240 replies to this topic

[acaballero]

Shiraz,

May I ask how did you obtain those tables? Did you use some formulas to calculate them yourself? If so what are the formulas?

I own an ASI1600MC not cooled and I would like to calculate the subexposures for my equipment.

Regards

[Shiraz]

Hi - good questions

the tables were generated using a fairly detailed end-to-end astrograph model - as far as I know, there are no simple formulas that do the same. The spreadsheet itself is available if you wish to play with it - I put it on the Australian astro site Iceinspace at http://www.iceinspac...orum/showthread. php?t=151000. Note that your spreadsheet software will warn you that this is unknown and potentially dangerous software - I can't do anything meaningful to assure you that it is safe, but it is a locked spreadsheet, written in Excel on an up to date Win10 machine

I think that the tables could be used as a reasonably good starting point for your color camera, if you use sub lengths 3x as long as those listed for the mono camera. Dark current will be a complication, but I would be surprised if it made much difference to the recommended sub lengths if your air temperature is not excessive.

as well as the recommended sub lengths, there is also a column that gives expected ADU values above bias. This applies to the camera regardless of whether it is color or not. To check your sub lengths, add your own camera's bias ADU value to the expected ADU value from the table - and then compare the result with the ADU value in the background sky of one of your lights. If your measured sky ADU is much higher than the expectedADU+bias, then you could consider reducing your sub length - and vice versa.

It should also be said that you will get some sort of result whatever value of sub length you use - the suggested values are an attempt to maximise the imaging efficiency by providing the best possible final image signal-to-noise ratio (SNR) in the shortest possible time. If you use longer subs, you will not get any improvement in the final image SNR in a given time, but will lose some dynamic range and clip more stars. If you use significantly shorter subs, you will take a lot longer to get a final image of the same SNR quality, but you will have less of a problem with bright stars clipping.

cheers Ray

Edited by Shiraz, 24 August 2017 - 06:44 PM.

[Mert]

I have a Little question on the data supplied by Pixinsight where

it states DN in stead of ADU, is that correct?

 

Then as you can see in the attached screenshot, I went way overhead

with this.

I took 240 seconds subs with the camera at -15ºC on a TS triplet APO

with reducer ( F5.5 ).

Checking on the subframe exposure table you shared I should have

given aprox. 1 minute of exposure?

The sky was pritty dark but I don't have a SQM to check.

 

FWIW the screenshot and the final image ( 23 frames ) attached.

Attached Thumbnails

• 
• 

Edited by Mert, 25 August 2017 - 01:33 PM.

[Shiraz]

Hi Mert - very nice image

I have always assumed that ADU and DN are interchangeable.

Cannot comment in detail on the exposure without knowing what gain you used and what the ADU was in your bias frames (assuming that the PI readout is from an uncalibrated sub).

Also, assume that you used a mono camera? If so, it is likely that you could have used shorter subs on your f5.5 scope and got a similar result.

The sub exposure tables do not tell you what you should do - that is your call. The suggestions show how short your subs can be before they start to reduce the final image SNR. You can go longer if you want and get the same final SNR in the same total time, but at the expense of more saturated stars. You could go shorter if you didn't mind taking significantly longer overall to get to the same final SNR and if you want less star saturation.

cheers ray

Edited by Shiraz, 26 August 2017 - 12:13 AM.

[Mert]

Hi Ray,

 

Thanks for pointing that out, indeed the screenshot was from an uncalibrated light ( 2016 )!

I use an ASI1600MC-C ( not a mono versión ) and used Unity gain 139 setting the

settings from the ascom driver, offset I don't remember right now .

 

Effectively I have noticed that when going with longer exposure times, the stars loose

their color, like they are burned.

 

Here is a screenshot from a Bias fame in Pix:

Attached Thumbnails

• 

Edited by Mert, 26 August 2017 - 04:15 AM.

[Shiraz]

Thanks Mert. definitely looks you could use quite a bit shorter subs if you wanted - at gain 139 the sky background could be as low as 450ADU (about a third of your 1300+), which means that you could use subs a third as long (ie of a bit over a minute rather than 4 minutes).

The symptom of reduced dynamic range is burn out of the cores of more stars - if you go for 1 minute subs at gain 139 you might have significantly less star clipping. Alternatively, you could reduce the gain to 0 and continue to use ~4 minute subs - as shown in the tables. The results would be much the same as from shorter subs at higher gain (the dynamic range after stacking optimised subs, is pretty much the same for gains from 0 to 139, as shown in post #39).

cheers ray

Edited by Shiraz, 26 August 2017 - 07:28 AM.

[Mert]

Thanks a lot Ray, that is great news!

This means I can go for shorter subs ( very interesting) and

do many more subs and as such have a better SNR and less noise.

My laptop will crunch as many images as I throw onto it without

a problem, it has a SSD and runs fast!

I will do testing, this makes the hobby much more interesting!!

 

Great stuff!!!! You made my day

[Shiraz]

thanks Mert. if you go for shorter subs, the subs themselves will look less impressive, but you will end up with the same final image SNR after stacking - but many fewer stars will be burned out. Have also found that the resolution can be better with short subs, so that may be another benefit.

It is sure an interesting hobby!

Cheers Ray

Edited by Shiraz, 26 August 2017 - 05:53 AM.

[Mert]

Thanks again for your effort sharing all this with us!!

[Mert]

One question left for sure Ray, I don't see any recommended values for the offset

in order to reach the values of the tables.

Is there any advisable value for the offset other then the default value I use?

( unity gain setting 139 and offset 10?? )

 

Thanks for any thoughts on this!!

[acaballero]

Ray,

Thanks for sharing your knowledge.

I can see lots of people are benefiting.

[Shiraz]

Mert, the offset setting just needs to be big enough to keep the image values above zero at all times. The actual ADU offset from the camera varies a bit unpredictably with gain (at least mine does) - the default values suggested by the maker have plenty of headroom to allow for read noise and offset variability - suggest that you use the graph in post#7 of https://www.cloudyni...or-zwo-cameras/ . I don't know enough about the QHY163 camera to make any suggestions, other than to use the values recommended by the maker.

acaballero, just realised that my editing seems to have messed up the link I gave you in post#52 - this is what it should have been http://www.iceinspac...ad.php?t=151000

cheers Ray

Edited by Shiraz, 26 August 2017 - 12:11 PM.

[Jon Rista]

I have always assumed that ADU and DN are interchangeable.
 

I just need to address this. DN and ADU are not universally interchangable. In PixInsight, DN are always 16-bit. IF you happen to have a camera with a 16-bit ADC, then yes DN and ADU represent the same thing. However there are a lot more cameras on the market these days that have 14-bit, 12-bit and in some cases even 10-bit ADC (the latter usually in certain camera modes). ADU means Analog to Digital Unit, which only has meaning within the context of the ADC that is producing them. If the ADC is 14-bit, then an ADU and a DN are most definitely not the same thing, and most definitely not interchangable. 

 

When discussing things like gain, read noise, dynamic range, exposure lengths, etc. then knowing what your ADU really are is very important. Because if you convert DN back to electrons when DN and ADU are not the same, you could be off by a significant factor.

[Shiraz]

thanks for that clarification Jon.

Thankfully, from a DSO imager's point of view, the 12 and 14 bit cameras are indistinguishable from ones with a 16 bit ADC - data is usually presented in 16 bits and the term ADU (traditionally referring to the digital data in the subs) can be used interchangeably with the DN term in PI.

To put your mind at rest, the original model on which this thread is based deals exclusively with signal at the photon/electron level.

cheers Ray

Edited by Shiraz, 26 August 2017 - 07:16 PM.

[kingjamez]

Can I get some clarification on what's important in exposure? I've been using the tables to get my ADU to an appropriate range for my gain setting. Tonight with my Ha filter, at gain 200 offset 50 I can get an ADU in the ~1900 range with a 3 minute exposure. However, my histogram doesn't cleanly separate from the left side of the histogram at this level. I have to go up to somewhere around 5 minutes and an ADU of 2500 is where I have to be to get any left side separation.

 

Which should I prioritize? Optimally swamping the read noise or getting left side histogram separation?

 

-Jim

[TareqPhoto]

Can I get some clarification on what's important in exposure? I've been using the tables to get my ADU to an appropriate range for my gain setting. Tonight with my Ha filter, at gain 200 offset 50 I can get an ADU in the ~1900 range with a 3 minute exposure. However, my histogram doesn't cleanly separate from the left side of the histogram at this level. I have to go up to somewhere around 5 minutes and an ADU of 2500 is where I have to be to get any left side separation.

 

Which should I prioritize? Optimally swamping the read noise or getting left side histogram separation?

 

-Jim

That is what making me stopping taking images, i feel values given either in charts or tables can't be as fixed for all time or conditions, and i feel i don't know what is correct or what i should follow, the only way for me is to test many values and combinations and see which is giving good one, don't know if the tables given are considering something like light pollution level and moon light degree and air quality and target zone location or whatever, i tried using the "DSO" default setting on SGP and different exposure time and i felt i have to modify something here or that but don't know which.

[Shiraz]

Jim, the two methods are trying to do the same thing - getting to a sub exposure where the imaging is most efficient. The very low read noise of this camera means that the sky noise (left hand side of the histogram) does not have to be very strong to overcome read noise and you can expect the histogram peak to be close to the left side, even at high gain and with Ha. The tables are based on quite conservative assumptions, so suggest that you "swamp the read noise" - your call though.
From the tables, at gain 200, your ADU values should be around 640 above the bias - you quote an ADU of 1900, does that include the bias offset and what bias level does your camera produce at 200/50? Your data suggest that bias offset is about 1000ADU - if right, that means that your 3 minute exposures were pretty close, but maybe a little longer than you really needed - ~2 minutes would have been sufficient for the conditions (you must have a bright sky).

Tareq. There are different tables for different light conditions, including moon - did the best I could, but of course they do not model in detail your exact conditions. They are intended to provide a starting point for sub exposures and then you can adjust the exposures up or down to get closer to the "expected ADU" values in your specific conditions. For example, if your sky brightness falls halfway between the table values, use a sub exposure halfway between those recommended on either side. Or, if you want to image in half moon conditions, start out with exposures halfway between the no moon and full moon tables. When you have taken a sub, adjust the sub exposure so that the final sky ADU minus the bias offset ADU for your camera (at chosen gain/offset) is close to the "expected ADU" value in any of the tables.
Suggest that you take images anyway - the only penalties for using different exposure are that the dynamic range will be reduced if the subs are too long or that you will take longer overall if your subs are too short - you will still get an image.

cheers Ray

Edited by Shiraz, 18 October 2017 - 07:10 PM.

[TareqPhoto]

Jim, the two methods are trying to do the same thing - getting to a sub exposure where the imaging is most efficient. The very low read noise of this camera means that the sky noise (left hand side of the histogram) does not have to be very strong to overcome read noise and you can expect the histogram peak to be close to the left side, even at high gain and with Ha. The tables are based on quite conservative assumptions, so suggest that you "swamp the read noise" - your call though.
From the tables, at gain 200, your ADU values should be around 640 above the bias - you quote an ADU of 1900, does that include the bias offset and what bias level does your camera produce at 200/50?

Tareq. There are different tables for different light conditions, including moon - did the best I could, but of course they do not model in detail your exact conditions. They are intended to provide a starting point for sub exposures and then you can adjust the exposures up or down to get closer to the "expected ADU" values in your specific conditions. For example, if your sky brightness falls halfway between the table values, use a sub exposure halfway between those recommended on either side. Or, if you want to image in half moon conditions, start out with exposures halfway between the no moon and full moon tables. When you have taken a sub, adjust the sub exposure so that the final sky ADU minus the bias offset ADU for your camera (at chosen gain/offset) is close to the "expected ADU" value in any of the tables.
Suggest that you take images anyway - the only penalties for using different exposure are that the dynamic range will be reduced if the subs are too long or that you will take longer overall if your subs are too short - you will still get an image.

cheers Ray

I will give that a try and see, hope i can get the right values.

[Greg F]

Trying to wrap my head around all this. If in a yellow zone with F5.4 and gain at 139 my exposure time with L would be around 35 seconds but 60 seconds would still be ok. But for NB in same zone I would have to be around 35 minutes. Or I could bump gain to 200 and be at 25 minutes. If so what do i lose by upping the gain? 

[kingjamez]

Trying to wrap my head around all this. If in a yellow zone with F5.4 and gain at 139 my exposure time with L would be around 35 seconds but 60 seconds would still be ok. But for NB in same zone I would have to be around 35 minutes. Or I could bump gain to 200 and be at 25 minutes. If so what do i lose by upping the gain? 

You loose dynamic range by upping the gain.

 

-Jim

[Greg F]

So keep it at Gain 139 and around 35 minutes for exposure would be good.

[Greg F]

So it seems i have been really over exposing at 20 minutes at 0 gain for LRGB

[Jon Rista]

You shouldn't need 20-35 minute exposures with this camera. Especially not for LRGB, you should be able to swamp the noise in significantly less time than that, even at a dark site. For narrow band, you might be able to use 10-15 minute exposures for narrow band, but also keep in mind that the purpose of a narrow band filter is contrast. At a dark site, there could be areas of the field that remain read noise limited, for all intents and purposes, forever, because there just isn't any on-band signal in those areas. That is what you want with narrow band, though...and you just have to make a judgement call about how long to expose (and the primary determinant there is star clipping.) 

 

Also, keep in mind, the ASI cameras usually have  a 2000 second maximum exposure limit, 33 minutes...so 35 minute subs are just out of the question regardless.

[Greg F]

Well guess there will be no 35 minutes, lol. I did not know there was a limit. I just did 30 minutes a few times last week trying to guide for the first time, would much prefer to stay at 10-15 minutes without having to guide although i just got a ZWO camera for guiding. Some much to learn, did not know that dark site I had to expose longer? I'll just keep reading what ya'll post and do what you say till i understand a little better. So just for clarification, 0 gain for LRGB i should try very short exposures, somewhere around 3 minutes or up the gain to around 139 and then go around 60 seconds. I don't understand dynamic range except it sounds like something that is preferable so 0 gain would be better then?

[Jon Rista]

What is your f-ratio?