#1  
Old 19-02-2008, 01:16 PM
Peter Ward's Avatar
Peter Ward
Galaxy hitchhiking guide

Peter Ward is offline
 
Join Date: Dec 2007
Location: The Shire
Posts: 8,090
QE & S/N comparasion

I did and A/B comparison with my Canon 5D and STL11k, As they say a picture can say a thousand words, and the pictures are here

http://www.atscope.com.au/BRO/tutorials/dslr.html

If the Canon was exposed for 90 seconds, i.e. the time it took to get the RG&B channels out of a Mono camera, then it would have performed better, but that was not the point of the exercise.

I was curious to see how the two cameras would deal with the same amount of (limited) flux, and was admittedly surprised by the obvious differences in signal and noise between the two.

Last edited by Peter Ward; 19-02-2008 at 03:36 PM. Reason: typo
Reply With Quote
  #2  
Old 19-02-2008, 01:39 PM
iceman's Avatar
iceman (Mike)
Sir Post a Lot!

iceman is offline
 
Join Date: Sep 2004
Location: Gosford, NSW, Australia
Posts: 36,760
Wow, very interesting indeed.

Proof is in the pudding!
Reply With Quote
  #3  
Old 19-02-2008, 03:29 PM
tornado33
Registered User

tornado33 is offline
 
Join Date: Jan 2005
Location: Newcastle, NSW, Australia
Posts: 4,106
Ahh yes but its not a modded camera Indeed I can see the nebula looks bluish with the 5D's infra red cut filter in place

Here is a 125 second ISO1600 image, with Houghys cooled modded 350D DSLR ,also had a UHCS filter in place. This was a test image taken to check focus and composition, and was done as a jpeg and not modified in any way, this is straight off the camera. 125 seconds with 10 inch f5.6 newtonian.
Attached Thumbnails
Click for full-size image (etacarina 120sec iso1600uhcs.jpg)
192.2 KB85 views
Reply With Quote
  #4  
Old 19-02-2008, 03:36 PM
Peter Ward's Avatar
Peter Ward
Galaxy hitchhiking guide

Peter Ward is offline
 
Join Date: Dec 2007
Location: The Shire
Posts: 8,090
Seeing as I don't own a modified camera I couldn't run that test....however 125 secs is 4x times longer than my test exposure...

Do you have data at 2800mm (F8) from a 30 second exposure?


Quote:
Originally Posted by tornado33 View Post
Here is a 125 second ISO1600 image, with Houghys cooled modded 350D DSLR ,also had a UHCS filter in place. This was a test image taken to check focus and composition, and was done as a jpeg and not modified in any way, this is straight off the camera. 125 seconds with 10 inch f5.6 newtonian.

Last edited by Peter Ward; 19-02-2008 at 03:44 PM. Reason: clarification
Reply With Quote
  #5  
Old 19-02-2008, 04:14 PM
Terry B's Avatar
Terry B
Country living & viewing

Terry B is offline
 
Join Date: Mar 2006
Location: Armidale
Posts: 2,789
Quote:
Originally Posted by Peter Ward View Post

If the Canon was exposed for 90 seconds, i.e. the time it took to get the RG&B channels out of a Mono camera, then it would have performed better, but that was not the point of the exercise.
Isn't that exactly the point?
You have 3 x more data in the RGB image so it will have much less noise.
Reply With Quote
  #6  
Old 19-02-2008, 04:29 PM
Peter Ward's Avatar
Peter Ward
Galaxy hitchhiking guide

Peter Ward is offline
 
Join Date: Dec 2007
Location: The Shire
Posts: 8,090
Quote:
Originally Posted by Terry B View Post
Isn't that exactly the point?
You have 3 x more data in the RGB image so it will have much less noise.
Que?

Clearly you missed the point, as there is *not* 3x the data.

Each filtered pixel got the same 30 seconds worth of red, green or blue light.

The downside with a mono camera is you have to take 3 successive exposures, but a single shot colour does all three channels at the same time.

The mono camera has higher spatial resolution, in this case 33 mega-pixels if you count pixels in the same way as in colour cameras.

As to the point, well there are several.
CCD's
Higher QE, lower noise, less noise structure, better colour saturation, higher dynamic range.
CMOS.
Low H-alpha sensitivity unless modified, Easier use, requires external guider.

Last edited by Peter Ward; 19-02-2008 at 04:30 PM. Reason: typo
Reply With Quote
  #7  
Old 19-02-2008, 04:48 PM
avandonk's Avatar
avandonk
avandonk

avandonk is offline
 
Join Date: Aug 2005
Location: Melbourne
Posts: 4,786
It is a meaningless 'test' as the Sbig has nine times the usable photons. The Canon only uses a third of its pixels for each colour.

Then test them at the same temperature!

Sorry try again by maybe illuminating the Sbig without any filtration for thirty seconds and the Canon for 30 secs through red, green and blue filters and converting the three images to luminance. Then the test would be more equitable. The IR filter still hampers the unmodded Canon.

Bert
Reply With Quote
  #8  
Old 19-02-2008, 05:12 PM
Peter Ward's Avatar
Peter Ward
Galaxy hitchhiking guide

Peter Ward is offline
 
Join Date: Dec 2007
Location: The Shire
Posts: 8,090
So, you are saying we effectively knobble the Canon by introducing a ND filter factor ahead of the Bayer (G_R_B_G) Matrix?

There is considerable interpolation during the debayering and it would be very rare to find an object that emits only one of the primary colours.

The off colour pixels do contribute to the image so I can't see how you get "9x the usable photons" from the CCD....if that were the case, then should have been little difference between the two cameras with broad band emitters (stars)....and it is clear that the CCD is still more sensitive there.

I also wanted to use each camera "as is". Turning off the peltier on the CCD would make as much sense as putting an ND filter on the Canon


Quote:
Originally Posted by avandonk View Post
It is a meaningless 'test' as the Sbig has nine times the usable photons. The Canon only uses a third of its pixels for each colour.

Then test them at the same temperature!

Sorry try again by maybe illuminating the Sbig without any filtration for thirty seconds and the Canon for 30 secs through red, green and blue filters and converting the three images to luminance. Then the test would be more equitable. The IR filter still hampers the unmodded Canon.

Bert
Reply With Quote
  #9  
Old 19-02-2008, 05:15 PM
skwinty's Avatar
skwinty (Steve)
E pur si muove

skwinty is offline
 
Join Date: Dec 2007
Location: Cape Town
Posts: 494
Hi All
IMO the tests are subjective to say the least.
For one, the QE is not the same across the spectrum being measured as well as the QE of CCD's and CMOS chips are very different.
to compare CCD and CMOS is compared oranges and apples.
I dont believe that CMOS has reached the performance of CCD's in astrophotography just yet although the gap is narrowing.
Regards
Steve
Reply With Quote
  #10  
Old 19-02-2008, 05:47 PM
Peter Ward's Avatar
Peter Ward
Galaxy hitchhiking guide

Peter Ward is offline
 
Join Date: Dec 2007
Location: The Shire
Posts: 8,090
Subjective in what way?

The telescope, mount and exposure times were identical.

Each instrument was unmodified, and used in a manner that would optimize the shot noise for each camera at 30 seconds.

I have no argument with one camera being optimized in its design to be used for deep sky imaging, and the other for daylight photography....and make no apologies for that.

The roll-over simply highlights the glaring differences in performance in
identical conditions.

If the weather holds, I'll try the same again, but with a back illuminated CCD this time

Quote:
Originally Posted by skwinty View Post
Hi All
IMO the tests are subjective to say the least.
Reply With Quote
  #11  
Old 19-02-2008, 05:59 PM
skwinty's Avatar
skwinty (Steve)
E pur si muove

skwinty is offline
 
Join Date: Dec 2007
Location: Cape Town
Posts: 494
Well subjective in the sense that the result is open to visual interpretation and most would report differently if questioned about specific aspects of the results.You say that the test was optimised for shot noise but you dont specify for example what was the gain setting for the CCD as opposed to the ISO of the DSLR.
The quantum efficiencies are different between the two sensors and the qe is not linear across the spectrum and to achieve comparable results surely you would have to increase the exposure time for the DSLR.
The CCD will surely outperform the CMOS over the same integration time.
Regards
Steve
Reply With Quote
  #12  
Old 19-02-2008, 06:19 PM
Zuts
Registered User

Zuts is offline
 
Join Date: Mar 2007
Location: sydney
Posts: 1,830
Hi Peter,

If you are trying to tell us that an 11,000 AUD camera cooled to -15 degrees C takes a better picture than a 1,500 AUD DSLR then I for one have no argument with you.

However when people compare say CPU's they do it on a price/performance or bang for buck ratio. Using this criteria they may say that a 2ghz CPU costing 50 AUD is better value for money than a 3ghz CPU worth 800 AUD.

To take the picture using the the STL took 90 seconds, the better example from the posts was the 125 second exposure as this is closer to the EFFORT required to take the photo.

Effort counts here. The BANG here is the quality of the final picture and how much BUCK (effort and money) one is prepared to pay.

I dont think anyone here would disagree that with a bit more EFFORT in the form of stacked exposures it is possible to get photo's approaching that of a CCD. If the gap is closed by modding and cooling the DSLR camera then the EFFORT is reduced.

I think that on a BANG for BUCK comparison a DSLR is probably better value than a 11,000 AUD SBIG.

This is only my opinion and of course other people are free to make there own conclusions.

Paul
Reply With Quote
  #13  
Old 19-02-2008, 06:45 PM
Kal's Avatar
Kal (Andrew)
1¼" ñì®våñá

Kal is offline
 
Join Date: Nov 2006
Location: Sydney
Posts: 1,845
I agree with Zuts on this. People know that CCD cameras perform better than digital SLR's. I can grab a $1000 intel CPU and show it performs better than a $150 AMD one - but people like me buy the AMD one because they know that from a price to performance ratio you get better value for money.

It is the same with DSLR Vs dedicated astro CCD cameras. The market is there, because the value for money is there.
Reply With Quote
  #14  
Old 19-02-2008, 06:53 PM
Peter Ward's Avatar
Peter Ward
Galaxy hitchhiking guide

Peter Ward is offline
 
Join Date: Dec 2007
Location: The Shire
Posts: 8,090
Gee, Paul your numbers are a bit out Try about $A8800 (STL) and $A3400 (5D body...where can you buy a 5D for $1500 ?? I'll buy two )

But, yes, I agree, the STL11k will cost a whole lot more than a 5D, but that wasn't the point....I could do the same test with a ST2000xm and despite less acreage in the ST, the QE/Noise/Sensitivity would have been similar to the STL at about half the price.

I suppose if there was a point to my comparing the two, was to discover: QE really matters.

For example, with a Marconi Back illuminated chip you loose just 5 or 6 H-Alpha photons out of a hundred, with a bog standard DSLR you loose about 85 of them.

It's pretty brain dead math, with all other aspects being equal, to see one detector will have a serious advantage over another.

Cheers
Peter


Quote:
Originally Posted by Zuts View Post
Hi Peter,

If you are trying to tell us that an 11,000 AUD camera cooled to -15 degrees C takes a better picture than a 1,500 AUD DSLR then I for one have no argument with you.

l
Reply With Quote
  #15  
Old 19-02-2008, 07:09 PM
skwinty's Avatar
skwinty (Steve)
E pur si muove

skwinty is offline
 
Join Date: Dec 2007
Location: Cape Town
Posts: 494
hi Peter
Your test was indeed interesting if not imo subjective.(just refer to the differences between the 2 cameras)
However its back to the Hyundai / BMW argument i suppose.
I would be interested to see the results of the back illuminated CCD test and any others you may care to perform and share with us.
It is a pity you do not have a modded 40d to work with as that camera has some further advantages over the 5d such as a better dynamic range and more sensitivity to Ha.
I would have liked the 40d to have a 16bit ADC but i suppose i must be happy with 14 vs 12 bit ADC and a Digi 111 processor vs Digi 11.
Regards
Steve
Reply With Quote
  #16  
Old 19-02-2008, 07:11 PM
Zuts
Registered User

Zuts is offline
 
Join Date: Mar 2007
Location: sydney
Posts: 1,830
Hi Peter,

Canon 5D, 2,700 AUD

http://www.dirtcheapcameras.com.au/s...ory=20&brand=2

ST11K I was assuming a class1 sensor 8,500 USD + GST + PPH. Maybe if you could sell me one of these for 8,800 AUD i might get one

http://www.optcorp.com/ProductList.a...9-320-324-1008

Paul
Reply With Quote
  #17  
Old 19-02-2008, 07:18 PM
Kal's Avatar
Kal (Andrew)
1¼" ñì®våñá

Kal is offline
 
Join Date: Nov 2006
Location: Sydney
Posts: 1,845
I think a good test would be a one shot color CCD camera against a modified DSLR. Both of these are in the hands of astronomers, and both have produced great results.

An STL-11000CM color Vs a modified 5D will be an interesting comparison.
Reply With Quote
  #18  
Old 19-02-2008, 08:00 PM
Peter Ward's Avatar
Peter Ward
Galaxy hitchhiking guide

Peter Ward is offline
 
Join Date: Dec 2007
Location: The Shire
Posts: 8,090
There are limits on how much gear I can get my hands on, but, the suggestion that a single shot colour CCD/DSLR comparison be done is an interesting one which I hope to follow up.

As for subjectivity vs objectivity. I must be losing something in the translation.

Objectivity pertains to facts, things you can easily measure. Good old scientific method.
eg. SLK McLaren Benz will do 0-100pkh in 3.8 secs.
Hyuandai won't. So what? Nothing new or surprising here.

Subjectivity pertains to how you feel about something or an opinion. Not easily measured.

So, getting back to the original post. Yep, it is pretty obvious given the same flux, a dedicated medium QE CCD astrocam shows higher dynamic range, better QE, lower noise etc than on off the shelf Premium DSLR

Everything else (cost, possible modifications, your shoe size) doesn't change the result.

Now, if I can just find that U47 power pack....

Cheers
Peter

Last edited by Peter Ward; 19-02-2008 at 08:07 PM. Reason: typo!
Reply With Quote
  #19  
Old 19-02-2008, 08:38 PM
CometGuy's Avatar
CometGuy
Registered User

CometGuy is offline
 
Join Date: Feb 2005
Location: Brisbane
Posts: 942
Peter,

Interesting comparison - no suprises really.

Since were looking at primarily a H-Alpha object I thought I'd throw in some Quantum efficiency theory The data I have on an unmodified 5D puts QE at less than 5% at H-Alpha, but then you have to remember only 1/4 of the pixels are sensitive to H-Alpha. With the red filter in place the STL11000 problably has 30% QE at H-Alpha but this is distributed across all its pixels. So we are looking at least 20x more H-Alpha sensitivity for same exposures in favour of the SBIG!

If we modify the 5D and increase the exposure to 90 seconds (as a number people are suggesting), we get a bit closer maybe to around half the sensitivity of the STL11000 in H-Alpha. Of course there are better CCD's for H-Alpha detection than the STL11000.

Peter, what I'd also be interested in though is seeing is a same exposure unfiltered STL11000 exposure versus a 5D with the red/green/blue channels combined in a Black and White image. I'd be interested in seeing difference in limiting magnitude of stars - have you got something along those lines?

Terry
Reply With Quote
  #20  
Old 19-02-2008, 08:59 PM
Peter Ward's Avatar
Peter Ward
Galaxy hitchhiking guide

Peter Ward is offline
 
Join Date: Dec 2007
Location: The Shire
Posts: 8,090
Terry, interesting analysis, but I don't think it's quite so bad for the Canon for reasons I stated earlier.

The distributive property of multiplication over addition tells me summing colour channels will not change the CMOS result (just change it to a grey scale image in photoshop to see what it would look like)....but not filtering the CCD will drop a ND filter factor of at least 0.4, making it look brighter still..... Now *that* would have deserved me some flack!

Quote:
Originally Posted by CometGuy View Post

Peter, what I'd also be interested in though is seeing is a same exposure unfiltered STL11000 exposure versus a 5D with the red/green/blue channels combined in a Black and White image. I'd be interested in seeing difference in limiting magnitude of stars - have you got something along those lines?

Terry
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 01:17 AM.

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