LDN 1228 - more data made a big difference

[andrew s]

@vlaiv thanks for taking the trouble to do the calculations on the nebula colour.

While it may not be exact given your concerns it does (as you well know) demonstrate that colour can be validated given good calibrated data.

Regards Andrew 

[ollypenrice]

10 hours ago, gorann said:

Thanks Olly, but what are you looking at, in daytime on a bar stol?

😁 The clouds in Florida, I suspect! Ralf took this at the Florida star party. Believe it or not it can resolve two Jovian cloud belts and I could read car number plates in it at over 200 metres.

Bar stool? How dare you!  That's an over-engineered pier, I'll have you know, with unique anti-vibration top plate.

Olly

[gorann]

12 hours ago, vlaiv said:

Well, it's not gray

Supposedly - it's above shades if my measurements and math are correct.

Reference star used was:

TYC 4590-1433-1

at 20:54:22.679+78:07:05.94

Gaia DR2 quotes this star to be 5570K of effective temperature. Corresponding sRGB linear triplet is:

1.135763
0.972170
0.875711

Photometric measurement of star gives:

68281.670806
84433.395310
96729.032408

and measured nebulosity linear RGB values are:

18.910516739
14.435421944
6.550098896

Resulting nebulosity RGB values are:

3.14548
1.6621
0.593

However, I'm rather skeptical about these results for two reasons:

1. Data is not calibrated (at least flat field is not applied - this makes background removal very difficult and probably not precise)

2. Data is 16bit. There is 6 hours of data in 4 minute exposures. This is 90 subs stacked and that is 6.5 bits improvement over camera bit count which is around 14 at gain 100 - so it's 20 bits. 4 Bits of precision lost due to 16 bit format.

 

 

 

Thanks Vlaiv,

Then the dark nebulosity in my image is rather close to what it should look like when watched from Earth. Regarding flats so have I never had any good experience with those with colour cameras. Seem to mess up the color balance. There are long threads obout doing OSC flats and how the channels should be separated and calibrated first. I decided it is not for me as long as there are no or few dust bunnies in the system, Vignetting can be fixed in processing.

[gorann]

10 hours ago, Datalord said:

It was too tempting to let this one go unanswered. I think I have gotten myself into the mode of "f value has no meaning on photons". I have an 11" RASA f/2 and a 12" RC f/8, 620mm and 2400mm. The number of photons are purely a matter of aperture. The number of photons per pixel is a matter of the camera and the aperture and is ultimately the only factor that makes sense. A 1m mirror with a camera that has a resolution of 0.52"per pixel collects more photons per pixel than my 12" with a resolution of 0.52"per pixel. Notice how they will have the same resolution and if the cameras have the same number of pixels, they will have the same image.

Contrary, if your RASA8 has a camera with 0.3"per pixel vs a 71mm refractor rig with a 2.5"per pixel camera, you're not saving all that time. You will have a much higher resolution, but SNR won't be improved 12 times. However, make both have 2.5"per pixel and yours will have a 12 times higher SNR.

The way I calculated this was simply to compare the 8" RASA with 400 mm FL with a commonly used wide field scope like a 400 mm f/5 Apo, so with 80 mm aperture (like an Esprit 80). Both should of course have the same camera to be comparable. In that case the RASA will have (200 x 200)/(80x80) = 6.25 times more light collecting capacity, so 6 hours will correspond to 37.5 hours with the smaller scope.

[Datalord]

17 minutes ago, gorann said:

The way I calculated this was simply to compare the 8" RASA with 400 mm FL with a commonly used wide field scope like a 400 mm f/5 Apo, so with 80 mm aperture (like an Esprit 80). Both should of course have the same camera to be comparable. In that case the RASA will have (200 x 200)/(80x80) = 6.25 times more light collecting capacity, so 6 hours will correspond to 37.5 hours with the smaller scope.

Right, the aperture. I was more reacting to the use of f/ in the argument somewhere. I've seen it a hundred times and while it's a derivative of the real reason, it can completely skew the point, especially for anyone new to the hobby.

[vlaiv]

39 minutes ago, gorann said:

Then the dark nebulosity in my image is rather close to what it should look like when watched from Earth

Actually if everything is right, above color is how it should look in outer space not from earth.

I do have doubts if above measurement is accurate. For example I used 5570K (according to Gaia DR2) star as a reference.

It should have following RGB values:

1.135763
0.972170
0.875711

So, linear R should be something like 30% stronger than B. Data from your camera has much stronger B than R:

68281.670806
84433.395310
96729.032408

Linear B is in this case 41% stronger than R. This is something that usually does not happen with OSC cameras. OSC cameras usually have stronger R than B. Even IMX571 QE graphs hint that R should be more sensitive:

R has both higher peak and broader curve. So does G. In above reference 5570K star, G is stronger than B. In QE graph, G is clearly stronger than B, yet in measured star values - B is stronger than G?

I don't see how this could happen unless something is wrong with either reference star or my measurement or the data.

[gorann]

1 hour ago, vlaiv said:

Actually if everything is right, above color is how it should look in outer space not from earth.

I do have doubts if above measurement is accurate. For example I used 5570K (according to Gaia DR2) star as a reference.

It should have following RGB values:

1.135763
0.972170
0.875711

So, linear R should be something like 30% stronger than B. Data from your camera has much stronger B than R:

68281.670806
84433.395310
96729.032408

Linear B is in this case 41% stronger than R. This is something that usually does not happen with OSC cameras. OSC cameras usually have stronger R than B. Even IMX571 QE graphs hint that R should be more sensitive:

R has both higher peak and broader curve. So does G. In above reference 5570K star, G is stronger than B. In QE graph, G is clearly stronger than B, yet in measured star values - B is stronger than G?

I don't see how this could happen unless something is wrong with either reference star or my measurement or the data.

There is a built in UV/IR filter is the ASI2600 (as I remeber) so maybe it has an effect on what the IMX571 sees?

Edited August 29, 2020 by gorann

[vlaiv]

3 minutes ago, gorann said:

There is a built in UV/IR filter is the ASI2600 (as I remeber) so maybe it has an effect on what the IMX571 sees?

Most UV/IR filters for astronomical cameras should be more or less uniform, like for example 1.25" L filters:

They should not significantly reduce green and red part of the spectrum ...

[gorann]

The file I sent you was only debayered, aligned and integrated, all in PI, so there should probably not be any colour adjustment done (but who knows what magic PI is doing?)

[vlaiv]

13 minutes ago, gorann said:

The file I sent you was only debayered, aligned and integrated, all in PI, so there should probably not be any colour adjustment done (but who knows what magic PI is doing?)

I don't think that there was any color adjustment done in stacking process. It's probably my choice of reference star or something that I over looked. Although, it is kind of strange to have such high blue values for star that should have more red than blue.

[ollypenrice]

3 hours ago, gorann said:

The way I calculated this was simply to compare the 8" RASA with 400 mm FL with a commonly used wide field scope like a 400 mm f/5 Apo, so with 80 mm aperture (like an Esprit 80). Both should of course have the same camera to be comparable. In that case the RASA will have (200 x 200)/(80x80) = 6.25 times more light collecting capacity, so 6 hours will correspond to 37.5 hours with the smaller scope.

 

3 hours ago, Datalord said:

Right, the aperture. I was more reacting to the use of f/ in the argument somewhere. I've seen it a hundred times and while it's a derivative of the real reason, it can completely skew the point, especially for anyone new to the hobby.

Goran, there's nothing wrong with that calculation but, like Datalord, I think F ratio is a complicated way of referring to aperture. It's only sensible to compare the same focal lengths, in which case the only physical variable is aperture. As we've concluded on previous occasions the ratio we really want is area of aperture per area of pixel.

Olly

[andrew s]

20 minutes ago, ollypenrice said:

 

Goran, there's nothing wrong with that calculation but, like Datalord, I think F ratio is a complicated way of referring to aperture. It's only sensible to compare the same focal lengths, in which case the only physical variable is aperture. As we've concluded on previous occasions the ratio we really want is area of aperture per area of pixel.

Olly

I think after a long discussion on this "etendue" was the key measure. Simply it is the equivalent of the product of aperture (how many photons/s) and usable  field of view (sky coverage in arc secs).

Provided the scope/detector has the resolution required (which is almost always the case with modern CMOS cameras) this captures the rate of accumulation of information from the sky. It helps balance the multitude of factors ,e.g. aperture, useful telescope FOV, chip size, focal length etc.

I commend it to the forum (and it's a French word @ollypenrice 🥂) .

Regards Andrew 

PS Might be better to use aperture^2 x field of view arc sec ^2 for comparing disparate systems.

Edited August 29, 2020 by andrew s

[vlaiv]

3 minutes ago, ollypenrice said:

 

Goran, there's nothing wrong with that calculation but, like Datalord, I think F ratio is a complicated way of referring to aperture. It's only sensible to compare the same focal lengths, in which case the only physical variable is aperture. As we've concluded on previous occasions the ratio we really want is area of aperture per area of pixel.

Olly

F/ratio works if (physical) pixel size is kept the same, not just focal length.

If you want ratio to define speed (something per something) - best thing to use would be aperture per pixel size in arc seconds (reciprocal of resolution in "/px) - or maybe easier way to remember would be just multiply aperture and resolution.

Now we can compare different combinations of gear like 200mm at 1"/px to 80mm at 2"/px.

200mm at 1"/px is just 200mm per 1px per arc second so 200 (or 200 x 1 = 200)

80mm at 2"/px is 80mm per 0.5 px per arc second is 160mm per 1px per arc second so 160 (or 80 x 2 = 160).

First system is 200 and second is 160 - first is faster by 1.25 in SNR (200/160) for given time (in light dominant regime) or 1.25^2 = 1.5625 in signal strength / photons per pixel.

We can do similar thing with respect to FOV like @andrew s pointed out above, rather than per pixel and that is useful if we bin / resize image after, this pixel approach is better if we want to compare systems without binning / resizing.

 

[Datalord]

4 minutes ago, vlaiv said:

Now we can compare different combinations of gear like 200mm at 1"/px to 80mm at 2"/px.

200mm at 1"/px is just 200mm per 1px per arc second so 200 (or 200 x 1 = 200)

80mm at 2"/px is 80mm per 0.5 px per arc second is 160mm per 1px per arc second so 160 (or 80 x 2 = 160).

First system is 200 and second is 160 - first is faster by 1.25 in SNR (200/160) for given time (in light dominant regime) or 1.25^2 = 1.5625 in signal strength / photons per pixel.

Yeah, this is a perfect example of how we should compare "speed". I wish I had had this clarity in the beginning. I took the long way around to the large mirrors I have now.

[vlaiv]

3 minutes ago, Datalord said:

Yeah, this is a perfect example of how we should compare "speed". I wish I had had this clarity in the beginning. I took the long way around to the large mirrors I have now.

Large mirror * large sensor = large speed

[andrew s]

23 minutes ago, vlaiv said:

Large mirror * large sensor = large speed

Hallelujah,  hallelujah....

(Provided the scopes usable field can cover the sensor .)

Regards Andrew 

[gorann]

1 hour ago, ollypenrice said:

 

Goran, there's nothing wrong with that calculation but, like Datalord, I think F ratio is a complicated way of referring to aperture. It's only sensible to compare the same focal lengths, in which case the only physical variable is aperture. As we've concluded on previous occasions the ratio we really want is area of aperture per area of pixel.

Olly

Hi Olly

for me it was kind of obvious that it had to be the same camera and FL to make sense, so maybe I should avoided the dreaded f-word and initially have talked about comparing 200 mm aperture with 80 mm at the same FL to be more pedagogic. In any case the fact is that my RASA8 captures >6  times more photons per hour than my 3" refractor with the same camera. But it makes a lousy travel-scope in comparison😉

[gorann]

53 minutes ago, andrew s said:

Provided the scope/detector has the resolution required (which is almost always the case with modern CMOS cameras) this captures the rate of accumulation of information from the sky.

Unfortunately that is still not really true. My ASI2600MC has 3.76 µm pixels, which is in the lower end of what is out there, so with the 400 mm FL of the RASA8 I am undersampling at 1.94 "/pixel.  On the positive side is that I do not have to feel too bad when I cannot get the guiding of my NEQ6 to fall below 1" RMS. With my 14 " Meade that would be totally useless and fortuantely it is on another mount. The other good thing is that smaller pixels on an APS-C chip would give astronomical file sizes. They are already 50 Mb coming off the camera and increase to 300 Mb after debayering in PI. So Debayering, aligning and calibrating takes half a day.

[vlaiv]

6 minutes ago, gorann said:

Unfortunately that is still not really true. My ASI2600MC has 3.76 µm pixels, which is in the lower end of what is out there, so with the 400 mm FL of the RASA8 I am undersampling at 1.94 "/pixel. 

I doubt that you are under sampling with 1.94"/px.

Look at this:

RASA is not quite what we would call diffraction limited system. If it were, then those dots in first row would be something like half of the smallest dot in this diagram (that blue for example at 500nm  / 10mm - or about 1 to 1.5 little squares in this grid).

In another words, most of these dots above are about 5-10 times larger than diffraction limited 8" aperture. For diffraction limited aperture of this size - optimal sampling rate (planetary) is about 0.26"/px

This means that equivalent sampling rate, without influence of atmosphere and tracking would be 1.3"/px to 2.6"/px (depending on wavelength and spot size above). Add seeing and tracking and you can see how it easily shoots above 2"/px.

But that is ok, RASA 8 is not meant to be diffraction limited imaging scope - it is meant to be fast wide field imaging scope and in wide fields resolution is sacrificed so larger spots don't really matter.

[gorann]

34 minutes ago, vlaiv said:

I doubt that you are under sampling with 1.94"/px.

Look at this:

RASA is not quite what we would call diffraction limited system. If it were, then those dots in first row would be something like half of the smallest dot in this diagram (that blue for example at 500nm  / 10mm - or about 1 to 1.5 little squares in this grid).

In another words, most of these dots above are about 5-10 times larger than diffraction limited 8" aperture. For diffraction limited aperture of this size - optimal sampling rate (planetary) is about 0.26"/px

This means that equivalent sampling rate, without influence of atmosphere and tracking would be 1.3"/px to 2.6"/px (depending on wavelength and spot size above). Add seeing and tracking and you can see how it easily shoots above 2"/px.

But that is ok, RASA 8 is not meant to be diffraction limited imaging scope - it is meant to be fast wide field imaging scope and in wide fields resolution is sacrificed so larger spots don't really matter.

First you got me worried, so thanks for that last line Vlaiv!

[ollypenrice]

In any event, no free lunch in AP...

Olly

[gorann]

20 minutes ago, ollypenrice said:

In any event, no free lunch in AP...

Olly

The RASA8 is at least a cheap lunch compared to the other "astrographs" on the current menue😊

[ollypenrice]

2 minutes ago, gorann said:

The RASA8 is at least a cheap lunch compared to the other "astrographs" on the current menue😊

Yes, perhaps so. However, one of the ironies of widefield systems is that the user often wants to go much, much wider with them and do mosaics. (Personally I have lost interest in my 530mm/full frame rig for single panel targets. I only want to use it for mosaics.) In this case the ability to cover large formats also comes into play so a system which needs multi panels to cover full frame ceases to be as fast as it looks.

Olly

 

 

[gorann]

8 minutes ago, ollypenrice said:

Yes, perhaps so. However, one of the ironies of widefield systems is that the user often wants to go much, much wider with them and do mosaics. (Personally I have lost interest in my 530mm/full frame rig for single panel targets. I only want to use it for mosaics.) In this case the ability to cover large formats also comes into play so a system which needs multi panels to cover full frame ceases to be as fast as it looks.

Olly

 

 

Fortunately Olly I have only been into this wide-field imaging for about a week so I think it will take a while before I run out of tragets and go into making mosaics. Right now I am more thinking about using my Esprits to sharpen up parts of what I get from the RASA when the moon is gone and nights get longer.

[ollypenrice]

36 minutes ago, gorann said:

Fortunately Olly I have only been into this wide-field imaging for about a week so I think it will take a while before I run out of tragets and go into making mosaics. Right now I am more thinking about using my Esprits to sharpen up parts of what I get from the RASA when the moon is gone and nights get longer.

You have it all to come, Goran... 😁👹🤣

When you have wide, there is only one way to go and that's wider...

Olly