#1  
Old 07-10-2015, 08:44 PM
Camelopardalis's Avatar
Camelopardalis (Dunk)
Drifting from the pole

Camelopardalis is offline
 
Join Date: Feb 2013
Location: Brisbane
Posts: 5,425
Thoughts on DSLR exposure time

Folks,

I've read a few of the _really_ interesting threads about finding optimal exposure times for CCDs, but wondered if there was a way to similarly estimate for a DSLR?

Having taken a bunch of bias and flat frames on my 1100D, I used the DSLR settings Pixie script to calculate the following values using ISO800 selected on the camera:

Gain = 0.36 e/ADU
ISO for unit gain = 288
Read noise = 3.92 e
Bias = 2048 ADU
Thermal noise = 3.73 e (using dark frames of 10 min at ~5C)

If I open an example light frame taken at the same temperature and ISO (a shot of the Grus 3/4 quartet with the Esprit) into PI and hover the pointer over the background, I see average pixel values of 0.032 - 0.035, with the small bright spots of galactic cores around 0.15.

My technique, for want of a better word (!), has been to expose so the histogram is away from the left hand edge, and preferably around 1/4 the way towards the right. Which works well with larger objects, like Andromeda and such, so maybe the example sub I looked at is not a good...example...as there are numerous bright stars in the field with pixel values around 0.25, which I suspect are dominating the luminance in the image, but at least there is some signal in the image that is exposing as intended (misguided or otherwise!)

In my example above, I figure I could have exposed for easily double the time and obtained a different signal. But since I'm using an uncooled DSLR, in generally cool ambient temperatures, is the signal any better?

So...at what point can I be sure that the signal I'm trying to capture is greater in value than the bias/read noise, and/or thermal noise? And where is the sweet spot?

Any insights gratefully received
Reply With Quote
  #2  
Old 11-10-2015, 12:09 AM
Shiraz's Avatar
Shiraz (Ray)
Registered User

Shiraz is offline
 
Join Date: Apr 2010
Location: ardrossan south australia
Posts: 4,918
you can still use the simple formula of http://www.iceinspace.com.au/forum/s...d.php?t=117010

TargetADU = BiasADU + 10*RN*RN/camera_gain

with the following qualifications:
- the formula assumes negligible dark current, but the DSLR will have dark current that increases as the chip warms through an exposure (ie, it isn't negligible and it isn't a constant effect)
- the DSLR pixels look at the sky through Bayer filters, so the measured sky brightness will depend on which filter the test pixel is under.

From your measured data, it seems reasonable to take the RN and add a bit to account for dark current fluctuations - and use that as the noise term in the formula. Assuming a system noise of ~6e yields a TargetADU of a bit over 3000, which is what you could aim for. When measuring the sky background to see how it compares with the targetADU, suggest you move the cursor around and try to estimate the miniumum sky level, rather than the average and then vary the exposure so that the min is at the TargetADU (this gives good SNR in the least sensitive colour channel). It would be better to use something that returns the actual ADU when measuring the sky background, rather than normalised values like PI gives. The whole-image histogram is not an ideal way to measure sky background if you have much vignetting. However, if you choose that approach and the field is fairly evenly illuminated, setting the sub length so that the left hand edge of the histogram is at about 3000 should give you a reasonable exposure.

Last edited by Shiraz; 11-10-2015 at 12:25 AM.
Reply With Quote
  #3  
Old 12-10-2015, 07:28 PM
Camelopardalis's Avatar
Camelopardalis (Dunk)
Drifting from the pole

Camelopardalis is offline
 
Join Date: Feb 2013
Location: Brisbane
Posts: 5,425
Thanks Ray, it was your thread that originally got me thinking and inspired me to ask more

I haven't yet found a way to analyse the raw pixel values in the raw files, but just going on PI's normalised values, but loading up one of the raws I took fresh this past weekend at a dark site and a bias file (and debayering both), the background values seem to hover around 0.033-4 whereas the bias file is 0.031...so it looks like exposing to about 25% just eeks it past the line. So if I understand it correctly it looks like I should expose for just a little longer from a dark site.

This is ignoring the dark current of course...which in the warmer evenings is noticeably higher than on the colder nights
Reply With Quote
  #4  
Old 12-10-2015, 08:07 PM
rustigsmed's Avatar
rustigsmed (Russell)
Registered User

rustigsmed is offline
 
Join Date: Mar 2012
Location: Mornington Peninsula, Australia
Posts: 3,950
Hi Dunk,

I see you have taken the scientific approach, from trial and error I can confirm that 30 minute Ha subs on an uncooled dslr is far tooooooo long

https://www.flickr.com/photos/803366...posted-public/


I wouldn't want to see the SNR or NSR on this one! And that is one of the chillier subs! 25 degs!

Cheers

Rusty
Reply With Quote
  #5  
Old 12-10-2015, 08:12 PM
Shiraz's Avatar
Shiraz (Ray)
Registered User

Shiraz is offline
 
Join Date: Apr 2010
Location: ardrossan south australia
Posts: 4,918
Quote:
Originally Posted by Camelopardalis View Post
Thanks Ray, it was your thread that originally got me thinking and inspired me to ask more

I haven't yet found a way to analyse the raw pixel values in the raw files, but just going on PI's normalised values, but loading up one of the raws I took fresh this past weekend at a dark site and a bias file (and debayering both), the background values seem to hover around 0.033-4 whereas the bias file is 0.031...so it looks like exposing to about 25% just eeks it past the line. So if I understand it correctly it looks like I should expose for just a little longer from a dark site.

This is ignoring the dark current of course...which in the warmer evenings is noticeably higher than on the colder nights
sounds like a plan to me Dunk - a bit more exposure may be useful provided the dark noise does not get in the way.

High dark current is not all bad - if you have a lot of noise, you can use shorter subs because the results will be so bad that it doesn't make any difference

Last edited by Shiraz; 12-10-2015 at 08:25 PM.
Reply With Quote
  #6  
Old 12-10-2015, 09:24 PM
Camelopardalis's Avatar
Camelopardalis (Dunk)
Drifting from the pole

Camelopardalis is offline
 
Join Date: Feb 2013
Location: Brisbane
Posts: 5,425
Quote:
Originally Posted by rustigsmed View Post
I see you have taken the scientific approach, from trial and error I can confirm that 30 minute Ha subs on an uncooled dslr is far tooooooo long

...

I wouldn't want to see the SNR or NSR on this one! And that is one of the chillier subs! 25 degs!
Eek yeah uncooled is a bit of a challenge but mine isn't too bad up to ambient around 10C. Much more than that and it's too noisy for my tastes now. I need to get cracking on the camera cooler mod

Quote:
Originally Posted by Shiraz View Post
sounds like a plan to me Dunk - a bit more exposure may be useful provided the dark noise does not get in the way.

High dark current is not all bad - if you have a lot of noise, you can use shorter subs because the results will be so bad that it doesn't make any difference
yeah I'd rather have some signal with my noise

I'm going to need to tackle the cooler situation as it's warming up now and I don't want summer to be a write off. That's one of the things that got me thinking and reading...I want to take subs as short as possible to balance the noise out with the number of subs.

Btw, any thoughts on dynamic range with DSLRs? From the way I figure, I'm losing one bit (using 14-bit raws) by bumping the ISO to 800, and another if I risk it at 1600. When will I start to notice the difference?
Reply With Quote
  #7  
Old 13-10-2015, 12:05 AM
glend (Glen)
Registered User

glend is offline
 
Join Date: Jun 2013
Location: Lake Macquarie
Posts: 7,033
I've been constantly shooting at ISO 800 with setpoint at 0C on warm nights and -5C on cooler nights. Both Rchesire and myself have compared 5 min darks shot at those temps and they are identical to bias/offset frames at 1/4000th of a sec. Why ISO 800, because Craig Stark did some testing and recommended it.

I have no problem imaging at f5 with the newt or the mak-newt, but at f8 with the RC it seems to take forever to get the same luminance, and that highlights all the little guiding issues I never worried about at f5.
Reply With Quote
  #8  
Old 13-10-2015, 01:42 PM
Camelopardalis's Avatar
Camelopardalis (Dunk)
Drifting from the pole

Camelopardalis is offline
 
Join Date: Feb 2013
Location: Brisbane
Posts: 5,425
Glen, you might want to try ISO 1600...the read noise is almost halved on the 450D, and since you're cooling your sensor anyhow the thermal noise should be well under control. That claws back much of the intensity you lose from f/5 to f/8.

I was recently lured into trying 800 with my 1100D, but the noise at 1600 is only ~30% higher for the same exposure time, so much of the doubling of signal comes for free. My main issues at the moment are thermal, as the weather is warming up. It's easy enough to use employ an intermission between subs of 60 seconds or so to give the camera some breathing time, and still accumulate data quicker than at ISO 800.

Of course, there's the question of dynamic range...which I don't fully understand, but I'm not convinced either way that it makes a difference from my usual imaging haunts.
Reply With Quote
  #9  
Old 13-10-2015, 02:38 PM
glend (Glen)
Registered User

glend is offline
 
Join Date: Jun 2013
Location: Lake Macquarie
Posts: 7,033
Thanks Dunk, that's a good idea, I will try ISO1600 running as cold as I can. I don't need a cool down interval for the sensor running at setpoint of -5C. As to night time ambient and summer seeing wobbles, I try to delay imaging until after midnight, to allow the earth to output retained heat and the atmo to stabilise. It helps a littlle. Not an issue in winter but then dew and optics heating is more an issue.
Reply With Quote
  #10  
Old 13-10-2015, 05:21 PM
Camelopardalis's Avatar
Camelopardalis (Dunk)
Drifting from the pole

Camelopardalis is offline
 
Join Date: Feb 2013
Location: Brisbane
Posts: 5,425
Yeah you should get better results Glen, as mine needs that breather in between subs to keep the temperature from escalating, or cooler ambient temps
Reply With Quote
  #11  
Old 13-10-2015, 06:09 PM
John K's Avatar
John K
Registered User

John K is offline
 
Join Date: Sep 2005
Location: Melbourne
Posts: 1,468
+1 for ISO1600.

In my experience with a 400D, somewhere between 2-5 mins depending on temperature if un-cooled.

If your DSLR is cooled then up to 10 minutes.

John K.
Reply With Quote
Reply

Bookmarks

Thread Tools
Rate This Thread
Rate This Thread:

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +10. The time is now 07:07 PM.

Powered by vBulletin Version 3.8.7 | Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Advertisement
Bintel
Advertisement
Testar
Advertisement