Jupiter - 30 October 2022

[geoflewis]

16 minutes ago, Magnum said:

Oh that's a good chart, yes the 2x does seem to have a negligible change doesn't it.

OK, so I would go straight for trying the new Baader at 1.3x, but I would put the ADC and or  filter wheel in front of the Baader 1.3x . if you keep the ADC in the middle like you have it now the spacing will be too much for the Baader and make it act at something like 2.5x. You will need to have a play around while previewing Jupiter, a quick and rough way to check is to adjust your ROI box in the capture program if given ROI is just touching the edges of the planet then you know thats how big it is. You can set any custom sized ROI in sharpcap. I like Jupiter to be around   500 pixels at the moment.

Lee

Thanks Lee, that’s the configuration I was thinking to use . I’m going to try the Baader lens on its own, screwed into the nosepiece of the camera (I’ve already confirmed that it fits ok). So the configuration will be scope -> ADC -> FW -> Barlow lens -> camera, but I’ll be experimenting as you suggest. I use FireCapture for capture, but can set the ROI in the same way you describe to measure the planet - I can’t believe that I never thought to do that previously 🙄. I’ll then reset the ROI with a bit to spare around Jupiter for the capture itself.

[Magnum]

Just now, geoflewis said:

Thanks Lee, that’s the configuration I was thinking to use . I’m going to try the Baader lens on its own, screwed into the nosepiece of the camera (I’ve already confirmed that it fits ok). So the configuration will be scope -> ADC -> FW -> Barlow lens -> camera, but I’ll be experimenting as you suggest. I use FireCapture for capture, but can set the ROI in the same way you describe to measure the planet - I can’t believe that I never thought to do that previously 🙄. I’ll then reset the ROI with a bit to spare around Jupiter for the capture itself.

I just added a bit more to my last reply while you were typing yours

[geoflewis]

21 minutes ago, Magnum said:

I just added a bit more to my last reply while you were typing yours

Just read the extra comments  - all excellent info, thanks

[astroavani]

A beautiful result, geoflewis friend!
Interesting that I only get good results using the C14 at f/22 with the PM 2X. I've tried many times in bad seeing night to use native d/f but the results were not good.
I think it must be related to sampling and ASI 290 which is very close to the 462 you use. As it has a very small pixel, it seems that a larger image makes the final result more pleasant.

[geoflewis]

6 hours ago, astroavani said:

A beautiful result, geoflewis friend!
Interesting that I only get good results using the C14 at f/22 with the PM 2X. I've tried many times in bad seeing night to use native d/f but the results were not good.
I think it must be related to sampling and ASI 290 which is very close to the 462 you use. As it has a very small pixel, it seems that a larger image makes the final result more pleasant.

Thanks Avani, it's very good to hear from you. The topic of correct sampling is indeed very interesting; I'll add you to the growing list of planetary imagers who find that sampling well below (ie. much finer) the theoretical numbers as calculated by @vlaiv, just works. There must be something in the maths theory that just isn't correct as all the best planetary imagers get their best results imaging at 0.1"/px or lower, where the maths is saying optimum for C14 is ~0.14". It's a huge difference in sampling rate and I'm at a loss to fathom why the theory is not supported by evidence in the field.

[vlaiv]

52 minutes ago, geoflewis said:

Thanks Avani, it's very good to hear from you. The topic of correct sampling is indeed very interesting; I'll add you to the growing list of planetary imagers who find that sampling well below (ie. much finer) the theoretical numbers as calculated by @vlaiv, just works. There must be something in the maths theory that just isn't correct as all the best planetary imagers get their best results imaging at 0.1"/px or lower, where the maths is saying optimum for C14 is ~0.14". It's a huge difference in sampling rate and I'm at a loss to fathom why the theory is not supported by evidence in the field.

Oversampling won't produce less detail / worse image as far as detail is concerned.

Issue with oversampling is as follows:

As you spread the light more than you need to - you get worse SNR per sub and you need to increase either number of subs or sub length in order to compensate.

Problem with either of those is that we are doing lucky imaging.

- longer subs means that one won't freeze the seeing, and only people with excellent seeing can afford to increase sub duration to compensate

- more subs / longer imaging runs - again, here we are at a loss. SNR improves with square of captured subs, yet number of usable subs grows linearly (there will be certain percent of good subs regardless if you image for 4 minutes of 8 minutes, so number of good subs grows linearly with time).

In any case - people with excellent skies won't feel those issues as much as the rest - but oversampling leads to poorer SNR in ending image (which in turn means less room for sharpening).

As to why @astroavani gets better results with over sampling versus regular sampling - I can't say, but I can point out that capture is only one variable before we get to final result.

There is stacking and sharpening to consider, and it might be due to either of these two that causes regular sampling to present poorer results.

In any case - if image is over sampled - that is easily identified in the image. High frequency components will be missing and there is simply no cheating of the physics - one cannot capture what is not there.

 

[geoflewis]

48 minutes ago, vlaiv said:

In any case - if image is over sampled - that is easily identified in the image. High frequency components will be missing and there is simply no cheating of the physics - one cannot capture what is not there.

Thanks Vlaiv, I am not a physicist, so can't challenge the theory on any basis other than experience, but I still contend that there must be something going on. We are not talking about hack imagers, but the world's best, so IMHO there must be a 'real' reason why they chose to sample at F22, F24, etc., with small pixel cameras like the 290 and 462 sensors. Look at the image from 5 Nov and more so the comments by Christopher Go.

http://astro.christone.net/jupiter/

He clearly considers it an extremely good image displaying very fine detail, so I am curious what 'high frequency components' he may have lost by imaging at F24  rather than F12-F14 that you contend is optimum for his configuration, which is essentialy the same as mine and Avani's?

Please know that I'm not just rejecting your analysis, but I can't for the life of me understand why the field evidence of these very experienced, talented imagers doesn't tie with the theory....

Edited November 7, 2022 by geoflewis

[vlaiv]

11 minutes ago, geoflewis said:

He clearly considers it an extremely good image displaying very fine detail, so I am curious what 'high frequency components' he may have lost by imaging at F24  rather than F12-F14 that you contend is optimum for his configuration, which is essentialy the same as mine and Avani's?

Here is comparison that should be clear enough:

First is part of the image that you pointed to by Christopher Go

Second is screen shot shows what sort of image can be recorded at that scale if all frequencies of the image are utilized.

You can clearly see the level of detail present in bottom image that is simply missing in top image. This points to the fact that above image by Christoper Go simply does not contain detail for presented resolution.

We can then go on to analyze frequency response between two images and to see what happens when we remove high frequency components from the bottom image.

Here are again two crops side by side (256 x 256 for easy FFT) and their respective frequency responses.

Left frequency response is concentrated in the center - nothing to the sides, while right one utilizes whole frequency space for this image.

We can now remove high frequency components for both images and restore images from their FFT representations to see what will happen.

Ok, now we have just very center (low frequency) part of both images - and left image is almost the same / unchanged, but right one is very different.

It is missing that high fidelity and now it looks as blurred as left image.

(vertical and horizontal banding is artifact of sharp cutoff - I just set higher frequencies to 0 - but in nature it is gradual drop towards the zero so we don't see that vertical / horizontal ripples).

In the end - we can clearly see - where there is telescope produced frequency cut off point in the left FFT:

ImageJ reports that this is at 4.14 pixels per cycle. Nyquist sampling is at 2 pixel per cycle, so this is 4.14 / 2 = x2.07 over sampled.

I can't say if above image by Christopher Go would be worse if sampled at proper resolution, but what I can tell you is that it is over sampled and it can be clearly seen both in image softness and in frequency domain as demonstrated.

Not only that it is over sampled by factor of x2 - it looks that due to seeing and processing - actual sampling rate is closer to x4 lower than what was used (majority of signal is concentrated in that central 1/4 circle)

If we reduce it to 25% of its size and compare to much sharper reference at the same size - we get comparable detail level and sharpness (except for color saturation - one conclude that same instrument / conditions were used for these two images)

 

[geoflewis]

@vlaiv Thanks again for the additional analysis and supporting images. It is beyond my ability to understand what you are saying, sorry. Well I understand what you are saying, but not the maths that 'proves' your analysis. It still leaves me questioning why the expreience of Chris Go, Damian Peach, et al, differs from your mathematical analysis. I think that they fine tuned their approach based on experience, so when experience/observation challenges the maths, I have to question the maths - it's almost like Ptolemaic v Copernican arguments. OK I know it's not, but hopefully you will understand my reasoning for doubting the maths....

[vlaiv]

8 minutes ago, geoflewis said:

@vlaiv Thanks again for the additional analysis and supporting images. It is beyond my ability to understand what you are saying, sorry. Well I understand what you are saying, but not the maths that 'proves' your analysis. It still leaves me questioning why the expreience of Chris Go, Damian Peach, et al, differs from your mathematical analysis. I think that they fine tuned their approach based on experience, so when experience/observation challenges the maths, I have to question the maths - it's almost like Ptolemaic v Copernican arguments. OK I know it's not, but hopefully you will understand my reasoning for doubting the maths....

I don't think that you need to look at math at all - just look at first two images and you will clearly see the difference.

Left is level of detail captured by over sampling, right is level of detail that is possible at that resolution. When you properly sample - you capture level of detail that is possible at given resolution.

10 minutes ago, geoflewis said:

I think that they fine tuned their approach based on experience, so when experience/observation challenges the maths,

I'm failing to see how does their experience challenge the math/physics? Is their image sharp at higher resolution than is supported by physics of light? No, or at least I'm failing to see it.

[geoflewis]

3 minutes ago, vlaiv said:

I don't think that you need to look at math at all - just look at first two images and you will clearly see the difference.

Left is level of detail captured by over sampling, right is level of detail that is possible at that resolution. When you properly sample - you capture level of detail that is possible at given resolution.

Who produced the image on the right and what equipment were they using please?

[Mandy D]

3 minutes ago, vlaiv said:

I don't think that you need to look at math at all - just look at first two images and you will clearly see the difference.

Left is level of detail captured by over sampling, right is level of detail that is possible at that resolution. When you properly sample - you capture level of detail that is possible at given resolution.

I'm failing to see how does their experience challenge the math/physics? Is their image sharp at higher resolution than is supported by physics of light? No, or at least I'm failing to see it.

Is a "free-energy" or "over-unity" machine possible? No! Physics does not allow you to get what is not there. Like you say, trust the physics.

[vlaiv]

Just now, geoflewis said:

Who produced the image on the right and what equipment were they using please?

It is HST image that I just googled as high resolution Jupiter image.

It is much larger (and still sharp) than one I used for comparison.

Here is original link:

https://stsci-opo.org/STScI-01FM5S4SWQQ9M5A15BHBMG0NW4.png

[geoflewis]

2 minutes ago, vlaiv said:

It is HST image that I just googled as high resolution Jupiter image.

It is much larger (and still sharp) than one I used for comparison.

Here is original link:

https://stsci-opo.org/STScI-01FM5S4SWQQ9M5A15BHBMG0NW4.png

Ok, now you really lost me. How is comparing HST image with ground based image relevant to this discussion?

[drjolo]

9 minutes ago, vlaiv said:

I don't think that you need to look at math at all - just look at first two images and you will clearly see the difference.

I think this comparison may be a bit misleading. The idea being discussed here is image quality losses due to the oversampling, and the example shows something different.

Perfect example would be to see the same image 2x and 4x sampled, possibly after processing and see the difference caused by the lower SNR in this second one. 

[vlaiv]

Just now, geoflewis said:

Ok, now you really lost me. How is comparing HST image with ground based image relevant to this discussion?

Very simple.

Don't think about comparison image as being HST image - think of it as - the most detail that can be recorded at certain sampling rate.

Fact that I used HST makes sure of that (as it is capable of recording at even higher sampling rates).

I wanted to show what image at certain sampling rate looks like - when all the detail / frequencies are present. It looks sharp as we would expect. We can then compare that image with images produced by amateur equipment to see if images with amateur equipment also capture all the details that is possible at that sampling rate.

It turns out that they don't - which means that given  sampling rate is not adequate for amateur equipment. People are over sampling.

That is the point of comparison and using of HST image (just to ensure we are not instrument limited at given resolution).

[vlaiv]

1 minute ago, drjolo said:

I think this comparison may be a bit misleading. The idea being discussed here is image quality losses due to the oversampling, and the example shows something different.

Why do you think it is misleading?

I just wanted to show that there is indeed over sampling.

But ok, let me do another set of images that will show what you want - loss of detail due to different sampling.

 

[drjolo]

5 minutes ago, vlaiv said:

Why do you think it is misleading?

Oversampling is different sampling for the same image given by the optics. 

In the example we have different images given by different optics. 

[vlaiv]

Left image is original, right image has been scaled down to 50% of original (thus "loosing detail") and then enlarged back to 100%

left image is original, right has been reduced to 25% (thus loosing detail) and then enlarged back to 100%

When reduced to 50% and scaled up back - there is no difference, but when we reduce it to 25% (1/4 of sampling rate) - we can start to notice softening, so some detail has been lost.

Same thing for HST image:

left at full resolution - right scaled down to 50% and scaled back up to 100% (We can see some softening at this stage - feature edges are not as sharp in right image).

this is 100% vs 25% -> 100%

Softening is more than obvious now.

[vlaiv]

2 minutes ago, drjolo said:

Oversampling is different sampling for the same image given by the optics. 

In the example we have different images given by different optics. 

Yes, but we are exploring what sort of detail can be recorded at given sampling rate.

We assume that both instruments can produce level of detail that requires given sampling rate (if not then it is indeed over sampling).

[drjolo]

Okay, I thought that discussion is about image quality loss due to the oversampling of the same image - like we have image from C14 at f/22 and f/15 for example. 

I can understand oversampled image has lower SNR compared to correctly sampled, but the detail loss caused by this is not so easy to model, and that is not shown in the examples you provided. 

Edited November 7, 2022 by drjolo

[vlaiv]

3 minutes ago, drjolo said:

Okay, I thought that discussion is about image quality loss due to the oversampling of the same image - like we have image from C14 at f/22 and f/15 for example. 

I can understand oversampled image has lower SNR compared to correctly sampled, but the detail loss caused by this is not so easy to model, and that is not shown in the examples you provided. 

There is no detail loss due to over sampling - at least not one explicit one.

You are quite right - it is far more difficult to model detail loss due to SNR loss and inability to sharpen as much as one would like.

I don't think there will actually be detail loss - only more noise at the same level of detail as no detail is really lost if one over samples - it is just that it can't be shown without showing the noise as well.

[geoflewis]

5 minutes ago, vlaiv said:

There is no detail loss due to over sampling - at least not one explicit one.

You are quite right - it is far more difficult to model detail loss due to SNR loss and inability to sharpen as much as one would like.

I don't think there will actually be detail loss - only more noise at the same level of detail as no detail is really lost if one over samples - it is just that it can't be shown without showing the noise as well.

Ok , so if there is no detail loss due to oversampling, then that suggests that I'm definitely better to 'risk' oversampling versus undersampling, which is what I think I may be doing without any amplification in my image train - I think we calculated I am currently at <=F12 or 0.14"/px. There may be worse SNR at F22 v F12, but will I actually see it, especialy if in excellent seeig I can stack >10k frames? However, maybe at F12 I am not fully utilising the optics and I can't ever recover what I didn't capture - is my logic still wrong?

[vlaiv]

5 minutes ago, geoflewis said:

Ok , so if there is no detail loss due to oversampling, then that suggests that I'm definitely better to 'risk' oversampling versus undersampling, which is what I think I may be doing without any amplification in my image train - I think we calculated I am currently at <=F12 or 0.14"/px. There may be worse SNR at F22 v F12, but will I actually see it, especialy if in excellent seeig I can stack >10k frames? However, maybe at F12 I am not fully utilising the optics and I can't ever recover what I didn't capture - is my logic still wrong?

Your logic is not wrong - if you want to capture all the detail that is potentially available - it is better to over sample than it is to under sample.

I just don't see why would you aim for F/22 or F/24 if you can sample optimally at F/14.5 given your pixel size. If you use barlow - then you can "dial in" magnification by changing barlow to sensor distance.

Maybe best approach is to just give it a try one way and the other and just choose what you feel is giving you best results. Most others are using that sort of an approach and don't really care that their images are over sampled, if it is easier for them to work that way.

 

[geoflewis]

Just now, vlaiv said:

Your logic is not wrong - if you want to capture all the detail that is potentially available - it is better to over sample than it is to under sample.

I just don't see why would you aim for F/22 or F/24 if you can sample optimally at F/14.5 given your pixel size. If you use barlow - then you can "dial in" magnification by changing barlow to sensor distance.

Maybe best approach is to just give it a try one way and the other and just choose what you feel is giving you best results. Most others are using that sort of an approach and don't really care that their images are over sampled, if it is easier for them to work that way.

Thanks Vlaiv, I think you may have hit the nail on the head. It may just be that it is easier to dial the focus with a bigger 'oversampled' image on screen. I certainly found this to be true when Mars was last close to us in 2020. I was operating at ~F22 for Mars in 2020, but only F12 for Jupiter and Saturn last year, but of course Jupiter and Saturn were much lower down and imaging them at all was very challenging. So far this year I have imaged all three at F12, but I am going to experiment with different amplification to see what differences I get, especially with Mars smaller diameter this year, but up at ~60° elevation.

Thanks for all your advice on this complex (to me anyway) topic.