D600 will have similar performance.

Close encounters was my first thought as well.

[gregbradley]

I think Peter the D600 will be quite similar with the D800E having a slight edge but not massive. They are both Sony Exmor sensors - same as the Sony A99 and RX1.

I shot panoramas again last night and ISO12800 turned out very well which was surprising. I am thinking it is the best ISO for this camera for this work. ISO6400 is very clean on these cameras and D600 from sample photos released looks very clean. 12800 is starting to break down a bit though with D600 but with some careful noise reduction it might still be good. D800 still holding on at 12800 but I'd say that's as far as it will go. D4 may go ISO25600 but then again maybe not.

Greg.

[Phil Hart]

Greg.. I'd still love to see you do some objective tests of these different ISO settings. Until you stretch some of the lower ISO images (eg 1600), you won't know whether you're gaining anything with the higher ISO settings - how they look on their own is not all that meaningful.

Here's an old test I did with the 40D: http://philhart.com/content/astrophotography-test-canon-20d40d-iso-setting-sub-exposure-length

It's hard to tell the difference between any ISO setting once you process them! I know this is true for the 5DMKII as well. Very little to gain going above ISO1600/3200 when shooting RAW, but a lot to lose in dynamic range IMO.

Phil

[gregbradley]

I took a sample ISO3200 and ISO 6400 and ISO 12800 the other night as you suggested. I'll post them soon.

Yes I am mindful of that dynamic range curve of these DSLRs where the higher the ISO the worse the dynamic range. There must be a sweet spot though and in my use so far I would estimate that to be ISO6400 on this camera. But looking at the tests I might be surprised. ISO3200 seems a sweet spot for 5D2 judging by posted images.

Greg.

[CometGuy]

Hi Greg,

This a great image with the hill in the foreground, nicely done! As other's commented reminiscent of scenes from "Close Encounters of the Third Kind" in which they used Devil's Tower Wyoming as a backdrop.

Regarding actual Quantum efficiencies you can find some good data on Christian Buil's site and on the Point Grey site. The Sony CMOS sensors have very high quantum efficiency as you mention - although I haven't seen anything on the 5D mk3 or 60D yet. I'll dig-up some links if your interested.

Terry

[gregbradley]

Thanks Terry.

I checked Christian's site and he only has a few Canon models there. But he does show 5D2 QE at around 28% averaged across the colours.
I tried to find that post of the various camera models from both companies and their respective QEs but no luck yet.

Greg.