72 replies to this topic

[HxPI]

That background profile looks beautiful!! Guess I'll have to add more cowbell, err frames, to the sequence! So Gain 76, cooling of course, 30-60 sec exposures, aggressive dithering, and hundreds of frames!! I'll give it a try. Thanks for the explanations Jon!

 

Ciao,

Mel

Edited by HxPI, 02 June 2017 - 11:06 AM.

[Jon Rista]

Well, use the gain and exposure length that makes the most sense with your equipment. I have an f/4 scope used at a red/white zone border. And I used Gain 76 because I was having some calibration issues at Gain 0, and wanted to minimize read noise more. But, it depends a lot on your skies and your aperture what exposure you will really need. Don't make the mistake of too-ridigly adhering to the exact settings I use. That's why I like the exposure rules, then can help you calculate the necessary exposure on your own by figuring out how long it would take to get a given minimum necessary signal level (and, if you need to, you can certainly expose for longer than that so long as you are not clipping too much.) The one thing I do recommend though is stacking a lot of subs...it just, really helps with this camera.

Edited by Jon Rista, 02 June 2017 - 11:14 AM.

[HxPI]

Understood. Thanks.

[John Miele]

I guess I'm just going to have to ditch using my beloved Nebulosity 4 with my ASI1600mm and move to SGP for one reason...Neb4 does not allow dithering unless you do it every frame. My aggressive dithering and settling time using 300+ short subs is killing me. Being able to dither every 4th frame would really help.

 

I am so comfortable with Nebulosity's controls and user interface, I have been resisting the switch...but I guess it's time to move on...

 

P.S. If anyone using Nebulsoity has found a way to dither every "nth" frame, pleeeaassee let me know!

 

Thanks

 

John

[Jon Rista]

John, have you asked Craig Stark if he could add sparse dithering to Nebulosity? If it's not currently a feature, there shouldn't be any reason it couldn't become a feature. 

[Marcelofig]

I guess I'm just going to have to ditch using my beloved Nebulosity 4 with my ASI1600mm and move to SGP for one reason...Neb4 does not allow dithering unless you do it every frame. My aggressive dithering and settling time using 300+ short subs is killing me. Being able to dither every 4th frame would really help.

 

I am so comfortable with Nebulosity's controls and user interface, I have been resisting the switch...but I guess it's time to move on...

 

P.S. If anyone using Nebulsoity has found a way to dither every "nth" frame, pleeeaassee let me know!

 

Thanks

 

John

I'm also putting aside Nebulosity, at least for capturing images. But mainly because it offers very few options.

 

Now I'm testing APT and it turns out to be a simple and intuitive program, easy to learn. You can, for example, among many other things, automatically control the filter wheel (you do not need to change them manually), it has advanced dithering and also includes platesolve.

 

Also the demo version is fully functional and no expiration date. Basically free, except for some advanced options.

 

(Later comes SGP, but for the moment it is better to start with the simpler).

[dangarnett]

@JonRista Great info Jon, I appreciate your insight. I didn't know Gain affected QE, good to know! I've been mainly imaging at 200 gain, 50 offset at f/4 to get around 30s subs. Which for LRGB sounds like it might be a bit too high. I struggle to get the avx mount to track smooth at 812FL so keeping the exposures short helps with tracking errors. Getting an Atlas Pro at some point....

 

I have a question about binning then. I might not understand quite how binning works exactly besides making 4 pixels into one large pixel to increase sensitivity. I was under the impression for RGB filters you increase the exposure to get to the desired background sky level (my red is a little higher than g and b) or you keep the same exposure length and bin 2x2? Is that right? I seem to keep around the same adu values binned or not binned and they are undersampled at 30s 2x2. 

[Jon Rista]

I wouldn't really say gain affects Q.E. The Q.E. of a sensor is a fixed curve. What I would say is gain affects sensitivity. Sensitivity is relative to BOTH Q.E. and read noise. If you increase Q.E. or reduce read noise, you are increasing the sensitivity of the camera either way. If you increase both, you are increasing the sensitivity of the camera quite considerably. This bears out if you think about it just a bit. If you have 9e- read noise, and your pixels each sense 2 photons. You have 2e- signal, and your SNR is 2/SQRT(2 + 9^2) = 0.22:1! That's pretty crummy! Now, if you have 1.3e- read noise, and you also sense 2 photons, then your SNR is 2/SQRT(2 + 1.3^2) = 1.04:1. That's a good start (from an engineering standpoint, an acceptable SNR starts at 1:1). The sensitivity of the camera increased, even though our quantum efficiency did not...for the same unit time, we still sensed the same amount of photons, two. The sensitivity increased because our read noise dropped. 

 

As for sensitivity by area. Larger pixels will gather more light than smaller pixels, yes. However, you only really actually gain sensitivity if there is also a corresponding reduction in read noise per unit area when you bin. With CCD cameras, this reduction in read noise does exist (some people claim 2x2 binning with a CCD gives you only one unit of read noise for all four pixels, however in practice that is FAR from reality...read noise will usually increase, as there are multiple sources of noise in a CCD sensor, including just the shifting of pixel charges around, so you might have around two units of read noise rather than four when binning 2x2). With CMOS cameras, hardware binning, while technologies do exist to do it at the hardware level with noise benefits, is usually done on the post-ADC digital values, so there is no improvement in read noise. So binning or not, there really isn't any improvement. There can be different ways of looking at SNR. There is the notion of pixel SNR, which solely deals with the signal to noise ratio within each individual pixel. However there is also object SNR, which deals with the total amount of light gathered by the scope for the given object. From an object SNR standpoint, it doesn't much matter how big the pixels are, so long as the object fits within the sensor. You will gather the same total amount of light from that object regardless of how big the pixels are.

[Ron in Michigan]

The initial poster said he increased gain for focus. I use SGP and have not figured out how to do this with my 1600MMc. 

It would sure speed up focus times - currently about 2.5 minute process. 

 

I don't see any way to set gain from inside SGP.  Is it done by configuring binning? (if so - how/what) where?

[Wwilmoth69]

To change the gain in sgp you go to your profile setting thedn camera then the settings tab to the right. It brings up the gain of set and usb speed
Or to the the right where u can connect individual parts there is a setting button I believe. I'll post a pic tonight when I get home for you I'm a visual guy





I just bought a 1600 and I'm trying to understand all this
Not only am I a newb but I'm a newb haha

I'm gonna read through this a few times as I'm trying to figure out what mean median to shoot for

This all just blew my mind so tonight I'm gonna have to go back through slowly and see if I get it

Edited by Wwilmoth69, 05 October 2017 - 05:54 PM.

[Wwilmoth69]

OK so I think I get it but what I don't understand is how do you find Out mean is while you're imaging because you say that it needs to be X amount after all of your calibration if my sky glow changes every night how do I estimate that before processing

Or do I just have to remember the mean for each settings and what the mean and median is during that image to figure out the average of what it normally would be at


Say at a gain of 170 gives a median of 1500 during imaging and after calib I get 45 I just adjust my exposure to mnatch 1500 during my exposures because that will alkways result in 45 and the sky glow just mandates my exposure time

Is that right?? Or does that even make sense lol

[dts350z]

These are some suggested numbers provided by other users. Not sure if you use SGP. If you are then just have a look at the Image Statistics window on the left (after you've taken a pic) to see what your subs Mean ADU number is.

 

Optimal Exposure:
 

Median ADU shown in SGP:

 

Gain 0 Offset 10:         400 ADU

Gain 75 Offset 12:       550 ADU

Gain 139 Offset 21:     850 ADU

Gain 200 Offset 50:    1690 ADU

Gain 300  Offset 50 :  2650 ADU

 

Better to be a little high than low.

 

For other Gains or offsets use the formula Jon indicated:

 

MinDN16 = (((ReadNoise * 20) / Gain) + BiasOffset) * 2^16/2^Bits

 

WHERE:

 

ReadNoise is the read noise in e-

Gain is the gain in e-/ADU

BiasOffset is the ADC offset in ADU

Bits is the bit depth of the camera

 

So, as an example, for the ASI1600 @ Gain 0 with 3.5e- read noise and 4.88e-/ADU at 12-bit:

 

MinDN16 = (((3.5 * 20) / 4.88) + 10) * 16 = ((70 / 4.88) + 10) * 16 = (14.4 + 10) * 16 = 25 * 16 = 400

You would need a minimum 400 DN 16-bit background sky @ Gain 0 with Offset 10.

 

 

You can get a reasonable approximation of the Read Noise and Gain from the ASI Graph:

 

1600-Gain-RN-DR-FW-vs-gain-716x1024.jpg

Is this for an "empty star field" or for your target? If for the target, say a bright nebula that fills the frame, I would be concerned that you would be exposing too low, with darker nebulosity lost in the noise.

 

If this for an "empty star field", in other words trying to measure the skyglow, then it makes sense to me.

 

Also, for RGB, if you do this per filter are undoing any attempt by the filter manufacturer to color balance the filters?

[Wwilmoth69]

Well last night I tried narrow band at unity gain

Lum came out 2000
Ha
S111 were at 360 and 16 min
I think I clipped the bottom end probably
Lum were 120seconds exposure and narrow I did 210 seconds

Guess I need to increase gain on the narrow to bump it up to the right of the histogram a lot lol
Could see great detail in ha on the bubble neb even though mean was low

I'll try and process it later and upload each of the filters after calibrating with darks and see what y'all think

First night with the camera went great just need to understand this whole adu thing better but ill get there

[dts350z]

 

These are some suggested numbers provided by other users. Not sure if you use SGP. If you are then just have a look at the Image Statistics window on the left (after you've taken a pic) to see what your subs Mean ADU number is.

 

Optimal Exposure:
 

Median ADU shown in SGP:

 

Gain 0 Offset 10:         400 ADU

Gain 75 Offset 12:       550 ADU

Gain 139 Offset 21:     850 ADU

Gain 200 Offset 50:    1690 ADU

Gain 300  Offset 50 :  2650 ADU

 

Better to be a little high than low.

 

For other Gains or offsets use the formula Jon indicated:

 

MinDN16 = (((ReadNoise * 20) / Gain) + BiasOffset) * 2^16/2^Bits

 

WHERE:

 

ReadNoise is the read noise in e-

Gain is the gain in e-/ADU

BiasOffset is the ADC offset in ADU

Bits is the bit depth of the camera

 

So, as an example, for the ASI1600 @ Gain 0 with 3.5e- read noise and 4.88e-/ADU at 12-bit:

 

MinDN16 = (((3.5 * 20) / 4.88) + 10) * 16 = ((70 / 4.88) + 10) * 16 = (14.4 + 10) * 16 = 25 * 16 = 400

You would need a minimum 400 DN 16-bit background sky @ Gain 0 with Offset 10.

 

 

You can get a reasonable approximation of the Read Noise and Gain from the ASI Graph:

 

1600-Gain-RN-DR-FW-vs-gain-716x1024.jpg

Is this for an "empty star field" or for your target? If for the target, say a bright nebula that fills the frame, I would be concerned that you would be exposing too low, with darker nebulosity lost in the noise.

 

If this for an "empty star field", in other words trying to measure the skyglow, then it makes sense to me.

 

Also, for RGB, if you do this per filter are undoing any attempt by the filter manufacturer to color balance the filters?

 

I'd like to get an answers to my above questions.

 

Also, using "median" value in SGP for dim targets might not be optimal? For instance, Sh2-136 the ghost nebula. It's a dim reflection nebula and it seems exposing much above this gain 200 offset 50 median value > 1690 in SGP gives better results (less noise) for the same amount of total integration time?

[dan_hm]

 

 

These are some suggested numbers provided by other users. Not sure if you use SGP. If you are then just have a look at the Image Statistics window on the left (after you've taken a pic) to see what your subs Mean ADU number is.

 

Optimal Exposure:
 

Median ADU shown in SGP:

 

Gain 0 Offset 10:         400 ADU

Gain 75 Offset 12:       550 ADU

Gain 139 Offset 21:     850 ADU

Gain 200 Offset 50:    1690 ADU

Gain 300  Offset 50 :  2650 ADU

 

Better to be a little high than low.

 

For other Gains or offsets use the formula Jon indicated:

 

MinDN16 = (((ReadNoise * 20) / Gain) + BiasOffset) * 2^16/2^Bits

 

WHERE:

 

ReadNoise is the read noise in e-

Gain is the gain in e-/ADU

BiasOffset is the ADC offset in ADU

Bits is the bit depth of the camera

 

So, as an example, for the ASI1600 @ Gain 0 with 3.5e- read noise and 4.88e-/ADU at 12-bit:

 

MinDN16 = (((3.5 * 20) / 4.88) + 10) * 16 = ((70 / 4.88) + 10) * 16 = (14.4 + 10) * 16 = 25 * 16 = 400

You would need a minimum 400 DN 16-bit background sky @ Gain 0 with Offset 10.

 

 

You can get a reasonable approximation of the Read Noise and Gain from the ASI Graph:

 

1600-Gain-RN-DR-FW-vs-gain-716x1024.jpg

Is this for an "empty star field" or for your target? If for the target, say a bright nebula that fills the frame, I would be concerned that you would be exposing too low, with darker nebulosity lost in the noise.

 

If this for an "empty star field", in other words trying to measure the skyglow, then it makes sense to me.

 

Also, for RGB, if you do this per filter are undoing any attempt by the filter manufacturer to color balance the filters?

 

I'd like to get an answers to my above questions.

 

Also, using "median" value in SGP for dim targets might not be optimal? For instance, Sh2-136 the ghost nebula. It's a dim reflection nebula and it seems exposing much above this gain 200 offset 50 median value > 1690 in SGP gives better results (less noise) for the same amount of total integration time?

 

Re. your first question: It is not just for an empty star field.  The reading should be the same regardless of what you're aiming at.  If you open your subexposure in Pixinsight and mouse over an area of background sky with the Readout tool (with 16 bit set as the integer range), you should get a reading close to the Median ADU that's shown in SGP.  

[Salacious B Crumb]

I’ve been reading and re-reading this now for a couple of months and it seems like I need to lengthen my exposure times beyond what people talk about here in order to get the needed ADU’s and histogram separation from the left side.

 

According to Bortle maps, I’m on the orange zone (#5) and the iPhone Dark Sky Meter app tells me my SQM is 20.3. I’ve pumped up my settings to 139/21, yet 3 min narrowband subs are not enough to give me any left-edge separation or needed ADU’s. The scope is a f/6.5.

 

 

 

 

Any thoughts?

 

-Sal

[FiremanDan]

I’ve been reading and re-reading this now for a couple of months and it seems like I need to lengthen my exposure times beyond what people talk about here in order to get the needed ADU’s and histogram separation from the left side.

 

According to Bortle maps, I’m on the orange zone (#5) and the iPhone Dark Sky Meter app tells me my SQM is 20.3. I’ve pumped up my settings to 139/21, yet 3 min narrowband subs are not enough to give me any left-edge separation or needed ADU’s. The scope is a f/6.5.

 

 

ADU + Histogram.JPG

 

 

Any thoughts?

 

-Sal

What filters are you using? I discovered my S2 was a bit lower than I thought at 76gain and 600s. 

[Salacious B Crumb]

These are still the ZWO cheapos. Trying to find a buyer for my Kidney so I can buy the AstroDon's...

 

 

- Sal

[kingjamez]

I’ve been reading and re-reading this now for a couple of months and it seems like I need to lengthen my exposure times beyond what people talk about here in order to get the needed ADU’s and histogram separation from the left side.

 

According to Bortle maps, I’m on the orange zone (#5) and the iPhone Dark Sky Meter app tells me my SQM is 20.3. I’ve pumped up my settings to 139/21, yet 3 min narrowband subs are not enough to give me any left-edge separation or needed ADU’s. The scope is a f/6.5.

 

 

ADU + Histogram.JPG

 

 

Any thoughts?

 

-Sal

Hi Sal,

I too have struggled with the recommendations. I've asked several times, and for the life of me I can't rectify the "optimal" ADU values and the desire to separate the histogram from the left side (which I take to mean starting to raise the noise floor). 

 

In the end you are trying to balance two things:

1. Getting the signal above the noise floor so that you know you aren't loosing SNR due to noise.  (getting the histogram to move from the left side)

2. Not clipping the brightest portions of the image due to a lack of dynamic range. (not blowing out your stars)

 

It is entirely possible that the two goals are incompatible, that if you move the histogram away from the left side, you'll blow out stars. Or vise versa.

 

I don't buy these median ADU charts, I don't see how they could possibly account for different content in the images. Take imaging the Orion Nebula vs a dim galaxy cluster for instance, median ADU can't tell the whole exposure story. 

 

Gain is inversely proportional to dynamic range, so one way of combating this problem is to lower your gain. However at least with CMOS cameras, read noise increases with lower gain so there is a bit of tail chasing.

 

I'm still very much in the learning by experimenting phase however I've had my best luck by prioritizing maximizing target SNR over clipping of the brightest stars. 

 

My technique is to expose at the absolute minimum time to safely separate the noise floor from the left side of the histogram reliably. Anything past initial separation and you are not gaining any additional SNR but you are blowing out more stars so you need to tune that as best you can. This often means I'm exposing well past the recommended ADU readings suggested here.

 

-Jim

Edited by kingjamez, 12 December 2017 - 10:58 AM.

[kingjamez]

I’ve been reading and re-reading this now for a couple of months and it seems like I need to lengthen my exposure times beyond what people talk about here in order to get the needed ADU’s and histogram separation from the left side.

 

According to Bortle maps, I’m on the orange zone (#5) and the iPhone Dark Sky Meter app tells me my SQM is 20.3. I’ve pumped up my settings to 139/21, yet 3 min narrowband subs are not enough to give me any left-edge separation or needed ADU’s. The scope is a f/6.5.

 

 

ADU + Histogram.JPG

 

 

Any thoughts?

 

-Sal

I'll also add that given your above screen shot. I'd up the gain to 200/50 (a good compromise between read noise and dynamic range) and try your 3 minute exposure again. Bump it up if needed until your minimum value gets above 16 (really 1 due to the 16 bit conversion in SGP) I use a little padding and shoot for around 50 or so as my minimum value that helps with reliable separation as the night progresses. 

 

-Jim

Edited by kingjamez, 12 December 2017 - 10:56 AM.

[glend]

I have put hundreds of hours on my ASI1600MM-C since May 2016, mostly shooting narrowband. I have  found that obsessing about settings is a waste of time. I shoot almost everything at Unity and the images turn out fine. There are times i use High Gain if  i am in a hurry, or Gain 76 Off 20 on a target like Eta Carinea, but usually Unity is fine. The camera is very flexible and worrying about precise ADU rules will drive you crazy. Just get out there and use it.

PS, my standard narrowband sub length is 300s, and lots of them on dim targets, but i usually drop back to 2min or less on brighter objects. I use SGP to manage the camera, works great. Baader filters.

Edited by glend, 12 December 2017 - 11:46 AM.

[Salacious B Crumb]

Yes, I should have mentioned that this was a Ha-filter and IC1396. The moon was not visible yet. And by no means I don't want my confusions to leave to another he-said / he-said debate. I'm just curious...

 

 

 

- Sal

Edited by Salacious B Crumb, 12 December 2017 - 11:32 AM.

[scopenitout]

These are still the ZWO cheapos. Trying to find a buyer for my Kidney so I can buy the AstroDon's...


- Sal

Astrodon's are good, but just don't lose track of how many kidneys you've sold.

[BillD17]

I’ve been reading and re-reading this now for a couple of months and it seems like I need to lengthen my exposure times beyond what people talk about here in order to get the needed ADU’s and histogram separation from the left side.

 

According to Bortle maps, I’m on the orange zone (#5) and the iPhone Dark Sky Meter app tells me my SQM is 20.3. I’ve pumped up my settings to 139/21, yet 3 min narrowband subs are not enough to give me any left-edge separation or needed ADU’s. The scope is a f/6.5.

 

 

ADU + Histogram.JPG

 

 

Any thoughts?

 

-Sal

What iPhone Dark SkyMeter app are you using?  I can't find one in the App Store.

 

I'm getting good results with 139/21 on an f/7 refractor with 1min subs for L and 3 min subs for R,G,B, and narrowband.  This is from  a suburban site.  It depends a bit on how bright the subject is, but my sky glow is well off the left edge of the histogram.

[Salacious B Crumb]