Dslr ISO settings

[dph1nm]

Early testing gave it a 75% reduction in noise over DIGIC 4 based systems, which is a big leap.

I always laugh when I see magazines which say this. Shot noise cannot be better - this is dictated by the physics of photon emission (in our case from the astronomical object). Read noise or dark (thermal) signal may be better, but most terrestrial photo mags or review sites never measure these. I suspect what the photo mags are talking about is the noise in processed jpegs. These are highly manipulated by the software, but best avoided for astro.

NigelM

[iwatkins]

No, this was in the RAWs. This was in the Canon Pro magazine, not some Amateur Photography magazine.

But indeed, it will be tested to death once the DSLRs actually come out. We shall see.

Cheers

Ian

[Gina]

Yes, it'll be very interesting

[Cath]

Turning the ISO up does effectively reduce the A/D quantization noise (compared to the wanted signal). But the bad thing about having too high an ISO setting is that you saturate the lighter parts of the image (the star light etc).

[iwatkins]

Indeed Cath.

Which is why I'm a big proponent of using multiple and tested plans for individual DSOs. E.g. M42, lots of ISO1600 subs for the nebulosity, then a few lower ISO/shorter subs for the smaller stars and then lower/ISO shorter subs for the larger/brighter stars. Then stack the resultant stacks in PS afterwards using luminosity masks to get good detail in all regions and more importantly, great star colour. Call it HDR if you like, but it does work.

Of course, a full plan will be different for each DSO.

Cheers

Ian

[themos]

That makes sense to me, too.

[dph1nm]

I would seem that to achieve the noise reduction the DIGIC 5 processor "analyses four times more image data to accurately produce each pixel of the image". No thanks!

NigelM

[RikM]

For my 1000D I got a good result on the Monkey Head by combining a set of ISO1600 subs and a set of ISO800 subs. 30x each at 5min stacked as two groups in DSS. I tried a set at ISO400 but post-stretch these looked noiser than the others. I didn't even include them for the star colour because it didn't give me any extra.

It is something I will certainly be trying again.

[themos]

"analyses four times more image data to accurately produce each pixel of the image"

Now that they've run out of "Megapixels! more megapixels!" ways of selling cameras, they are starting to bin pixels, effectively. It's kind of funny. The engineers at Canon must be tearing their hair out at the marketers' shenanigans.

[iwatkins]

It's out: Canon announces EOS 5D Mark III 22MP full-frame DSLR: Digital Photography Review

Sample differing ISO images: Canon EOS 5D Mark III low light ISO samples: dpreview review samples: Galleries: Digital Photography Review

Cheers

Ian

[rocketandroll]

...didn't someone say there was a new 1D coming out too... a 1D-X or something that would go to ISO280,000?

Someone quoted about £5.5K for the body, and due for release about now?

Ben

[iwatkins]

Yep, end of April. Expandable to ISO204,800 and yours for £5200. Ouch!

New 5D is a more sensible buy if you do normal photography *and* astro photography. If you only do astro photography then I would say a dedicated CCD is probably still a better option.

But if the 7D is replaced with a similar system then it would be a more difficult choice, well for me it would be.

Cheers

Ian