What is the sweet spot for an ASI2600 camera?

[osbourne one-nil]

I've rephrased my original question into something more specific  - what would you say the sweet spot is for an ASI2600MC Pro camera, in terms of scope? Running the CCD suitability guide in Astronomy.Tools, it would look to be something around the 500mm focal length mark, which results in around a 1.5" per pixel. Is that the sort of thinking behind this?

If so, the best corrected/largest aperture resulting in that "/pixel ratio would be something like the Esprit 100. Being a triplet, would that give me a noticeably crisper image than my SD81 doublet? Is the aperture and 800mm focal length of my R200SS being wasted by the "limitation" of my camera's small pixels?

Basically, what's the best combination, assuming the camera is the starting point on this occasion? 

Edited February 19, 2023 by osbourne one-nil

[Adam J]

On 19/02/2023 at 09:48, osbourne one-nil said:

I've rephrased my original question into something more specific  - what would you say the sweet spot is for an ASI2600MC Pro camera, in terms of scope? Running the CCD suitability guide in Astronomy.Tools, it would look to be something around the 500mm focal length mark, which results in around a 1.5" per pixel. Is that the sort of thinking behind this?

If so, the best corrected/largest aperture resulting in that "/pixel ratio would be something like the Esprit 100. Being a triplet, would that give me a noticeably crisper image than my SD81 doublet? Is the aperture and 800mm focal length of my R200SS being wasted by the "limitation" of my camera's small pixels?

Basically, what's the best combination, assuming the camera is the starting point on this occasion? 

If you can get hold of a spot diagram then spot size is also an important factor in understanding resolution. Of the spot size is significantly larger than the pixels then you won't achieve the theoretical resolution even in perfect seeing.  The other thing with the CCD suitability guide is that it makes no account of OSC sampling due to the Bayer matrix and so only really works at all for mono cameras. 

Adam

[scotty38]

Personally I'd focus (sorry) on the other side and ask if the combination gives you the FOV you'd like. Unless I have the wrong end of the stick there's no point having a technically perfect setup if you don't like the view you get. I'd also bet that conditions are more likely to kill any perfection that is there anyway.

I've seen all sorts of camera/scope combinations produce some wicked images but similarly I've scoured Astrobin for my camera/scopes of desire and seen some shockers 🙂

[rnobleeddy]

Is the idea of pixels being "too small" still relevant?

For anyone who images with a range of scopes, isn't it better to get the smallest pixels and then just bin? That way, you can avoid undersampling for widefield.

My logic is:

- read noise on the modern cameras is low, so software binning isn't really a problem.

- Fractional binning allows you get back to the correct sampling rate (last time this came up, there was a suggestion that is was unproven, but empirical results have been good)

- The range of pixel sizes available in modern CMOS cameras is relatively small, and the current generation is so much better than CCDs or earlier CMOS cameras (higher QE, back illumination etc) that the impact of using one of these is far bigger than the gains in avoiding binning

 

 

 

[osbourne one-nil]

Thanks all - I'm finding it difficult to ask the right question! 

I shot M81 a week or so ago, with everything the same (number of subs, gain, total integration time etc) other than the telescopes. One was shot with my 81mm f/7.7 apo and one with my 8" f/4 Newtonian. Once the images were scaled to match, and I did this both ways (scaled the refractor image up and compared and scaled the Newt's image down and compared), there was really not a lot in it. Ok, they were taken on different nights, but conditions were pretty similar. 

 

I'm surprised and far slower, far smaller refractor can even get close to a much faster and much larger Newtonian. This is what led me to ponder if to get any benefit from the Newt, I'd need larger pixels....or finally get my head around binning. The other option is, of course, to just have one scope, never have to fiddle on with swapping things over and collimation and have an easy life!

[osbourne one-nil]

On 24/02/2023 at 11:16, Adam J said:

If you can get hold of a spot diagram then spot size is also an important factor in understanding resolution. Of the spot size is significantly larger than the pixels then you won't achieve the theoretical resolution even in perfect seeing.  The other thing with the CCD suitability guide is that it makes no account of OSC sampling due to the Bayer matrix and so only really works at all for mono cameras. 

Adam

This is where my brain starts to melt. Actually, my brain does that way before this stage. 

 

Here's the best spot diagram I can find, and I have both the Extender PH (which is the top one) and the Coma Corrector 3 (3rd one down) and I shoot with an APS-C sized sensor. Hmmmmm!

[Adam J]

4 hours ago, osbourne one-nil said:

This is where my brain starts to melt. Actually, my brain does that way before this stage. 

 

Here's the best spot diagram I can find, and I have both the Extender PH (which is the top one) and the Coma Corrector 3 (3rd one down) and I shoot with an APS-C sized sensor. Hmmmmm!

Cant really tell from that but it is highly unlikely that you will produce spot sizes below 6um from a F4 Newtonian from what I have seen in other scopes. Quad flat field refractors can get down to 3-4um but fast Newtonians just don't perform as well as that due to the central obstruction and compromises within the coma corrector design. So basically what I am saying is that you are not going to be resolving images to the pixel level with that scope and a 3.75um pixel camera and so the camera pixel size is not going to be the limiting factor.

In the end the 2600mc is going to work fine with either of those scopes certainly the Bayer Matrix is not going to be limiting you, your over thinking it...just get the camera. 

Adam

[Clarkey]

22 hours ago, Adam J said:

In the end the 2600mc is going to work fine with either of those scopes certainly the Bayer Matrix is not going to be limiting you, your over thinking it...just get the camera. 

Definitely.  I have the Rising Cam version of the 2600 and use it at a number of FL's. Yes, there is a theoretical sweet spot, but there are a lot of other things that will impact the quality of your images long before the pixel scale starts to come into play.

[osbourne one-nil]

23 hours ago, Adam J said:

 your over thinking it...just get the camera. 

Adam

I've already got it! I was just having a think about my setup. I'll stick with what I have!

[osbourne one-nil]

58 minutes ago, Clarkey said:

Definitely.  I have the Rising Cam version of the 2600 and use it at a number of FL's. Yes, there is a theoretical sweet spot, but there are a lot of other things that will impact the quality of your images long before the pixel scale starts to come into play.

Excellent - thanks. I was in Penrith last night....how many lights does Gilwilly have these days?!

[Clarkey]

1 hour ago, osbourne one-nil said:

.how many lights does Gilwilly have these days?!

I know, it's like Blackpool. Worst offender is the golf driving range. Apparently, golfers need the sky illuminated!

 

[Anthonyexmouth]

2 hours ago, Clarkey said:

Definitely.  I have the Rising Cam version of the 2600 and use it at a number of FL's. Yes, there is a theoretical sweet spot, but there are a lot of other things that will impact the quality of your images long before the pixel scale starts to come into play.

How are you getting on with the rising cam? I've always had ZWO cameras, mainly because FLO is just a few miles from me and can cycle to pick up kit. But seeing as the Altair version and RC versions are so dramatically cheaper I can't justify what is £800 extra  for the tilt plate on the ZWO version and am I right in thinking the Rising Cam is the same as the AA one?

[ollypenrice]

The RASA 8 is a dream with this camera.  FL is 400mm but small scale resolution is good. The RASA is not diffraction limited but it has so much more aperture than most 400mm FL scopes that it still has room to out-resolve even the diffraction-limited smaller ones. You also get so much signal in the time that you have more room for sharpening in post-processing.

As for the effects on resolution of the Bayer matrix, I would say they are insignificant.  You'll hear people argue that an RGGB set of pixels resolves as one large pixel but this is not true. The debayering software interpolates the missing information and, yes, this is a form of highly educated guessing but it is remarkably effective in the real world. In the days of CCD, I preferred mono over OSC but not because of resolution. I compared my mono and OSC Atik 4000s regularly and never found anything to choose on resolution.

I'm now using two variants of this camera, one in the RASA and another in a Samyang 135. It is, in both cases, a cracking instrument.

Olly

[Clarkey]

4 minutes ago, Anthonyexmouth said:

How are you getting on with the rising cam? I've always had ZWO cameras, mainly because FLO is just a few miles from me and can cycle to pick up kit. But seeing as the Altair version and RC versions are so dramatically cheaper I can't justify what is £800 extra  for the tilt plate on the ZWO version and am I right in thinking the Rising Cam is the same as the AA one?

I'm not sure about it being the same as the AA one - possibly the same internals, but externally different. Initially I had some problems, but it turned out to be down to a couple of things. Firstly, I think I suffered from some dew on the sensor glass, which I didn't realise at the time as it had never been an issue with the ASI1600. Once I made sure the internal dew heater was on (and I added the ZWO stick-on one for good measure) all was good. I also had a bit of problem with flats which I think is down to some slight sensor tilt. It's not bad, so I am just living with it. Also, I am not used to OSC processing - mono is much easier to get a good colour balance. The only real negative that I have found is that there is some dust inside the camera which I cannot remove without dismantling. It calibrates out OK with flats, but it is a bit annoying. Overall, I think the camera is very good for the money compared with the UK offerings. There is always the risk of something going wrong and then you have to deal with the minefield of import / export etc. However, as it was not far off half the price of the ZWO offering I felt it was worth the risk. I guess I was lucky as I paid no import duty or VAT so it was a flat rate price too. 

[Adam J]

2 hours ago, ollypenrice said:

The RASA 8 is a dream with this camera.  FL is 400mm but small scale resolution is good. The RASA is not diffraction limited but it has so much more aperture than most 400mm FL scopes that it still has room to out-resolve even the diffraction-limited smaller ones. You also get so much signal in the time that you have more room for sharpening in post-processing.

As for the effects on resolution of the Bayer matrix, I would say they are insignificant.  You'll hear people argue that an RGGB set of pixels resolves as one large pixel but this is not true. The debayering software interpolates the missing information and, yes, this is a form of highly educated guessing but it is remarkably effective in the real world. In the days of CCD, I preferred mono over OSC but not because of resolution. I compared my mono and OSC Atik 4000s regularly and never found anything to choose on resolution.

I'm now using two variants of this camera, one in the RASA and another in a Samyang 135. It is, in both cases, a cracking instrument.

Olly

I think it would be an interesting experiment to put a 400mm triplet refactor  mono camera combination next to a RASA 8 with a OSC and the same pixel size and perform a comparison against a bright detailed target. Its very difficult looking at astrobin as as there is such a wider variation in processing ability. 

Ultimately Bayer interpolation is little better if at all than any up sampling algorithm as might be applied to any image in Photoshop. I would also point out that only the Red and Blue pixels resolve to a single larger pixel, the green is different as it has higher sampling. In terms of OCS broadband RGB imaging the algorithm will perform better as it uses the green channel to improve B and R as a kind of luminescence.

But if using a duel band filter and extracting the Ha and OIII channels that methodology does not apply, in fact debeyering of that type doesn't occur at all and cant without mixing the Ha and OIII channel data (don't want to do that). In this instance you will and do end up with Ha super pixels that are then up-scaled to the original resolution using a bi-linear interpolation, OIII being a little more interesting than this.

Its is in the use of duel band filters that you see the most significant reduction in resolution vs mono narrow-band. I suspect that back in your Atik 4000 imaging days duel band filters where not available, but in the end without trying a duel band on the ATIK4000s you are unlikely to see as significant a reduction in resolution. 

Using a RASA OSC vs mono is unlikely to make a big difference to resolution based on specification, that being because resolved detail will not be limited based on pixel size. The reason for this is that the spot size is much larger than the 3.75um pixels on the 2600, to the point that its larger than even the RGGB super pixel grouping.  The same reasoning is behind the false belief that the ASI183mm or the ASI294mm pro in its bin 1 mode will gain you resolution. I have looked at many spot diagrams and not a single one when the spot size is at 2.3um or less. Top end performance is normally 3-4um more often 6um significantly larger than 2.3um pixels. 

Below is the spot diagram for the RASA 8, while in green its a lovely 3.6um, I will interpolate that for Ha (between the 600-700nm spots), it will have at least have doubled to 7um spots and thats on axis, that is about the same size as a Ha super pixel on the 2600mc pro. All in all the leaves you with a resolving power in Ha of about 3.6arc seconds. To put that into context the Daws limit is about 0.6 arc-seconds for an unobstructed aperture.  The interesting thing is that a 80mm aperture quad refactor with a mono camera would be expected to perform at closer to 2 arc-seconds per pixel. 

In a nut shell I am one of the people that you will hear saying the RGGB resolves less than Mono, but as above that is also dependent on how you use it, the scope its paired with and the filter type being used, what it isn't is a blanket untrue statement. 

However, all of the above is a diversion and in the end we have the cameras available to use with a limited veraity of sensor sizes and pixel sizes. Hence, my advice to the OP stands, the 2600mc is a good camera so although its interesting to discuss the details above from an intellectual perspective I think you need to stop overthinking it as viable alternatives to the 2600mc are thin on the ground and I have seen excellent images across a wide range of focal length and diverse setups.  Tine taken thinking about this stuff is imaging time lost. 

Adam

Edited February 28, 2023 by Adam J

[osbourne one-nil]

59 minutes ago, Adam J said:

 Time taken thinking about this stuff is imaging time lost. 

Time living in Cumbria is imaging time lost!

[ollypenrice]

1 hour ago, Adam J said:

I think it would be an interesting experiment to put a 400mm triplet refactor  mono camera combination next to a RASA 8 with a OSC and the same pixel size and perform a comparison against a bright detailed target. Its very difficult looking at astrobin as as there is such a wider variation in processing ability. 

Ultimately Bayer interpolation is little better if at all than any up sampling algorithm as might be applied to any image in Photoshop. I would also point out that only the Red and Blue pixels resolve to a single larger pixel, the green is different as it has higher sampling. In terms of OCS broadband RGB imaging the algorithm will perform better as it uses the green channel to improve B and R as a kind of luminescence.

But if using a duel band filter and extracting the Ha and OIII channels that methodology does not apply, in fact debeyering of that type doesn't occur at all and cant without mixing the Ha and OIII channel data (don't want to do that). In this instance you will and do end up with Ha super pixels that are then up-scaled to the original resolution using a bi-linear interpolation, OIII being a little more interesting than this.

Its is in the use of duel band filters that you see the most significant reduction in resolution vs mono narrow-band. I suspect that back in your Atik 4000 imaging days duel band filters where not available, but in the end without trying a duel band on the ATIK4000s you are unlikely to see as significant a reduction in resolution. 

Using a RASA OSC vs mono is unlikely to make a big difference to resolution based on specification, that being because resolved detail will not be limited based on pixel size. The reason for this is that the spot size is much larger than the 3.75um pixels on the 2600, to the point that its larger than even the RGGB super pixel grouping.  The same reasoning is behind the false belief that the ASI183mm or the ASI294mm pro in its bin 1 mode will gain you resolution. I have looked at many spot diagrams and not a single one when the spot size is at 2.3um or less. Top end performance is normally 3-4um more often 6um significantly larger than 2.3um pixels. 

Below is the spot diagram for the RASA 8, while in green its a lovely 3.6um, I will interpolate that for Ha (between the 600-700nm spots), it will have at least have doubled to 7um spots and thats on axis, that is about the same size as a Ha super pixel on the 2600mc pro. All in all the leaves you with a resolving power in Ha of about 3.6arc seconds. To put that into context the Daws limit is about 0.6 arc-seconds for an unobstructed aperture.  The interesting thing is that a 80mm aperture quad refactor with a mono camera would be expected to perform at closer to 2 arc-seconds per pixel. 

In a nut shell I am one of the people that you will hear saying the RGGB resolves less than Mono, but as above that is also dependent on how you use it, the scope its paired with and the filter type being used, what it isn't is a blanket untrue statement. 

However, all of the above is a diversion and in the end we have the cameras available to use with a limited veraity of sensor sizes and pixel sizes. Hence, my advice to the OP stands, the 2600mc is a good camera so although its interesting to discuss the details above from an intellectual perspective I think you need to stop overthinking it as viable alternatives to the 2600mc are thin on the ground and I have seen excellent images across a wide range of focal length and diverse setups.  Tine taken thinking about this stuff is imaging time lost. 

Adam

I've never used NB filters with an OSC camera and hadn't considered this scenario. Point taken.

What I have done is a side by side comparison of an OSC versus mono image of the Trapezium region of M42 in the same model of camera (bar the Bayer matrix difference) and the same optics. This was for a magazine article. Try as I might, I could discern no difference in perceptible final resolution of detail.

I heartily agree that the best thing to do in this game is get on with it!

lly

[gorann]

After CMOS took over I get the feeling that astrophotographers are less occupied with pixel scale. I see experienced astrophotographers using these new 3.76 µm pixel cameras on any focal length between 135 mm (like Olly, me and others) and 3.5 meters (like John Hayes with his 20" scope in Chile). So, my advice is just to get on with it. You can experiment with software binning after you taken your images. ASI 2600MC is a great camera (I have three) but there are other less expensive and good alternatives with the same sensor (I also have an Omegon variant of it that works great).

Edited February 28, 2023 by gorann

[Lumos]

I also use the ASI 2600MC with the RASA 8 and love the results. However I also use the 2600MC and the 2600MM with the Esprit 150 and love those results too. As the advice above said, choose focal lengths that motivate you and let the pixel scale take care of itself.

[tomato]

I’m doing the same as @Lumos, and agree 100% with the conclusions.

[GalaxyGael]

There isn't one really in practice, it's a darling sensor at the moment and one of the best in the OSC flavour, with something wide and fast or long integration to replace lrgb for some people. It's a camera used by all levels of experience and good enough for essentially all focal lengths, running the gamut between as captured undersampling and rank oversampling...and people can still process the data well in all cases.

File size for DSO lucky imaging might be an issue but looks like that's not an issues for you, so enjoy the imaging. I love mine, but also like my atik 460ex OSC too, and love it even more so with my 300mm 60 mm scope at 3 "/px. Lovely smooth bright images, with fov being the only real difference for me in reality between the two cameras and the weight difference (I like that tank of a cc'd camera and I have it so I use basically). The 2600 at aps c has the fov benefit.

Tilt plates better than the zwo one readily available, and the risingcam and omegon version are solid, as is player one version from what I have heard. Zwo is handy if you want to use asi air.

Get a bucket load of data regardless of your skies and you'll be hard pushed to find a different camera that will be better in a way that is stark, or at all, my 2c.

[Astrobug]

Im using the 2600mm on a Newton 200 f5  and a friend of mine on a c9 with a focal reducer. We did both m81 last winter. On the c9 your could still see some extra details. Under the belgium sky I would think the sweet spot on good nights is around 1300mm FL

Edited July 24, 2023 by Astrobug