Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

What's the verdict with ICX-694 cameras?

This topic has been archived. This means that you cannot reply to this topic.
53 replies to this topic

#1 Phil Hosey

Phil Hosey

    Vanguard

  • *****
  • topic starter
  • Posts: 2311
  • Joined: 29 Oct 2008

Posted 19 May 2014 - 06:26 PM

Since field of view isn't an issue for me (if I want more FoV I'll use a short FL scope), let's leave it out for the moment. I recently sold my QSI 683 because first, I found that I really didn't like the all-in-one setup and to change things I would have to buy expensive parts from QSI, add to that the fact that all my filters are 1.25" which start to vignette around f/5 (I want to use faster scopes). I'm looking heavily at the 694 based cameras and I want to make sure I understand the arguments for each. I know 8300 has more area (FoV), but I'm more interested in other attributes like pixel size, QE, read noise, dark current, etc. I've read that the higher QE of the 694 is offset by the smaller pixels and then there were some counter-arguments about how QE is measured etc. 694 ownders rave about how much more sensitive it is compared to the 8300. 8300 owners talk about how much more FoV they have. I'm still leaning toward the Sony chip, so my next question is, Atik, SX or QSI? QSI has a good rep but if I went that route again I would get just the camera and use someone else's filter wheel. I already have an OAG that could be used. ATIK has a good rep for low noise cameras and I've had one of their cameras before (314L+) and liked it alot. SX has been hit or miss for me. Some of their older designs have had very irritating issues and tests report from cameras as late as the SXVR generation all had higher read noise than advertised. Having said that though, the Trius SX-694 looks like a different animal with more features and they've apparently fixed the issue from the previous generations, but I've seen only a handful of images on Astrobin, there seem to be many more ATIK 460EX users out there. SX claims they use class 1 CCDs while ATIK advertises class 2. Unless there's a compelling reason for me to stay with 8300 sensor other than FoV, I'm pretty sure I'm going with 694, just not sure which one. Would love to hear from folks who've owned these cameras, or made the switch in either direction.

#2 Mr.Magoo

Mr.Magoo

    Lift Off

  • -----
  • Posts: 16
  • Joined: 19 May 2014

Posted 19 May 2014 - 07:10 PM

I have an Atik 460EXM. I like it. I was using it initially with the SX filter wheel but I had issues with the SX filter wheel when imaging under -20c. It would freeze and stop turning. I then bought the Atik filter wheel and OAG and they mate like a glove.

#3 Phil Hosey

Phil Hosey

    Vanguard

  • *****
  • topic starter
  • Posts: 2311
  • Joined: 29 Oct 2008

Posted 19 May 2014 - 07:51 PM

I have an Atik 460EXM. I like it. I was using it initially with the SX filter wheel but I had issues with the SX filter wheel when imaging under -20c. It would freeze and stop turning. I then bought the Atik filter wheel and OAG and they mate like a glove.


It got -20C here once.. DSLR really liked it. Fortunately that is a rare occurance where I live.

#4 Konihlav

Konihlav

    Surveyor 1

  • -----
  • Posts: 1675
  • Joined: 05 Mar 2009

Posted 20 May 2014 - 03:21 AM

Phil Hosey: the major advantage of ICX694 over KAF8300 family is much lower readout noise which is HUGE improvement for narrow band imaging. Another advantage is the QE. The downsides you are aware of are the FOV (chip size) and little hit by smaller pixel size which has direct impact on the final SNR when you scale the same image from these two different chips to same image scale... but as you know it's always a compromise, I personally much favor ICX-694 over 8300 because for larger FOV I prefer to use a full frame big beast camera that has 6-8 times bigger chip surface and in LRGB imaging is much much better then the super sony ICX694 provided you own the high-end telescope to give excellent image quality on such a big chip surface (large image circle). But that's another story (reason for me to own the best from both Worlds).

Anyway, I like Atik over SX. But recently I managed to persuade one new user of QSI with this SONY chip family (not 694 but the 814 one) to provide to me a set of biases and flats in order to make my measuring comparison, see:
http://blog.astrofot...ch/?page_id=782

and I was nicely surprised that the new QSI cameras perform very well. So, if I am in the USA I may, maybe prefer QSI over Atik as I always recommend people to stay within continent - because if something goes wrong the resolution takes much less time and is painless without much paperwork...

If there's someone with new QSI 660 camera I'd still like to run my measuring test run if I am given the biases and flats :-) though I myself have already "all these questions answered" and do not need to extend my research - but for other people it still might be useful source of comparison based on same approach, same methodology rendering useful and trustful output.

#5 orion69

orion69

    Viking 1

  • -----
  • Posts: 998
  • Joined: 09 May 2010

Posted 20 May 2014 - 04:19 AM

What about Sony ICX814 chip vs ICX-694 vs KAF8300?
Any major dissadvantages to 694?

814, 694 FOV is OK for me...

#6 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 23857
  • Joined: 10 Jan 2014

Posted 20 May 2014 - 04:49 AM

Just for some basic math around Q.E. vs. pixel size:

The KAF-8300 has 5.4µm pixels, while the ICX-694 has 4.54µm pixel size. In terms of area, the KAF has a 41.5% advantage.

The ICX-694 has 77% Q.E., while the KAF has 56% Q.E., and the advantage goes to the Sony at 37.5%.

Read noise is similar between the two, ~7e- for the ICX, ~9.3e- for the KAF. I don't think the ~2e- difference here is going to be particularly significant, although the Sony sensor's lower read noise may be enough of a margin to matter for your particular applications.

Quantum efficiency is a simple measure of the ratio of incident photons at the photodiode that release an electron (photon to charge conversion ratio). Area is not a factor here. Given that, in terms of total sensitivity, the larger pixels of the KAF actually result in a 4% sensitivity advantage over the Sony. So you are right, the smaller pixels of the Sony sensor to cost it, and the increase in Q.E. really only compensates for that (purely speaking in terms of sensitivity.)

There is more to think about than just sensitivity, though. The KAF obviously has a different image scale than the Sony at any given focal length. You'll need to determine on your own, based on whatever scopes you use, which image scale is more appropriate.

I think the true advantage of the Sony sensors is their low dark current. At 0.003e-/s/px, compared to the 0.04-0.02e-/s/px of the KAF, the Sony sensor really has amazingly low dark current. For a 600 second exposure, the Sony pixels won't even accumulate 2e- worth of dark current noise, while the KAF will accumulate 12 to 24e- worth of dark current noise. At less than two pixels of dark current for a 10 minute exposure, I truly believe the Sony sensors don't need dark frames.

So, if the ability to ditch dark frames entirely is a serious goal, then I'd say ICX-694 based CCD cameras will probably serve you well. If you need the smaller pixels for a better imaging scale, then it might still serve you well. If your looking for a real sensitivity increase, assuming you use both sensors at a similar imaging scale, the KAF-8300 sensor is still just as sensitive, as despite having lower Q.E., it has larger pixels, so it can still gather pretty much the same light in the same time. I don't think the ICX-694 is really even advantageous for narrow band imaging vs. the 8300...they are both going to gather the same amount of light in a given exposure time, be it 10 minutes or 20 minutes. The big difference is that over a 20 minute exposure, the Sony is going to trounce the 8300 in terms of dark current accumulated.

#7 Konihlav

Konihlav

    Surveyor 1

  • -----
  • Posts: 1675
  • Joined: 05 Mar 2009

Posted 20 May 2014 - 05:13 AM

What about Sony ICX814 chip vs ICX-694 vs KAF8300?
Any major dissadvantages to 694?

814, 694 FOV is OK for me...

The 694 is better choice then 814, the 814 has really small pixels and what actually matters in this case is the chip surface and that is equal.

#8 Konihlav

Konihlav

    Surveyor 1

  • -----
  • Posts: 1675
  • Joined: 05 Mar 2009

Posted 20 May 2014 - 05:19 AM

Phil Hosey: I have to say that I like your approach "if I want more FoV I'll use a short FL scope" :) that's simply true.

Jon Rista: I have seen some of your posts in other thread and it looks to me like you have no real (personal, direct) experience with many CCD cameras and obviously not expertise in narrow band imaging (have no clue how different is RN 5e- vs RN 10e-). Also, I noticed that you not always read what other people write to you (e.g. QSI and KAF-8300 cameras - they've been doing them for 3-4 years already!). I posted my opinion you have posted yours, period. I don't like these discussions which always end with a flame war. End of off topic from my side.

#9 JJK

JJK

    Gemini

  • -----
  • Posts: 3465
  • Joined: 27 Apr 2008

Posted 20 May 2014 - 05:26 AM

Phil Hosey: the major advantage of ICX694 over KAF8300 family is much lower readout noise which is HUGE improvement for narrow band imaging. Another advantage is the QE. The downsides you are aware of are the FOV (chip size) and little hit by smaller pixel size which has direct impact on the final SNR when you scale the same image from these two different chips to same image scale... but as you know it's always a compromise, I personally much favor ICX-694 over 8300 because for larger FOV I prefer to use a full frame big beast camera that has 6-8 times bigger chip surface and in LRGB imaging is much much better then the super sony ICX694 provided you own the high-end telescope to give excellent image quality on such a big chip surface (large image circle). But that's another story (reason for me to own the best from both Worlds).

Anyway, I like Atik over SX. But recently I managed to persuade one new user of QSI with this SONY chip family (not 694 but the 814 one) to provide to me a set of biases and flats in order to make my measuring comparison, see:
http://blog.astrofot...ch/?page_id=782

and I was nicely surprised that the new QSI cameras perform very well. So, if I am in the USA I may, maybe prefer QSI over Atik as I always recommend people to stay within continent - because if something goes wrong the resolution takes much less time and is painless without much paperwork...

If there's someone with new QSI 660 camera I'd still like to run my measuring test run if I am given the biases and flats :-) though I myself have already "all these questions answered" and do not need to extend my research - but for other people it still might be useful source of comparison based on same approach, same methodology rendering useful and trustful output.


Do you have data that suggests the dark frame calibrated 8300 is too noisy for NB imaging? Due to unusually cloudy weather this past Winter, I wasn't able to critically test every aspect of a new FLI MicroLine 8300 monochrome CCD camera, but what little data I recorded in NB H--alpha looked decent, even w/o flat field calibration.

If a scope can cover the FOV of an 8300 chip, any the imager wants to image at that scake, then any QE advantage of a chip that's half the size is moot.

#10 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 23857
  • Joined: 10 Jan 2014

Posted 20 May 2014 - 05:38 AM

Jon Rista: I have seen some of your posts in other thread and it looks to me like you have no real (personal, direct) experience with many CCD cameras and obviously not expertise in narrow band imaging (have no clue how different is RN 5e- vs RN 10e-). Also, I noticed that you not always read what other people write to you (e.g. QSI and KAF-8300 cameras - they've been doing them for 3-4 years already!). I posted my opinion you have posted yours, period. I don't like these discussions which always end with a flame war. End of off topic from my side.


I'm not sure what flame war you are talking about. You are the one who just made things personal. I don't have any direct experience with CCD imaging, however I do understand sensor design. I believe the read noise difference between an ICX-694 sensor and the KAF-8300 is closer to 7e- vs. 9e-, rather than 5e- vs. 10e-, but...let's not start a war, shall we?

:confused:

#11 Konihlav

Konihlav

    Surveyor 1

  • -----
  • Posts: 1675
  • Joined: 05 Mar 2009

Posted 20 May 2014 - 05:41 AM

JJK: sure I do have the data, results, test and comparisons. Majority is, in some form, on my blog page. Without real experience I would be just talking guesses and the Internet, CN, other discussions are full of myths and incomplete information from just beginner users. This is simple fact - that it's a very hard job to find relevant information on any topic in the Internet garbage growing every day.

valid sum up's (facts):
- QE is important for both broad band imaging and narrow band imaging
- RN is of insanely high importance in narrow band imaging (and lucky imaging) only

important parameters for your newly selected CCD camera (just few of them):
- get as big chip surface as possible
- get as big telescope aperture as possible with corrected image circle (corrected FOV) to match the big chip size

I personally do not like the myths about matching CCD cameras to telescope because telescopes change (or focal reducers change) as you sell and buy new ones. My philosophy is to get the best for the given purpose (it renders the need for at least 3 cameras and 3 telescopes :) ).

on CN, there are 50+ threads full of "flame wars" on which CCD camera to select. Everyone's needs differ.

#12 JJK

JJK

    Gemini

  • -----
  • Posts: 3465
  • Joined: 27 Apr 2008

Posted 20 May 2014 - 05:52 AM

JJK: sure I do have the data, results, test and comparisons. Majority is, in some form, on my blog page. Without real experience I would be just talking guesses and the Internet, CN, other discussions are full of myths and incomplete information from just beginner users. This is simple fact - that it's a very hard job to find relevant information on any topic in the Internet garbage growing every day.

valid sum up's (facts):
- QE is important for both broad band imaging and narrow band imaging
- RN is of insanely high importance in narrow band imaging (and lucky imaging) only

important parameters for your newly selected CCD camera (just few of them):
- get as big chip surface as possible
- get as big telescope aperture as possible with corrected image circle (corrected FOV) to match the big chip size

I personally do not like the myths about matching CCD cameras to telescope because telescopes change (or focal reducers change) as you sell and buy new ones. My philosophy is to get the best for the given purpose (it renders the need for at least 3 cameras and 3 telescopes :) ).

on CN, there are 50+ threads full of "flame wars" on which CCD camera to select. Everyone's needs differ.


I'll check your blog out tonight, but honestly, I haven't had any issues with NB imaging with an FLI ML8300, and I've barely scratched the surface (e.g., I haven't bothered using flat field calibrations).

I've attached a severely shrunken (for the CN file size requirement) work-in-progress image of IC443 (AP Traveler, FLI ML8300, Astrodon 3 nm H-a filter) with only a dark frame calibration (no flat fields). The full-size image looks much better due to less compression artifacts. Also, I am not yet expert in image processing.

I fully understand the virtue of greater QE. However, if a chip with 76 % QE is half the size of one with 56 % QE (peak values), and if you want to cover the same FOV as the larger chip, the gain in QE with the smaller chip is moot.

Attached Thumbnails

  • 6535635-IC443 36x20min Ha m30C DC_AT_HT_IT_USM_small file_CN.jpg


#13 Konihlav

Konihlav

    Surveyor 1

  • -----
  • Posts: 1675
  • Joined: 05 Mar 2009

Posted 20 May 2014 - 06:01 AM

Jon Rista: we shall not, no-one would benefit. But anyway, I have to say it now: the difference is even like 3.75e- (for example old SONY ICX-285) over 4.5e- (new high-QE SONY ICXblabla we talk about now) versus 8e- (SBIG, MII) to 10e- (FLI) to 12e- (SX, QHY) of RN of cameras with the same chip (KAF-8300) from different manufacturers :-) I had like 20+ cameras in my own hands and a lot more raw data from other users and friends.

ONE IMPORTANT STATEMENT: all the comparisons here are nice and valid, but what matters most is the pure fact that 99.99% of beginners imaging with a CCD camera do vanish (completely diminish) all the subtle details we argue about here. Why? Because they tend to use aggressive noise reduction in the post processing of their images. And then, everything becomes irrelevant :)

other point (going to end for today as I have to work, I have no time left to spend in these discussions) is that lot of people even do not focus well before they start to acquire images. And unfocused image can only be downsized to 50% or 30% or less. Then ALL noise factors are again completely irrelevant.

Unless you, like me, want full size image of 4000 pixels wide done in short time with almost no noise and high detail per pixel... then you start to chase for the best with no compromise. But the experience is non-transferable and everyone has different wishes. CS to all of you guys!

#14 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 23857
  • Joined: 10 Jan 2014

Posted 20 May 2014 - 06:15 AM

Jon Rista: we shall not, no-one would benefit. But anyway, I have to say it now: the difference is even like 3.75e- (for example old SONY ICX-285) over 4.5e- (new high-QE SONY ICXblabla we talk about now) versus 8e- (SBIG, MII) to 10e- (FLI) to 12e- (SX, QHY) of RN of cameras with the same chip (KAF-8300) from different manufacturers :-) I had like 20+ cameras in my own hands and a lot more raw data from other users and friends.

ONE IMPORTANT STATEMENT: all the comparisons here are nice and valid, but what matters most is the pure fact that 99.99% of beginners imaging with a CCD camera do vanish (completely diminish) all the subtle details we argue about here. Why? Because they tend to use aggressive noise reduction in the post processing of their images. And then, everything becomes irrelevant :)

other point (going to end for today as I have to work, I have no time left to spend in these discussions) is that lot of people even do not focus well before they start to acquire images. And unfocused image can only be downsized to 50% or 30% or less. Then ALL noise factors are again completely irrelevant.

Unless you, like me, want full size image of 4000 pixels wide done in short time with almost no noise and high detail per pixel... then you start to chase for the best with no compromise. But the experience is non-transferable and everyone has different wishes. CS to all of you guys!


I think some of your points are getting off topic a bit, but I think I understand where your coming from regarding read noise. As I mentioned in my post, the difference in read noise between an ICX-694 and KAF-8300 "may be enough of a margin to matter for a particular application." Narrow band would certainly qualify. If I understand where your coming from, your saying that because NB imaging increases image contrast, and the deepest background sky can end up exceptionally dark when using very narrow bands (i.e. 3-5nm), then sure, read noise would affect the background sky more in an NB image than in an LRGB image.

However, I am wondering if you might be exaggerating the difference between a ~5e- RN camera and a ~10e- RN camera? You mention on your blog that the difference between 4e- and 10e- means that you would need 6x more exposure time to get the same result with a 10e- sensor. Could you explain that to me?

(Note, this is an honest question, I'd like to understand your reasoning, not start a flame war.)

In my understanding, expose for twice as long, and you double the light gathered. Technically speaking, if you exposed for 2.5x longer with the 10e- sensor, you would have the same S/N as with the 4e- sensor. If you exposed for 6x longer with the 10e- sensor, you would have 2.5x the S/N. That simple fact should apply even for the dark background sky. If your background sky level is at 32e- on the 4e- sensor, then doubling (and a half) the exposure time on the 10e- sensor should make the background sky level 80e-...either way, 8:1 S/N. Is there some other factor involved in NB imaging that changes this fact? The only obvious relationship between a 4e- and 10e- RN that I could find is the fact that the 10e- sensor has 6e- more read noise...but unless each electron worth of noise equates to an exposure time multiple...I'm confused as to why you would need 6x more exposure time to get the same S/N.

#15 Konihlav

Konihlav

    Surveyor 1

  • -----
  • Posts: 1675
  • Joined: 05 Mar 2009

Posted 20 May 2014 - 06:57 AM

JJK: it's fine you have no problems :) it would be pity and shame if you would. The 8300 is not sooo bad Kodak chip, it's just noisy (as we talk about the noise now), but it has reasonable QE for Ha, OIII and also for SII. As I said, since everyone reduces noise with very aggressive blurring of the background, the discussion here becomes irrelevant.

Jon Rista: I don't have the equation here now, nor remember correctly the exact final form of it, but here are some thoughts (to honestly answer at least something now, until I find energy and time to post the charts on my blog page):
- with narrow band imaging we talk about background sky level of 0.0000000nothing, it's truly so low and this is the only case when readout noise matters so much (opposite, in LRGB and polluted skies when you do not give a ... to how big the RN is as it becomes unimportant)
- the QE obviously plays huge role in NB too, but some huge difference is in say KAI-11002 which has 31% of QE versus these SONY with 67%+
- simply said, you may just square root the 4*4=16 vs 10*10 = 100 and 16 is like 6x less then 100. this is kinda simplified math :)
- there are MANY other factors involving final result, one of them is (one of the most important, not yet mentioned) is the LENGTH of SUB-exposition. The longer SUB you make the less obvious the difference I talk about is. But the problem is that we amateurs make subs in form of 20-30 min at most, not 120 minutes sub in order to equal the 4e- RN chip with our 10e- RN camera...

that's the other fact, not yet mentioned, you need to make long individual sub expositions with higher RN camera then with a low RN camera (reason why lucky imaging works with SONY chips perfectly well). For instance, with 4e- RN camera it makes sense for me to make like 10-12min SUBS at most with 3nm filters. With 10e- RN camera it makes sense to expose an individual sub for 120 minutes...

note how big FW (full well) capacity does have these noisy chips - much bigger then the SONY ones... guess why :) it correlates to each other.

this topic would require 100 of pages to describe all aspects. That's the reason I speak about it rarely as I do not want to start the flame of answering questions of users coming into this discussion.

we should stick with "what's the verdict" of ICX694 cameras :-) to keep the original poster HAPPY!

#16 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 8959
  • Joined: 12 Oct 2007

Posted 20 May 2014 - 07:02 AM

You mention on your blog that the difference between 4e- and 10e- means that you would need 6x more exposure time to get the same result with a 10e- sensor. Could you explain that to me?



This kind of thing is easy to explain - and I don't think it is important. In certain situations higher read noise will make it take long time to match another camera - but what matters is how good the image is - in a given time. If the image with one camera is 90% as good as another after 2 hours - and 2 hours is all you wanted to spend - then there isn't much difference - even if it would have taken a year of additional exposure to match it.

When you combine this approach with JJK's comments - which I think are on target regarding the overwhelming importance of camera size - then the benefits of the 8300 are more clear compared to its slight disadvantages.

I have the Atik 383L+ and - yes - I was mainly motivated by the sensor size and the fact that I am not in some mythical game of exposing until I exactly match the SNR I would have had with a different camera.

And note that in many situations, read noise has little impact at all.

Frank

#17 blueman

blueman

    Photon Catcher

  • *****
  • Posts: 6925
  • Joined: 20 Jul 2007

Posted 20 May 2014 - 01:16 PM

FOV is a big deal to me. I have used small chips with small pixels and to capture the bigger nebula I need all the FOV I can muster. It is possible to use a telescope to increase the FOV too of course.
I think that the KAF8300 has gained such popularity because it is cost effective and does a nice job as can be seen by so many good photos taken with them. It has a large enough size to allow you to use a lot of scopes with it but cheap enough not to break the bank.
I am not sure there is a perfect chip or camera out there. They all seem to be a compromise to some degree.
Sony chips have always been quiet and produce nice images. But they continue to make small chips. I would really like to see them make some larger ones with a larger pixel. Imaging at sub arc sec is tough for most of us. Seeing conditions that support this kind of image scale are rare anymore. So larger pixels would help with that.
Blueman

Attached Thumbnails

  • 6536341-M-42_LRGB_WL.jpg


#18 JJK

JJK

    Gemini

  • -----
  • Posts: 3465
  • Joined: 27 Apr 2008

Posted 20 May 2014 - 02:17 PM

JJK: it's fine you have no problems :) it would be pity and shame if you would. The 8300 is not sooo bad Kodak chip, it's just noisy (as we talk about the noise now), but it has reasonable QE for Ha, OIII and also for SII. As I said, since everyone reduces noise with very aggressive blurring of the background, the discussion here becomes irrelevant.

Jon Rista: I don't have the equation here now, nor remember correctly the exact final form of it, but here are some thoughts (to honestly answer at least something now, until I find energy and time to post the charts on my blog page):
- with narrow band imaging we talk about background sky level of 0.0000000nothing, it's truly so low and this is the only case when readout noise matters so much (opposite, in LRGB and polluted skies when you do not give a ... to how big the RN is as it becomes unimportant)
- the QE obviously plays huge role in NB too, but some huge difference is in say KAI-11002 which has 31% of QE versus these SONY with 67%+
- simply said, you may just square root the 4*4=16 vs 10*10 = 100 and 16 is like 6x less then 100. this is kinda simplified math :)
- there are MANY other factors involving final result, one of them is (one of the most important, not yet mentioned) is the LENGTH of SUB-exposition. The longer SUB you make the less obvious the difference I talk about is. But the problem is that we amateurs make subs in form of 20-30 min at most, not 120 minutes sub in order to equal the 4e- RN chip with our 10e- RN camera...

that's the other fact, not yet mentioned, you need to make long individual sub expositions with higher RN camera then with a low RN camera (reason why lucky imaging works with SONY chips perfectly well). For instance, with 4e- RN camera it makes sense for me to make like 10-12min SUBS at most with 3nm filters. With 10e- RN camera it makes sense to expose an individual sub for 120 minutes...

note how big FW (full well) capacity does have these noisy chips - much bigger then the SONY ones... guess why :) it correlates to each other.

this topic would require 100 of pages to describe all aspects. That's the reason I speak about it rarely as I do not want to start the flame of answering questions of users coming into this discussion.

we should stick with "what's the verdict" of ICX694 cameras :-) to keep the original poster HAPPY!


I don't use aggressive blurring of the background. For the Jellyfish Nebula image, I only applied a light wavelet transform in the linear state to minimize noise in the fainter regions. After applying a Histogram Transformation and Curves Transformation (non-linear processes), I applied an Unsharp Mask to get back some of the sharpness I lost with the Wavelet Transform. If anything, that brings back some noise in the fainter areas. Obviously, with an object this complex, shrinking a 4MB JPEG to 200 KB injects compression artifacts, which looks like noise. The original image in TIFF format looks far better.

My general imaging plan is to take a statistically significant amount of data (pretty hard this past *BLEEP* Winter & Spring). More, not less data, with any CCD camera is the key to beating down some noise sources. I don't think read noise is as big an issue as you suggest.

If SONY comes out with larger format chips, I'll given them a try. Until then, I'll happily use my terrible 8300 monochrome camera and will order a terrible 16803-based camera this month.

#19 Phil Hosey

Phil Hosey

    Vanguard

  • *****
  • topic starter
  • Posts: 2311
  • Joined: 29 Oct 2008

Posted 20 May 2014 - 04:41 PM

After hearing a lot arguments and counter-arguments about the differences between the KAF-8300 and the ICX694 it is still almost a toss-up for me. I had two goals in mind when I sold my QSI683, one was to split up the monolith so to speak as I have grown to dislike the all-in-one aspect of the camera. In fact that was the only thing I disliked about it. The other was to try something different, ie, the 694. Partly because I could continue to use 1.25" filters and because I *thought* I could get cleaner and smoother images with the sony chip, but so far nothing has shown me that it is substantial enough of a difference to justify giving up so much real-estate. So now I'm leaning toward staying with the 8300 for now.

#20 Konihlav

Konihlav

    Surveyor 1

  • -----
  • Posts: 1675
  • Joined: 05 Mar 2009

Posted 20 May 2014 - 04:42 PM

blueman, JJK: I understand that you don't like to hear what I wrote (AFAIK I just wrote KAF-8300 is very noisy which is just simply true, not much to argue about) and you do not have to support your 8300 based camera by providing some of your images (there are billion of them on thee Internet). I really believe it works well and very well for you and maybe it's your only camera. After you use and test many more cameras you learn more. You both did not pay attention that I said, the case where the noise matters is narrow band imaging and lucky imaging only. In LRGB space, the 8300 outperforms (due to bigger FOV) ICX-694, especially under a heavy polluted skies. What I wanted to point out is how BIG the difference in total exposition time is in this narrow band imaging case comparing 5e- vs 10e- RN cameras. That is something 99.99% do not have clue about and do not realize.

just for your curiosity. There's one CCD chip that is worst ever on planet Earth, real mess. And I have, myself, intentionally purchased a camera equipped with such a mess. Why? chip size 36x24mm and price about 5000 USD (very cheap camera). Since I use it with F/3.7 system and ONLY for LRGB imaging, I really win every time I shoot with it under dark sky with LRGB filters. Narrow band with it makes no sense, it's totally blind.



Phil: Peter in Reno promised to send some biases and flats to me so as I can measure his new CCD and tell if the newest QSI cameras are truly perfect (so far I had one from them and it was very, very good, even a hair better than Atik).

If you want to make a big leap forward in equipment improvement and your image quality improvement (apart from post processing skills) I would recommend 36x24mm CCD camera (or even the 36x36, but that gets very expensive with filters). Bad fact is that such a leap forward calls for an expensive astrograph :-(

nowadays I have a system that is 10x more effective in LRGB imaging then my old little Borg with ICX-694. The answer for this was the F/3.7 10" aperture system with super noisy KAI-11002 that requires lot of cooling and precise, very precise callibration...

#21 Konihlav

Konihlav

    Surveyor 1

  • -----
  • Posts: 1675
  • Joined: 05 Mar 2009

Posted 20 May 2014 - 04:46 PM

for instance, this is just 5.5 hours of total exposition:
Posted Image

and this again, 5.5 hours total:
Posted Image

and ICX-694 with ASA 10" (as I said, my philosophy is "best" camera and best telescope):
Posted Image

more in my gallery...

#22 Konihlav

Konihlav

    Surveyor 1

  • -----
  • Posts: 1675
  • Joined: 05 Mar 2009

Posted 20 May 2014 - 04:51 PM

Phil: unfortunately it's always about money and budget. But if you want to keep the kinda slow refractor, maybe you should try to upgrade the focal reducer to something like 0.66x to make your F-stop ratio faster which would, hand-in-hand go well in your overall "gain" with ICX694 replacement...

#23 Phil Hosey

Phil Hosey

    Vanguard

  • *****
  • topic starter
  • Posts: 2311
  • Joined: 29 Oct 2008

Posted 20 May 2014 - 05:45 PM

Phil: unfortunately it's always about money and budget. But if you want to keep the kinda slow refractor, maybe you should try to upgrade the focal reducer to something like 0.66x to make your F-stop ratio faster which would, hand-in-hand go well in your overall "gain" with ICX694 replacement...


Pavel,
I have a 10" f/4.7 newtonian, it is a little faster than the refractor with a reducer and much more aperture. I'll need to upgrade the focuser but thinking this could give a little boost anyway.

#24 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 23857
  • Joined: 10 Jan 2014

Posted 20 May 2014 - 06:00 PM

blueman, JJK: I understand that you don't like to hear what I wrote (AFAIK I just wrote KAF-8300 is very noisy which is just simply true, not much to argue about) and you do not have to support your 8300 based camera by providing some of your images (there are billion of them on thee Internet).


I think this statement is a little disingenuous and a little unfair. The KAF-8300 is only "very noisy" RELATIVE to the new Sony sensors on the market. Sony has changed a lot of things, they brought new life to the game, and only relatively recently at that. Prior to Sony's entrance into the global sensor manufacturing market, KAF sensors were known as having very low noise. I mean, if you look back in history not even that far, it wasn't uncommon in the past for sensors to have as high as 40e- worth of read noise. Enter the KAF series of sensors into THAT market, and suddenly their 7-10e- worth of read noise is exceptionally good.

In the grand scheme of things, while KAF sensors aren't as good as the newest of the new generation of sensors, they are still quite good overall. Look at the last five years or so of images taken with KAF sensors...some of THE MOST exceptional astrophotography ever created, including narrow band images, were created with KAF sensors.

Yes, Sony's ICX line is better technologically, in some ways, but just because something better came along doesn't suddenly mean that the "old" KAF sensors are now worse than they were before. They are still immensely capable sensors, they are just no longer one of the best.

I just think a proper context is necessary here. I've heard the same kinds of arguments for too many years when debates about Sony Exmor vs. Canon sensors crop up. As soon as Exmor hit the market in the D800 and D600, suddenly somehow Canon cameras became perceptually "worse" than they were before, despite the fact that there are MILLIONS of photos that show how incredibly capable every Canon sensor since the 1Ds III and 5D II came along are.

It's fallacious and disingenuous to say KAF sensors are very noisy simply because they don't measure up to a new generation of sensors. They are still the same old sensor that astrphotographers have been producing truly phenomenal images with for years...that fact hasn't changed.

I hope I can claim an objective stance here, given that I do not yet own a CCD camera (I'm only just on the market for one.) I would LOVE to have a Sony sensor given their technical strengths, however I'm on the fence about them for my own purposes given their technical weaknesses. I've not yet made any decision about whether to get a CCD with a Sony sensor or a Kodak sensor.

I really believe it works well and very well for you and maybe it's your only camera. After you use and test many more cameras you learn more. You both did not pay attention that I said, the case where the noise matters is narrow band imaging and lucky imaging only. In LRGB space, the 8300 outperforms (due to bigger FOV) ICX-694, especially under a heavy polluted skies. What I wanted to point out is how BIG the difference in total exposition time is in this narrow band imaging case comparing 5e- vs 10e- RN cameras. That is something 99.99% do not have clue about and do not realize.


I also think your being a bit disingenuous about read noise. Your own data indicates that ICX sensors have read noise ranging from ~4.5e- to as high as 10.6e-. Concurrently, your own data indicates that KAF sensors have read noise ranging from ~7e- through around 10.8e-. The story isn't just as simple as ICX is 5e- and KAF is 10e-. Pick the right brand and model of camera, and you can get a KAF sensor with a "< 7e- RMS" RN rating, and pick the wrong brand and model of camera, and you can end up with an ICX sensor with a "< 10e- RMS" RN rating.

It's ok to be a fan of a particular technology, for sure! Nothing wrong with having an affinity with something. But I think it's unfair to assume that that new and preferred technology suddenly makes older technology no longer capable of doing what people have been doing with it for years. I don't believe there is any real evidence to support the notion that you MUST HAVE 6x or so the exposure time on a sensor with 10e- read noise...on the contrary, given some of the phenomenal NB images created with a wide range of Kodak sensors, I think quite the opposite is true. I understand the technical argument of why more read noise MAY be a problem, due to how close to true black the background sky gets with narrow band filtration, but I'm just not sure that there is any fundamental evidence to indicate the problem is that big of a problem in real-world practical situations. I also believe that the problem can be mitigated with more total integration time.

It sounds like your personal holy grail is to achieve the shortest possible exposures and least total integration time possible for your images. It's a laudable goal, and your work is great, however I don't think it's the only possible option, nor do I think that the quality of your work negates the quality of others work done with KAF sensors.

I understand your an advocate for Sony ICX sensors, and they are certainly good for what they are, but they are not the end-all-be-all of sensor design. They have their shortcomings, some of which can be quite significant depending on the photographer's ultimate goals. The small total area and generally small pixel size of the ICX sensors is often a drawback. Similarly, while KAF sensors do have more read noise, they also have their strengths. Efficiency of cost per total sensor area, for one, is a BIG benefit of the KAF sensors (particularly the 8300, given how cheap you can pick up a camera with one of those in it for these days), and with adequate cooling, you can mitigate noise considerably (-40°C cooling on the LOW end KAF-8300's, and -55°C cooling on the midrange models, you can certainly get the necessary cooling to start leveling the playing field.)

I know you don't want a flame war, but you need to be a little more objective. You have some clear biases, and I don't think they are helping you offer the most honest and helpful assessment of the options, not in a way that will help the OP make a decision that really, truly benefits HIS OWN work. It comes off more as evangelism for a particular type or brand of product. Again, that's not to say you are necessarily wrong, on any or all counts, just...maybe disingenuous. (It may indeed be that an ICX-694 is what he needs, but he should be fully equipped with all the real facts and capable of making up his own mind about it, rather than succumb to your own powerful opinion. ;P)

#25 orion69

orion69

    Viking 1

  • -----
  • Posts: 998
  • Joined: 09 May 2010

Posted 20 May 2014 - 06:56 PM

...and ICX-694 with ASA 10" (as I said, my philosophy is "best" camera and best telescope):
Posted Image


My idea was to improve narrowband imaging by buying either 694 or 814 chip.
I'm glad you posted this version of Pacman nebula but frankly I'm little puzzled now...

Here is my version with setup from my signature:

Posted Image


To be honest I don't see that much improvement considering the time used was same (I did use longer subs), you have better filters and of course 10" vs 6". Also I was shooting @ f/6.67 and was not using Riccardi reducer 0.75x which I didn't have at the time...
Of course I don't know your sky quality but mine was nothing special, also processing is of course different.

Please everyone, do not consider this as "which image is better" thing, I'm just trying to decide if buying Sony chip is good idea for me.
From what I see maybe you saved me some big chunk of money. :)


CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics