#21  
Old 25-04-2006, 09:05 PM
CometGuy's Avatar
CometGuy
Registered User

CometGuy is offline
 
Join Date: Feb 2005
Location: Brisbane
Posts: 942
My comparision is sort of valid since despite 6x light collecting advantage (the effective aperture is 145mm due to the large secondary), the exposure is 1/6th as long. Given that the sky will be darker with the 300D image due to the shorter exposure, you would then expect it to see slightly fainter stars if the camera performance is similar. This is what you actually see.

As far as extended detail is concerned, your optical system is only a stop slower so there is a factor of 3 against the 300D image. Hence the 300D image looks a lot noiser.

I know this is not precise and doesn't take into account a number of other factors, but it is a ballpark indication.

I disagree re 3 x 3 not being equivalent to a single 9 minute (it will be >90% as good provided it has been aligned correctly).

I am suprised with your observation re the SXV-H9, but then the newest DLSRs seem to have very uniform sensitivity and don't seem to require the same level of calibration many astroccds need. Perhaps this is the reason...

Terry
Reply With Quote
  #22  
Old 26-04-2006, 03:01 AM
Gama's Avatar
Gama
Registered User

Gama is offline
 
Join Date: Dec 2005
Location: Melbourne
Posts: 1,121
Well, i'd say 6 times the light grasp is pretty big. With 1 extra stop making a world of difference.
Its not the noise thats making the 300D less superior to the D200, but more its sensitivity. The 300D does not have a very good QE, thus it takes a little longer to accumulate a signal. The D200 has a higher Q.E and in turn is able to produce a image faster in the same given time. Of course the Hot filter will effect the H.a and I.R color signal compared to the 300D.
As for stacking, there is a difference, a 9 minute exposure is different to 3x3. You will as i said get the detail come thru above the noise threshold and a clearer picture. But it will not compensate for going deep in magnitude as long as the sky is dark and tracking is good.
Theoretical and practical are two different things. I agree it comes close, and produces a less noisy picture, but its not the real Mc Coy.
The SXV is a very sensitive camera, but as you said, the newer DSLR cameras are really starting to show their strengths. After i do a LRGB image using the SXV, it just doesnt have that smooth texture you see in books and expensive Astro cameras. This is because of the difference in the pixel count.
A 10 Mpixel image compared to a 1.4 Mpixel is a huge huge difference to image quality. Then binning it will create a very nice image in the end.
Reply With Quote
  #23  
Old 27-04-2006, 06:35 PM
Gama's Avatar
Gama
Registered User

Gama is offline
 
Join Date: Dec 2005
Location: Melbourne
Posts: 1,121
OK, i have taken a few images of M83 with the D200 and NR (Noise Reduction) OFF, as i said i want to see the total noise this camera generates including amp glow etc at room temperature.
I took 8 X 2 minute exposures with a room temperature D200 at prime focus of my 14" RCX @f8, i then stacked them (Median) and then done Strech, Unsharp and then a 3x3 BIN to shrink the size down a lot... and then a little crop to get rid of a boo boo i did when i bumbed the scope half way into the exposures... I also over streched the image to show the amp glow better, so its not as bad as it looks.
Any how heres the image. I like it, very good color for 16 minutes of exposure @f8. I also have attached an equivalent from the Canon 20D which i also done the same way except the exposures were done to 28 minutes at ISO400 to reduce noise. The color inbalance in the Canon was due to MaximDL not setting the correct balance.
Remember i over enhanced the Nikon to show the glow, but in reality its not as bad as it here, and remember this is when the camera is at room temp (21 deg). I later took a dark frame and subtracted the noise and the image was fine.
Attached Thumbnails
Click for full-size image (M83 Nikon Raw RCXF8.jpg)
129.6 KB214 views
Click for full-size image (M83 Raw canon RCX F8.jpg)
130.4 KB185 views
Reply With Quote
  #24  
Old 27-04-2006, 08:52 PM
CometGuy's Avatar
CometGuy
Registered User

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

Very impressive! I like your D200 version not only because of the colour balance but because it looks more natural. But I notice that the 20D has been processed with an unsharp mask or some other filter?

Anyway, woulld you be able to post crops around the galaxy?

Terry
Reply With Quote
  #25  
Old 28-04-2006, 12:01 AM
Gama's Avatar
Gama
Registered User

Gama is offline
 
Join Date: Dec 2005
Location: Melbourne
Posts: 1,121
Correct, i used unsharp, as i said i processed both pics the same way so not to imbalance the 2 pictures when enhancing, but i did over stretch the D200 so as to show the actual spots where the glow is found
Here is the crop of the full size, looks crappy when its compressed this much.
I'll do a 1 hour exposure soon on this object in the next few days.
Attached Thumbnails
Click for full-size image (M83 Nikon Raw RCX F8 crop .jpg)
134.8 KB150 views
Reply With Quote
  #26  
Old 02-05-2006, 02:04 AM
Gama's Avatar
Gama
Registered User

Gama is offline
 
Join Date: Dec 2005
Location: Melbourne
Posts: 1,121
OK, final test i done was a Noise test with the lens cap on.
The camera is going back to Nikon tomorow, and im very sorry to see it leave, with the crappy weather we had the last 4 days i couldnt do 1 more astro image, oh poo.
So i did a noise test with both cameras at 23 deg room temp (I really give them hell with temp). The results speak for themselves, the D200 really comes out very strong with these tests. I did not do any screen streching or enhancing. Its RAW and just saved and uploaded.. Thats it.
Another drawback for the Nikon, is you cant switch off the power while its doing the NR capture, cause all it does is stop the NR image and switch off. It does save the file but removes an equivalent of NR of time that you let it before you switched it off. By this i mean if you wait 10 seconds before switching off the camera, then it only removes 10 seconds of equivalent NR.

Anyway, i'll submit my complete analysis of this camera to Nikon Australia, and hopefully with their permission i could release it to everyone around the globe.
Having said that, i am NOT employed by Nikon or any of its agents nor related to any emploees to the said company. I just beleived that this camera should (And did) have a huge huge impact in Astrophotography.
Problem is the misconceptions out there that Nikon are inferior to Canon.
Well, wake up and smell the coffee, its not. As Comet Guy said and i totally agree, the cost of the camera and its ability was unfair to compare it too a 20D. But hey, thats all i had, and so do many others, plus its only a few hundred dollars more than the 20D and over $1500+ less than the 5D. Not that i dont like the Canon 5D, to which im still confidant the D200 will also be a better Astrophoto camera over it (Up to ISO 1600), just get rid of the Hot Filter.
If Canon wants to send me a 5D for testing, i'll give it the same fair go and also display the results.. Hmmm, whats their telephone number again...

Too much Tech Talk, time for Cave man Talk.
Ugg, 20D Good, D200 much Gooda Ugg.

Me Get one, Ugg.
Attached Thumbnails
Click for full-size image (Canon Bin 3x3 316 sec NR off.JPG)
91.2 KB80 views
Click for full-size image (D200 Bin 3x3 360 sec NR Off.JPG)
106.2 KB77 views
Click for full-size image (Canon Bin 3x3 319 sec NR on.JPG)
106.4 KB72 views
Click for full-size image (D200 Bin 3x3 360 sec NR On.JPG)
90.0 KB92 views
Reply With Quote
  #27  
Old 02-05-2006, 05:50 AM
iceman's Avatar
iceman (Mike)
Sir Post a Lot!

iceman is offline
 
Join Date: Sep 2004
Location: Gosford, NSW, Australia
Posts: 36,760
That's a significant difference in noise there!

I look forward to reading your full analysis, if it's able to be released. Thanks for the ongoing commentary, it's been really interesting.
Reply With Quote
  #28  
Old 02-05-2006, 08:30 AM
Gama's Avatar
Gama
Registered User

Gama is offline
 
Join Date: Dec 2005
Location: Melbourne
Posts: 1,121
It is a big difference, but thats what your paying for with the additional costs between the cameras. This was one of my concerns with the Cmos sensor over the CCD sensor. I knew that the CCD should perform better here even though the Cmos has much less heat to vent, but if its designed correctly and heat sunk right, then heat build up on the surface of the sensor wont be as bad.
Further, the CCD sensor design is quite well thought off, it has so many on board functions that dont seem to draw much current compared to the older versions.
But no matter what the hype, the proof is in the pudding !.
Reply With Quote
  #29  
Old 02-05-2006, 10:22 AM
Jonathan
Registered User

Jonathan is offline
 
Join Date: Jan 2005
Posts: 442
Great report Gama, thanks for all the info. The results are very encouraging and my next camera is sure to be a D200 now. (If only Nikon could supply enough of them to meet demand )
Reply With Quote
  #30  
Old 02-05-2006, 07:43 PM
CometGuy's Avatar
CometGuy
Registered User

CometGuy is offline
 
Join Date: Feb 2005
Location: Brisbane
Posts: 942
The reason the D200 dark frame looks so good is:

1. Its well known that Nikon apply a median-type filter to long exposure images.
2. Nikon do not apply a brightness offset to raw images (unlike Canon), this means the dark frame will look much darker with the Nikon.

Both these actions damage fainter detail in an astronomical image.

As a comparision I have attached a dark frames from my 350D, 1020secs at ISO1600 (at 24c) process 2 ways. Version 1 is a straight raw, but with the image offset is brought back to 0 to match the Nikon.

The second image has both 1) 0 offset and 2) a median filter comparable to the built in filter of the Nikon. I have then increased the contrast 10x (i.e push processed the image 2.5 stops in film speak) otherwise it would appear black.

Sorry I disagree, it is not hype that Canon have made a great achievement with their sensors. The noise levels and quantum efficiency are as good as any CCD sensor (its not fair to compare specialised monochrome sensors which have very high Quantum efficiencies but are too expensive/impractical to be used in a digicam).

As a last comment compare this 15 minute exposure of M83 made by Bill Christie using a 350D and an 8" f4 telescope to your image with a D200 and 14" f8 telecscope:

http://www.zodiaclight.com/galleria/...rnPinwheel.htm

Consider, this was done with 30 second individual frames...

Terry
Attached Thumbnails
Click for full-size image (darkfs_1.jpg)
62.1 KB75 views
Click for full-size image (darkfs_3.jpg)
56.3 KB65 views
Reply With Quote
  #31  
Old 02-05-2006, 09:08 PM
Gama's Avatar
Gama
Registered User

Gama is offline
 
Join Date: Dec 2005
Location: Melbourne
Posts: 1,121
Firstly, your comment about median filter insertion for the D200 isnt true. It is a Raw output. You are refering to the D70, D50 etc. You can disable it with the D200.
Secondly, if you look at the Histograms for both you will see that the 20D has substantial more signal in noise that fills the entire grid, where as the D200 has very little signal level and only occupies a tiny portion of the Histogram grid. Thus i said that i did not do any stretching what so ever, so the Canon due to its higher noise signal falls over the set points and thus shows up more red. If i adjust the set point for the Histogram levels then im altering the cameras initial signal levels it determines and saves. This is what you did, you modified the histogram to suite your noise levels. It would be good to see the actual histogram for such a long exposure..
Histograms contain your data and noise information, so the less Noise signal you have to start with the better your signal will be and also handle enhancement processing.
Check the images i posted for the 2.
I did this test because it shows just how much noise is actually effecting a real signal comming in. The less signal (Noise) you have here is obviously desirable.
Your exposure of 1020 secs has built up a large noise level. A proper image shows exposure shifts of the histogram to the right more, where as a less signal, the histogram is shifted to the left.
This backround will obviously not show up on your image because you modified the levels to suit your exposure. If there was a signal with that noise then the signal would also be effected in the same way.
The 20D failed, im sorry you just cant accept that. You always trying to defend Canon with conspirecy theorys and counter reasons to everything i test and display here.
No conspirecy theory, no bias to Nikon, as i said i own and use the 20D for Astrophotography. But call a spade a spade. Nikon have brought out a good camera for us Astrogeeks, all be it with a cruddy CCD filter.
As for the Cmos Vs CCD, i was not comparing the monochrome CCD version. In fact i compared it to the SXV-M25 which has the same chip as the D70. This CCD chip has a much higher QE than the Canon Cmos sensor. Again, no tricks, no smoke and mirrors.
The image of the M83 with the 8" f4 is exactly that, f4. What are you trying to say ?, nothing to do with the camera, cause its f4 to f8, makes no bananas that mines a 14" when his is an f4 astrograph. Now if i placed my f.5 reducer to bring my 14" to f4 as well, then what do you think the outcome will be when i place any camera on....
Terry, take my hand and come over to the DARK SIDE .hehe Darth Gamma

8 out of 10 beer drinkers cant be wrong !.
Attached Thumbnails
Click for full-size image (Canon 20D Histogram.jpg)
146.5 KB63 views
Click for full-size image (Nikon D200 Histogram.jpg)
144.8 KB72 views

Last edited by Gama; 02-05-2006 at 09:36 PM.
Reply With Quote
  #32  
Old 03-05-2006, 01:04 AM
avandonk's Avatar
avandonk
avandonk

avandonk is offline
 
Join Date: Aug 2005
Location: Melbourne
Posts: 4,786
Do you want to see a 20D compared to a 5DH? There is not even a race. I own both. I don't know how anything compares to a Nikon. I still have an open mind. Nikon though is going to stick to the C sized sensor and unfortunately this is not big enough if you want to do widefields. The alternative is to get the full frame Sbig (if you want a CCD) and from what I have have seen the 5DH holds it own quite well with even this camera at the same temperature!
Don't even mention the price differential. When a Nikon can do this with a 100ED I will sit up and take notice.

http://members.optusnet.com.au/~avan...GTR_CFA_ON.jpg

Where is the noise?

Bert

Last edited by avandonk; 03-05-2006 at 01:33 AM.
Reply With Quote
  #33  
Old 03-05-2006, 05:34 PM
Gama's Avatar
Gama
Registered User

Gama is offline
 
Join Date: Dec 2005
Location: Melbourne
Posts: 1,121
Well, you dont know anything about Nikon ?, Im glad to point you to look at what a D70 can do. 5DH images look good, the D70 look as good if not better ..Oh i opened a can of worms !!.
http://www.pbase.com/terrylovejoy/erwins_images .
Again i use the late Erwins images, and very sad he's no longer with us. Because this man had a great passion for his hobby and mainly used a D70. His images have made it to proffesional levels with some of his images posted thru out the proffesional Astronomy communities and papers.
In any case these images are with a MODIFIED Nikon D70. The same sensor as the SXV-M25.
So Bert, stand up and take notice.

But seriously, it doesnt make a hoot in hell what camera does what. In the end its what we can afford and end up using time and time again.
Sure, its good to post a modified version, but then compare it to another modified version.. 5DH is drool material, but wait till i rip the filter of my forthcomming D200, i'll then do another comparison against itself.
Im more than confidant it should do better than the 5DH, but im not making any bets, yet...
Its unfortunate i dont have a 5D or a 5DH to test, i'd really like to see the specs on the Cmos chip inside the 5D, especially its Q.E., and noise level.

Either way, this post was not about Holden Vs Ford, and so i'd like to keep it like that. Cause then we'll have someone with the 6303 or a even better a back illuminated sensor camera throwing their 2 cents in too.
Most good DSLR can take excellent images, as long as you give it enough exposure and software enhance the end result.
Actually could you please do a 330 sec dark (With lens cap on) and take one with NR off and one with NR on and post the images as well as the histogram (Before editing, RAW). I'd like to see the results..
Reply With Quote
  #34  
Old 03-05-2006, 07:40 PM
avandonk's Avatar
avandonk
avandonk

avandonk is offline
 
Join Date: Aug 2005
Location: Melbourne
Posts: 4,786
This may be of interest Gama also check out the rest of his site

http://clarkvision.com/imagedetail/d...gnal.to.noise/

Bert
Reply With Quote
  #35  
Old 03-05-2006, 08:56 PM
CometGuy's Avatar
CometGuy
Registered User

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

In your original post/s you didn't show much tact considering many of us own Canon gear, so are you really suprised by the response? But the questions and challenges coming from me are going to be the same ones you will face on other forums. I know your comment that the Nikon D50 is too noisy will get challenged as well (IMO it could be the true DSLR champion, especially if Duncan Munroe gets his 'mode 3' automation device marketed ).

The ONLY serious testing I know comparing the Canon CMOS and Sony CCD Quantum Efficiency was done by Christian Buil. He found little difference in quantum efficiency between the Canon and Nikon Sensors (if you like I'll point out the specific parts where this is evaluated, but assume you have already done that). He also has determined the transmission curves for both the Nikon and Canon hotfilters, and the outcome is still the same if removed.

The right offset you see in the 20D histogram is not dark signal, it is a numeric offset. Try doing a series of different exposures and you will see that it remains almost constant. Doing what I did and removing this offset is absolutely valid for the comparision. The D200 histograms does look impressive, but again is the filtering at work? A small crop of the dark frame, contrast boosted to clearly show the texture would give au some idea.

At the end of the day I am not interested in a Nikon DSLR until they bring out a camera that allows true raw mode

Terry
Reply With Quote
  #36  
Old 04-05-2006, 12:23 AM
Gama's Avatar
Gama
Registered User

Gama is offline
 
Join Date: Dec 2005
Location: Melbourne
Posts: 1,121
First, to Terry, in regards to Q.E, how one person tests Q.E (Including me) and what the Q.E specs from the manufacturer are different, unless they are tested correctly. The Sony CCD line have much better Q.E than Canons Cmos, i cannot be definite with the 5D or up models as i and any other person dont have any specs to their Q.E's.

Second, what does the Hot filter have to do with a Dark frame ?, cause you have the lid ON taking no light...I guess you just werent thinking at the time..
Third, Christian Buil's Q.E results are "Relative" values, and NOT absolute, even though the D70 is nearly 3 times higher in Q.E in his report, its not a true indication of the Cmos sensors actual range.
Fourth, i have taken a dark with NR ON for 30 secs and you can see its not as right shifted, in turn the image is darker. I have also attached a copy of the Image from the Starlight Xpress homepage that did the same test as me, except longer exposure and i didnt strech, comparing the KAI2000 CCD over the Sony ICX285 (SXV-H9) for Noise. Same Test, without streching !

Now for Bert, ive read it Bert, mainly headache reading, but really not too helpfull here. Some of the data is assumed (Indicated by ?), which is why i didnt take too much on board with it, and some specs are out, the Full well capacity for the D70 is actually a little worse than documented, and some others estimated.

Secondly, THE D200 AINT IN THERE, hehe, a shame really, he could have done most of the work for me. Still, you really need to test it yourselves.

But Bert, i really would like to see some histograms and pics please. I would like too see just how well it does actually perform. Remember to stay under 1600 ISO, as over this the 5D is leathal.
I came so close to buying one, in fact i bidded on it on ebay, but lost the bid. Then i read about the D200 which is over $1500 cheaper, so i thought i'll test it before i waste my money on a camera no ones tested or poduced any images worth posting.
Thus my position here.
Attached Thumbnails
Click for full-size image (canon 30sec dark.jpg)
142.3 KB54 views
Click for full-size image (Darkframecomparison.jpg)
121.6 KB80 views

Last edited by Gama; 04-05-2006 at 01:51 AM.
Reply With Quote
  #37  
Old 04-05-2006, 12:27 AM
Gama's Avatar
Gama
Registered User

Gama is offline
 
Join Date: Dec 2005
Location: Melbourne
Posts: 1,121
Oh yeah, Terry, it really doesnt matter how much info or data you give some one, cause like you said there are people out their that just wont accept it, being a die hard Canon guys, or Nikon guys, or Pentax guys etc.. Im not really biased, all i did is do some tests side by side the same way, with no hidden techniques. Thats all !.
I took images the same way i did with my SXV-H9 and Canon 20D and prefered the D200.


And HOLDENS RULE BABY !.

mumble mumble but i drive a Ford mumble mumble..
Reply With Quote
  #38  
Old 04-05-2006, 01:09 AM
Gama's Avatar
Gama
Registered User

Gama is offline
 
Join Date: Dec 2005
Location: Melbourne
Posts: 1,121
For the old Cmos Vs CCD, heres a paper that goes thru everything on the 2 sensors, very indepth. Tells the pro's and cons for both sensors and why they are better.
http://huhepl.harvard.edu/~LSST/gene...paper_2003.pdf
Also in fairness it is 3 years old now.
Reply With Quote
  #39  
Old 04-05-2006, 05:12 AM
CometGuy's Avatar
CometGuy
Registered User

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

Near the bottom of Christian's 10D vs D70 page:

"At the 500nm wavelength, isolated by an interferential filter of 20nm range, it appears that the Nikon D70 produces 0.848 times less signal than the Canon 10D in the green channel for a same ISO sensitivity. This being computed by correcting the instrumental effect (pixel size effect, lens type)."

He does not mention he corrects it for differences in amplifier gain which he gives as 2.98 e-/ADU for the D70 and 2.41 for the 10D. So the actual relative green response is:

0.848 x 2.98/2.41 = 1.05x (OK a 5% relative advantage to the Sony

In this case he goes on to say that noise is somewhat higher in the D70, etc, etc so the final outcome is that there is similiar performance (but that the camera has to be worked using 'mode 3') .

These tests were done with the hotfilter in place, but what I was saying is that removing them would not effect the above Quantum efficiency result (i.e the hotfilter transmissions are very similiar).

I have seen QE figures for the Sony monochrome CCD's at 67% from the starlight site for the sensor used in the SXV-H9 - I certainly believe that. For the SXV-M25 it is 'equivalent to 60% QE at peak of green filters' and that there is 50% rolloff (I interprete this as 50% QE) at 400nm and 650nm. This sounds like they are describing the underlying monochrome sensor. So you need to consider the effect of the bayer filters, I think.

Terry
Reply With Quote
  #40  
Old 04-05-2006, 05:56 AM
CometGuy's Avatar
CometGuy
Registered User

CometGuy is offline
 
Join Date: Feb 2005
Location: Brisbane
Posts: 942
Update : I also Noticed Christian updated his 5D evaluation (I was looking for QE data for the 5D) with some test notes on the D200 and mentions the presence of the same 'median' type filter as per the D70. The link is:

http://astrosurf.com/buil/5d/test.htm

Gama, I reckon you should try to verify this yourself in the raw files you already have. If it is there then perhaps it should be mentioned to Nikon that this undesirable for scientific applications, and maybe they might release a firmware update.

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 01:38 AM.

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