Is CMOS (rather than CCD) the future?

[Tim99]

What about something like this rather than a DSLR.  What can you tell me?

The ZWO ASI224MC Color CMOS Imaging Camera follows in the footsteps of the ASI120 when it comes to cutting-edge functionality like a high speed USB 3.0 port, an ST4 guide port, a standard 2” interface with M42 x 0.75 internal threads, a 1/4” tripod thread mount on the back, and an IR-cut window. However, the ZWO ASI224 camera has one important difference, and that is the inclusion of a Sony IMX224 Color CMOS sensor. This image sensor has a lot going for it when it comes to high quality solar, lunar, planetary, and deep sky imaging.

[pete_l]

 ...  And the 460 chip is not that small. On my 80mm ED refractor (FA 480 mm) I should even be able to fit in M31, but in some cases I may try to do some mosaics.

If you can image M31 in one frame you can get all the other Messiers, too. 90% of them are smaller than 40 arc-min and half are smaller than 10 arc-min. So how much money are people willing to spend to capture all the dark sky around the object of interest?

While there are obviously some objects and areas of the sky that benefit from massive fields, there are plenty more where it is unnecessary. Plus if you do wish to frame your objects, the mosiac of the field surrounding the target doesn't need anywhere near as many subs as the central portion, since it contains virtually no information.

[wuthton]

Sorry - mis-read a post, deleted.

[ollypenrice]

Talking about big CMOS sensors is all well and good, but ultimately pointless if you dont have the telescope to take advantage of it - as large, flat fields are not cheap either.

Yes Taks and Vixens to name two.

Derek

The WO Star 71 can cover full format.

With regard to the number of megapixels, this only matters if you have a short enough focal length to make having them worthwhile. If you have a longer focal length they are a positive disadvantage. In my view the size of your pixels is over-rated as an obsession. Have a look at the Pelican's neck in the full size version of this image taken with SGL member Fordos Moon this week.

https://ollypenrice.smugmug.com/Other/Best-of-Les-Granges/i-Q4NKT3j/0/O/NAM%20PELICAN%2014Hrs.jpg

That's at 3.5 arcsecs per pixel, a resolution many would dismiss as contemptible. There's no trickery here, but the way. I do blend higher res images into widefields but not in this case.

Olly

[ollypenrice]

If you can image M31 in one frame you can get all the other Messiers, too. 90% of them are smaller than 40 arc-min and half are smaller than 10 arc-min. So how much money are people willing to spend to capture all the dark sky around the object of interest?

While there are obviously some objects and areas of the sky that benefit from massive fields, there are plenty more where it is unnecessary. Plus if you do wish to frame your objects, the mosiac of the field surrounding the target doesn't need anywhere near as many subs as the central portion, since it contains virtually no information.

Dark sky? Where do you find that? 

One of the pleasures of a big chip is finding that there isn't any!

Anyone wishing to do deep sky astrophotography by taking a single sub without calibration should buy a Sony chip. That doesn't describe me...

lly

[perfrej]

Anyone wishing to do deep sky astrophotography by taking a single sub without calibration should buy a Sony chip. That doesn't describe me...

lly

Well, that's one statement

I tend to agree. Real estate is King!

/per

[ollypenrice]

  My point, really, is just that once you have the calibration sorted (which doesn't take long and, in my case, doesn't involve doing anything new more than once a year) you get a clean image from a stack of Kodaks.

However, what remains to be seen is whether Julian's recent and remarkable Squid was due to his Sony sensor or his Astrodon NB OIII filter (or both?)

Olly

[Zakalwe]

How does sensitivity factor into this? The Kodak chips are a far bit less sensitive than the Sonys, aren't they? So for the peeps that are blessed with seemingly endless cloud free nights they can throw hours of data at the job. For those under crummy UK skies, is the extra sensitivity of the Sony a much more important factor?

[juliangeorgeshaw]

The Squid? Did someone mention the Squid?

Here is some evidence bearing on Olly's question (does the principal advantage imaging the Squid accrue from using a Sony chip or from using an Astrodon 3nm Oiii filter?) and Zakalwe's question (does the extra sensitivity of the Sony chip vs. the 6303 make a difference in general?):

Nicholas Outters (the Squid's discoverer) used a Tak FSQ-106, Astrodon 3nm and 5nm Oii filters and an SBIG STL 6303 chipped camera. You can see his results in this document:

http://www.outters.fr/images%20site%20astro/decouverte-OU4-eu.pdf

For contrast my image is here:

http://www.astrobin.com/203894/

I used a Tak FSQ-85, Astrodon 3nm Oiii filter and an Atik 460exm Sony chipped camera, binned 2x2 (so my 4.54 micrometre pixels are grouped as 9.1 micrometre arrays which is almost the same as the 6303's 9 micrometre pixels). I took 20x20 minute subs but I threw away about 5 or 6 of them.

As Outters shows, the Astrodon 3nm filter does produce better results than the Astrodon 5nm filter with 8x30min subs. (By the way, this is consistent with the superiority of Sara's images using the Astrodon's much narrower band, NB filters images versus e.g. Baader NB filters which is what converted me to Astrodon. I am glad I paid extra for a 3nm Oii filter even though I did so with the intention of imaging in moonlight which doesn't really work, too noisy. IMO Astrodon NB filters are worth it, I only wish someone would make them cheaper).

Then Outters uses 25x30 minute subs with the 3nm Astrodon filter and the 6303 chipped camera.

IMO my naive attempt is much better than his even though I have less data and less aperture. So IMO the jury is in: using a Sony chip vs. the 6303 chip makes a big difference in favour of the Sony, at least on this target (the difference is certainly not due to the nut behind the camera, I am a newbie).

As Zakalwe says there is a huge difference in the measured sensitivities of the two chips. I know that theory doesn't always translate into reality in Astro imaging but it would surprise me if the sensitivity didn't make a difference on at least some targets and I think the Squid is one of them.

I do lust after the real estate of the 6303 but IMO the Squid is the most interesting object in the sky that I have come across so far (as a non-astronomer friend said it really does look like a squid. And to think that it is so large and yet was only discovered 4 years ago, by an amateur!). So for me "no squid = no deal"  (I will happily sacrifice the Bat for a good Squid) but others may differ.

[Horwig]

I don't know if this adds to the discussion or not, but these are two of my attempts at M81 and 82, one was a Kodak sensor, the other a Sony

the Kodak version was 50 minutes of luminance through a big Newtonian at f3.5 in 200 sec subs.

The Sony version was 120 minutes of luminance through a refractor at f5, in 90 second subs.

Both give roughly the same sampling in us/pixel.

I'm letting you work out which is which

Huw

[Stu_2011]

I don't know if this adds to the discussion or not, but these are two of my attempts at M81 and 82, one was a Kodak sensor, the other a Sony

m81+82-2.jpg

m81-82_c.jpg

the Kodak version was 50 minutes of luminance through a big Newtonian at f3.5 in 200 sec subs.

The Sony version was 120 minutes of luminance through a refractor at f5, in 90 second subs.

Both give roughly the same sampling in us/pixel.

I'm letting you work out which is which

Huw

Hi

Both great pictures, the top one is the Sony I guess, lots more detail to my old eyes. Also the diffraction spikes in the lower image gives a bit of a clue....

Thanks

Sent from my iPad using Tapatalk

[juliangeorgeshaw]

This is a great example Horwig, evidence always trumps assertions.

To my eyes the top image is MUCH better. If, as I suspect, it's made with the Sony chipped camera it confirms that there really IS a trade off between the real estate of the Kodak chips and the sensitivity of the Sony chips (which is what one would expect comparing the sensitivity charts for the two chips). [if the two images are the other way around I will look a bit silly ]

Is real estate king? Or is it sensitivity? It depends on your target doesn't it but for me it's sensitivity. The relatively modest extra real estate of the 8300 chip versus the Sony doesn't do much for me (albeit it would have given my M31 more breathing space). And Olly's magnificent ultra wide images are a bit beside the point because they use Kodak 11000 chips and need 2" filters (and you could buy a number of Atik 460s for the price of an 11000 chipped camera kitted out with Astrodon NB filters).

[Horwig]

Yes, the spikes on the bottom image give it away. The cameras were a QSI 6120 and an SX 35, so 3.1 micron pixels against 9 microns, The scopes were a 40cm Newtonian and an FSQ106.

In theory, the gathered photons would have been very roughly the same, the Newtonian is f3.5, the FSQ is f5, but the image detail in the top image is much greater.

However, put the SX camera on a wide enough scope and you get something entirely different, as seen from Olly's great work above.

Must try the QSI on a short stills lens.

Huw

[ollypenrice]

However Julian did it, his Squid is the best I've seen.

Collaboration, Sir, your OIII with our widefeild HaLRGB?

lly

[buzz]

It is an interesting comparison - but these are both CCD's, used on different equipment, with different exposure lengths etc. The colors pop on one but the data is on luminance, so I am a little confused, especially since the OP is on CCD vs CMOS.

We are very used to analyzing the data for well depth, read noise, thermal noise for CCD's. Until one has a proper back to back test with the same imaging workflow, operating temperature etc, it is difficult to draw a conclusion.

I had a Trius with the ICX694 Sony chip and replaced it with a QSI683. I admit the Sony was a quieter chip, evidenced by the calibration frames but do not notice it on the final result after calibration.

I have a APS-C Fuji X-T1, which I know has good red sensitivity. I have a t-thread adaptor for it and will try and compare it to the similar sized 8300 chip, both on the bench, and in practice (no TEC applied). There are some variables here too, like RAW frame conversion but it should be possible to get a measure of dynamic range and noise.

This comparison - by sheer chance of the same galaxy pair was taken with comparable WO refractors, at different times, different processing, one with the 6 Mpixel SX and the other with the 8 Mpixel QSI. It would be interesting to see what others make of this comparison. I know one has dust spots - I used old flat frames, but that is not the point.

[gorann]

Woow, I just found that my old thread on CMOS vs CCD (where the advice finally led me to buy a CCD, ending at an ATIK 460 with a Sony CCD) has suddenly woken up, been revitalized and taken into a new exciting direction: Sony vs Kodak CCDs. Very interesting reading and very pleasant viewing.  Maybe someone fortunate enough to have both a Kodak 8300 and a Sony ICX694 based camera would take the trouble to do a comparison using the same scope and similar processing on the same object? When I finally decided on the Sony, one reason was that the comments I has read indicated that there may be more variability in the quality of the Kodak chip (many loved it but some were quite disappointed about noise), while the Sony chip consistently received good reviews. Could it be that you need to be a bit lucky if you go for a Kodak?

[Horwig]

Buzz, love your pair of images. One thing it does prove is that your processing skills are much more repeatable than mine 

Huw

[buzz]

Woow, I just found that my old thread on CMOS vs CCD (where the advice finally led me to buy a CCD, ending at an ATIK 460 with a Sony CCD) has suddenly woken up, been revitalized and taken into a new exciting direction: Sony vs Kodak CCDs. Very interesting reading and very pleasant viewing.  Maybe someone fortunate enough to have both a Kodak 8300 and a Sony ICX694 based camera would take the trouble to do a comparison using the same scope and similar processing on the same object? When I finally decided on the Sony, one reason was that the comments I has read indicated that there may be more variability in the quality of the Kodak chip (many loved it but some were quite disappointed about noise), while the Sony chip consistently received good reviews. Could it be that you need to be a bit lucky if you go for a Kodak?

I have had problems with both sensors! Both have had banding issues on the bias that would not calibrate out. Both were fixed with firmware updates. It is as much about the implementation, as the chip itself. I have had two different cameras with the Kodak chip, the SXVR H18 and the QSI683. The H18 is now discontinued; practically it had a design flaw that housed the shutter immediately above the sensor. The swiping action of the shutter deposited dust over the sensor and it needed regular cleaning. The QSI houses the shutter externally, and has shutter design that does not contact moving parts.

My pair of images were put up to kinda show that the chip is only one step on the path to the final image and lots of things can occur to mess things up. The upper delicate one was with the QSI683 chip, using SGP, PHD2 and PixInsight on a FLT98 scope. I need to check but I think the lower one was with the Sony, the FLT132, but using Maxim for acquisition, guiding and processing.

The good thing is that sensor performance is measurable with scientifc method, so subjective comparisons are just the final confirmation. When I get the opportunity, I'll pitch the Fuji vs the Kodak. I just need to remind myself of the process to calculate the parameters.

[perfrej]

This kind of comparison is almost impossible without doing everything identically and on the same night, preferably in parallel...

I find the Kodak chips decent enough to allow calibration of the problems, and I want the real estate. Here's a rather not so well processed QSi-683 version based on 15-minute subs from the TEC-140 on my GM2000HPS in Provence....

/per

[juliangeorgeshaw]

However Julian did it, his Squid is the best I've seen.

Collaboration, Sir, your OIII with our widefeild HaLRGB?

lly

Olly I would be hugely flattered if you incorporated my data in your and Yves' fantastic image. Just let me know the best way to get it to you. (I am in NZ now with my mighty Baby Q which is best thing I ever bought which is thanks to your recommendation).