Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Some musings (& images) on the ASI462MC's noise etc...

  • Please log in to reply
53 replies to this topic

#26 Kokatha man

Kokatha man

    Hubble

  • *****
  • topic starter
  • Posts: 15,072
  • Joined: 13 Sep 2009
  • Loc: "cooker-ta man" downunda...

Posted 28 July 2020 - 08:48 PM

 

I notice also that Darryl's raw frames above are also very blue looking - I'd be interested if Darryl could post a raw frame (like the one above) but prior to the RGB balance he did for the noise comparison between the two raw frames. Alternatively, what RGB settings in Registax did he use to get them to have a similar colour balance? FWIW, I have found that even with Wred=62 and Wblue=99, I still have to multiply the blue level in Registax by 1.1 to give me the "correct" colour balance. I now use these settings consistently for all my images.

 

Andrew

Sorry, I can't recall the Reggie colour values for our comparisons Andrew...

 

Here's that frame with only the levels raised for clarity...

 

SingleFrame+Levels.png


  • Tulloch and Jim in PA like this

#27 Tulloch

Tulloch

    Vanguard

  • *****
  • Posts: 2,356
  • Joined: 02 Mar 2019
  • Loc: Melbourne, Australia

Posted 28 July 2020 - 08:55 PM

Sorry, I can't recall the Reggie colour values for our comparisons Andrew...

 

Here's that frame with only the levels raised for clarity...

 

attachicon.gifSingleFrame+Levels.png

Thanks Darryl, is it possible to attach the raw frame only as a png? I'd just like to run it through R6 and see what comes up.



#28 Kokatha man

Kokatha man

    Hubble

  • *****
  • topic starter
  • Posts: 15,072
  • Joined: 13 Sep 2009
  • Loc: "cooker-ta man" downunda...

Posted 28 July 2020 - 09:08 PM

Untitled-2.png



#29 Dunkstar

Dunkstar

    Viking 1

  • -----
  • Posts: 680
  • Joined: 26 Mar 2012
  • Loc: Brisbane, Australia

Posted 28 July 2020 - 09:16 PM

Technically I understand "munging" to be the act of chewing & digesting so I suspect that we won't be able to discern very much from the ensuing data...but someone who can divine from animal or bird droppings might be able to have some input! bigshock.gif smiley-char145.gif

 

Dunk, I use that 52/95 combo but I believe Andrew (Tulloch) arrived at a different & slightly better ratio - but I do any corrections in post-processing that I think Reggie hasn't done sufficiently.

 

Having said that there is a somewhat "bluish" tint to this new camera's image appearance during preview & after stacking...

 

Using an ir filter with any of these colour cameras does produce weird colours. (you need to capture in colour however & then convert to mono to avoid that grid effect from the Bayer matrix) But I'm not quite sure what you are aiming at with your 800nm comments...I'm a bit slow atm. wink.gif

 

Isaac, there's a "cog" icon with "More" next to it in the FireCapture control panels, click on that to get the red/blue sliders... smile.gif

That's exactly what I was getting at...we feed the camera a crumpet, the software chews on it and spits it back out. We'd most likely recognise it as a crumpet but would it be very desirable? Sorry, a bit early for philosophy step.gif

 

My 800nm comment was aiming at the colour balance...if the response at 800nm through each R/G/B pixel is almost the same, I'd expect to see essentially a monochrome image. What I omitted to consider before my morning coffee was that we don't know for sure where the cut-off for each colour filter is deeper into the the IR where the pixels may still be sensitive.

 

The best way to check the correct white balance would be to image a colour card in daylight with the sun at around 45 degrees in a cloudless sky to simulate D50 conditions, the way I did it for my ASI224MC and eventually came up the the values Wred = 62 and Wblue = 99. 

https://www.cloudyni...olour-accuracy/

 

(Post #44 on page 2 was where I finally got to test those values if you are looking for that reference smile.gif).

 

I notice also that Darryl's raw frames above are also very blue looking - I'd be interested if Darryl could post a raw frame (like the one above) but prior to the RGB balance he did for the noise comparison between the two raw frames. Alternatively, what RGB settings in Registax did he use to get them to have a similar colour balance? FWIW, I have found that even with Wred=62 and Wblue=99, I still have to multiply the blue level in Registax by 1.1 to give me the "correct" colour balance. I now use these settings consistently for all my images.

 

It does sound like ZWO have made some changes to the sensitivity of the blue channel, based on user feedback that the raw blue levels are too low - once we have a good handle on the difference between the 224 and 462 for a similar image (and we could always start with the two images I mentioned before) then we could produce a decent estimate for the Wred and Wblue values to use in FC for the 462.

 

You can change the white balance in the FC settings tab shown below.

 

Those values are just a simple multiplier of the raw values from the camera. Green is set to a middle "50", the others are relative to the Green channel.

 

Andrew

Awesome, thanks Andrew!

 

So if I set the white balance for R and B to 50, then I'm not tainting the raw data? Or is the white balance purely for our entertainment while capturing and the raw data isn't affected at all?



#30 Dunkstar

Dunkstar

    Viking 1

  • -----
  • Posts: 680
  • Joined: 26 Mar 2012
  • Loc: Brisbane, Australia

Posted 28 July 2020 - 09:23 PM

This is what I'm getting out of the Astronomik ProPlanet 807.

 

Yes, the seeing truly was awful bawling.gif

Attached Thumbnails

  • Dunk-PP807.png


#31 Dunkstar

Dunkstar

    Viking 1

  • -----
  • Posts: 680
  • Joined: 26 Mar 2012
  • Loc: Brisbane, Australia

Posted 28 July 2020 - 09:28 PM

And for entertainment value, here's with the ProPlanet 742 (salmon) and the 642 (much more red!).

 

Couldn't escape the seeing though.

 

 

Attached Thumbnails

  • Dunk-PP742.png
  • Dunk-PP642.png


#32 Tulloch

Tulloch

    Vanguard

  • *****
  • Posts: 2,356
  • Joined: 02 Mar 2019
  • Loc: Melbourne, Australia

Posted 28 July 2020 - 09:37 PM

Thanks Darryl. R6 thinks that the colour channels need to be multiplied by (with normalised values in brackets).

 

R - 1.24 (1.32)

G - 0.94 (1.00)

B - 0.85 (0.90)

 

So, if you used Wred = 52, Wblue = 95 during the capture, I'd suggest that the "correct" values in FireCapture should be closer to Wred = 69, Wblue = 86 to give "correct" white balance straight out of the camera.

 

Andrew

Attached Thumbnails

  • Darryl ASI462MC R6 autoRGB post-87685-0-48028100-1595988512.png

Edited by Tulloch, 28 July 2020 - 09:41 PM.


#33 Tulloch

Tulloch

    Vanguard

  • *****
  • Posts: 2,356
  • Joined: 02 Mar 2019
  • Loc: Melbourne, Australia

Posted 28 July 2020 - 09:41 PM

So if I set the white balance for R and B to 50, then I'm not tainting the raw data? Or is the white balance purely for our entertainment while capturing and the raw data isn't affected at all?

Hi there, there is no "tainting" of the data - all colour cameras multiply their channels in order to give a more-or-less "correct" white balance. Since the sensitivity of the sensor does not match that of the human eye, using these values for Wred and Wblue can be used to approximate a true colour cast.

 

You can choose 50 and 50 if you like, but it really makes no difference, they are just numbers...

 

Andrew


  • Dunkstar likes this

#34 Kokatha man

Kokatha man

    Hubble

  • *****
  • topic starter
  • Posts: 15,072
  • Joined: 13 Sep 2009
  • Loc: "cooker-ta man" downunda...

Posted 28 July 2020 - 10:25 PM

...there has been some discussion on what exactly you should be setting those values to when using these colour cameras as essentially mono cams with ir filters...a couple of us did some tests with the ASI224MC back in its earlier days to see what differences there were: I think the focus was on Neptune captures at the time...can't remember who did the most trials but I wound up just leaving the sliders at the "standard" setting, accepting those weird colourations in the preview & AS!3 stages & converting to a mono image in R6 or P/shop Dunk...I don't think it was causing any noticeable problems that way & as I said, was mainly to avoid the Bayer grid pattern when trying to capture as mono...

 

Andrew, no problems with that revised r/b with higher red & lower blue you've come up with via your approach...I'll leave that to you but I don't think I'll divert from my post-processing druthers, which I readily admit are somewhat less "scientific" than your own approach & its rationale...even if we do sorta end up similar..! :lol:


  • Dunkstar and Tulloch like this

#35 Tulloch

Tulloch

    Vanguard

  • *****
  • Posts: 2,356
  • Joined: 02 Mar 2019
  • Loc: Melbourne, Australia

Posted 28 July 2020 - 10:32 PM

Andrew, no problems with that revised r/b with higher red & lower blue you've come up with via your approach...I'll leave that to you but I don't think I'll divert from my post-processing druthers, which I readily admit are somewhat less "scientific" than your own approach & its rationale...even if we do sorta end up similar..! lol.gif

No worries Darryl, my only point is that since you have to choose some values anyway, they might as well be correct (or close to it) from the start smile.gif.


Edited by Tulloch, 28 July 2020 - 10:55 PM.

  • Kokatha man likes this

#36 Dunkstar

Dunkstar

    Viking 1

  • -----
  • Posts: 680
  • Joined: 26 Mar 2012
  • Loc: Brisbane, Australia

Posted 29 July 2020 - 12:19 AM

Hi there, there is no "tainting" of the data - all colour cameras multiply their channels in order to give a more-or-less "correct" white balance. Since the sensitivity of the sensor does not match that of the human eye, using these values for Wred and Wblue can be used to approximate a true colour cast.

 

You can choose 50 and 50 if you like, but it really makes no difference, they are just numbers...

 

Andrew

Thanks Andrew waytogo.gif

 

I’m no doubt letting OCD cloud my judgement, but just want the raw data to be as was on the sensor, rather than it being laundered on the way through. 

 

Of course, I know that would mean having to change it in the digital domain to get the right colour balance, but that’s one of the last things I do.



#37 Tulloch

Tulloch

    Vanguard

  • *****
  • Posts: 2,356
  • Joined: 02 Mar 2019
  • Loc: Melbourne, Australia

Posted 29 July 2020 - 12:22 AM

Thanks Andrew waytogo.gif

 

I’m no doubt letting OCD cloud my judgement, but just want the raw data to be as was on the sensor, rather than it being laundered on the way through. 

 

Of course, I know that would mean having to change it in the digital domain to get the right colour balance, but that’s one of the last things I do.

No worries Dunk, you might be interested in having a look at the tutorials and writings of Christophe Pellier below

https://www.planetar...imaging.com/en/


  • Dunkstar likes this

#38 Isaac

Isaac

    Vostok 1

  • -----
  • Posts: 110
  • Joined: 11 Nov 2006
  • Loc: Barberà del Vallès, Barcelona, Spain

Posted 29 July 2020 - 05:13 AM

Thanks Darryl. R6 thinks that the colour channels need to be multiplied by (with normalised values in brackets).

 

R - 1.24 (1.32)

G - 0.94 (1.00)

B - 0.85 (0.90)

 

So, if you used Wred = 52, Wblue = 95 during the capture, I'd suggest that the "correct" values in FireCapture should be closer to Wred = 69, Wblue = 86 to give "correct" white balance straight out of the camera.

 

Andrew

I see white balance is a hard issue to handle. I've followed something like this, changing a little bit of the three channels, but I've got always the same color for rings and Saturn, so the complete image is very "yellow". Any hint whether I can get a better result? Of course I'll change white balance in FC on the next session!

 

UaRM2KZ.jpg

 

Btw, since this post is also about noise, I tried M57 and this is what I achieved from 1,5 second exposure time and almost 10 minutes (the image is cropped). Keep in mind I didn't have a clear night, there was some bad seeing (which was getting better, but I was forced to test the camera lol.gif ):

 

 xs2DVa6.jpg


Edited by Isaac, 29 July 2020 - 05:14 AM.

  • FelipeA likes this

#39 DMach

DMach

    Surveyor 1

  • -----
  • Posts: 1,650
  • Joined: 21 Nov 2017
  • Loc: The most light-polluted country in the world :(

Posted 29 July 2020 - 06:16 AM

Says the man who has (probably) never experienced a 'real' winter! lol.gif

 

Steve

True ... the winters in Melbourne (where I grew up) are miserable, but perhaps not "real".

 

I did have a holiday in Finland last December - does that count? lol


  • yock1960 and Tulloch like this

#40 Tulloch

Tulloch

    Vanguard

  • *****
  • Posts: 2,356
  • Joined: 02 Mar 2019
  • Loc: Melbourne, Australia

Posted 29 July 2020 - 07:05 AM

Hi Isaac, the colour balance I use for Saturn with my ASI224MC is similar to yours, which is based on the colour settings I developed for Jupiter (Wred = 62, Wblue = 99, Blue x 1.1 in Registax). Saturn is a yellowy green-ish planet, and using the Auto-balance setting in Registax doesn't work well (for me).

 

The colour settings I used for these planets are shown below, with the planets at a similar elevation in the sky, at a similar date and time and the same Wred and Wblue values used for both. I reckon they are colour accurate (or at least accurate enough for me lol.gif). I don't play around with the gamma setting in Registax, I may make a little shift in the central "level" in Photoshop, but not much.

 

Of course, these settings are just what works for me, you will probably find better settings for the elevation of the planets at your location, and adjust the colour balance to suit your own tastes.But I think that if you use the Auto-balance tool the Registax, and then use the same exact same colour shift values for Saturn (given that they will be at a similar position in the sky right now), you should be pretty close.

 

Nice image of the M57 by the way, that is one object that is pretty much out of my reach, given my location at 38*S.

 

Hope this helps.

 

Andrew

Attached Thumbnails

  • Registax Saturn RGB settings.jpg
  • Registax AutoRGB Jupiter settings crop.jpg

  • TorstenEdelmann, Isaac and KpS like this

#41 Tulloch

Tulloch

    Vanguard

  • *****
  • Posts: 2,356
  • Joined: 02 Mar 2019
  • Loc: Melbourne, Australia

Posted 29 July 2020 - 07:07 AM

True ... the winters in Melbourne (where I grew up) are miserable, but perhaps not "real".

 

Well, it felt pretty frick'n real last time I was out there freezing my fingers off lol.gif . OK, no snow (like the year we spent in Sheffield), but cold enough ...

 

Andrew


  • Dunkstar and DMach like this

#42 yock1960

yock1960

    Fly Me to the Moon

  • *****
  • Posts: 5,065
  • Joined: 22 Jun 2008
  • Loc: (Crossroad of clouds) Ohio, USA

Posted 29 July 2020 - 09:29 AM

True ... the winters in Melbourne (where I grew up) are miserable, but perhaps not "real".

 

I did have a holiday in Finland last December - does that count? lol

Finland? Yes, that counts! I do know that we all get acclimated to our environment.  I remember moving from approx. 39N ( a bit south of where I am now) to 43N when I was a teenager and thinking the kids were nuts for not wearing coats at recess in Spring. More recently, when out in Arizona (@~5000ft elevation) and mid 40'sF , folks wearing jackets and gloves and I'm wearing shorts and a t-shirt! lol.gif

 

Steve 


  • Dunkstar, AstroDan2015 and gfstallin like this

#43 Isaac

Isaac

    Vostok 1

  • -----
  • Posts: 110
  • Joined: 11 Nov 2006
  • Loc: Barberà del Vallès, Barcelona, Spain

Posted 30 July 2020 - 05:32 AM

Hi Isaac, the colour balance I use for Saturn with my ASI224MC is similar to yours, which is based on the colour settings I developed for Jupiter (Wred = 62, Wblue = 99, Blue x 1.1 in Registax). Saturn is a yellowy green-ish planet, and using the Auto-balance setting in Registax doesn't work well (for me).

 

The colour settings I used for these planets are shown below, with the planets at a similar elevation in the sky, at a similar date and time and the same Wred and Wblue values used for both. I reckon they are colour accurate (or at least accurate enough for me lol.gif). I don't play around with the gamma setting in Registax, I may make a little shift in the central "level" in Photoshop, but not much.

 

Of course, these settings are just what works for me, you will probably find better settings for the elevation of the planets at your location, and adjust the colour balance to suit your own tastes.But I think that if you use the Auto-balance tool the Registax, and then use the same exact same colour shift values for Saturn (given that they will be at a similar position in the sky right now), you should be pretty close.

 

Nice image of the M57 by the way, that is one object that is pretty much out of my reach, given my location at 38*S.

 

Hope this helps.

 

Andrew

Thanks for your help!

 

Now I'm working with my data, and it seems I have my channels swapped. I realized the red one was pretty awful and the blue one was really good (well, seeing had a poor quality). I've used color mixer in Photoshop and "voilà". lol.gif

 

I'll plug my camera later and change my adjustments in FC. 

 

Thanks again.



#44 Dunkstar

Dunkstar

    Viking 1

  • -----
  • Posts: 680
  • Joined: 26 Mar 2012
  • Loc: Brisbane, Australia

Posted 31 July 2020 - 05:46 AM

Well, it felt pretty frick'n real last time I was out there freezing my fingers off lol.gif . OK, no snow (like the year we spent in Sheffield), but cold enough ...

 

Andrew

 

 

Finland? Yes, that counts! I do know that we all get acclimated to our environment.  I remember moving from approx. 39N ( a bit south of where I am now) to 43N when I was a teenager and thinking the kids were nuts for not wearing coats at recess in Spring. More recently, when out in Arizona (@~5000ft elevation) and mid 40'sF , folks wearing jackets and gloves and I'm wearing shorts and a t-shirt! lol.gif

 

Steve 

This will sound funny to you guys then... first time I went to an astro gathering a couple of hours inland from the coast here at 26*S, I arrived to a beautiful sunny afternoon at about 25*C...but was woefully unprepared for what happened when the Sun went down. I could not change clothes quickly enough. Several hours into the night I was wearing 7 layers on my upper body and 4 layers on my legs, I could barely walk....and I still felt cold unless I did actually walk undecided.gif  It dropped to -6*C that night.

 

Of course, the folk that cling to the coast here have no idea how hostile it can be such a short distance away, and you see people wearing puffer jackets when it drops to 21*C lol.gif


Edited by Dunkstar, 31 July 2020 - 05:47 AM.

  • eros312 and Tulloch like this

#45 Dunkstar

Dunkstar

    Viking 1

  • -----
  • Posts: 680
  • Joined: 26 Mar 2012
  • Loc: Brisbane, Australia

Posted 31 July 2020 - 11:07 PM

Have to say I'm finding the 462 to be a bit on the pale side, especially with Saturn... this is after a bit of a saturation push undecided.gif

 

And I didn't get the colour fringing on the rings with 290 mono. Maybe unsurprisingly. Doesn't seem apparent with Jupiter though.

 

Incidentally, what's the accepted norm for sharpening?

Attached Thumbnails

  • Dunk-Saturn-462test.png

Edited by Dunkstar, 31 July 2020 - 11:07 PM.

  • eros312, KpS, gfstallin and 4 others like this

#46 calypsob

calypsob

    Fly Me to the Moon

  • *****
  • Posts: 5,982
  • Joined: 20 Apr 2013
  • Loc: Virginia

Posted 01 August 2020 - 06:51 AM

Im still wondering if this thing has ampglow

#47 Dunkstar

Dunkstar

    Viking 1

  • -----
  • Posts: 680
  • Joined: 26 Mar 2012
  • Loc: Brisbane, Australia

Posted 03 August 2020 - 06:18 AM

Im still wondering if this thing has ampglow

Yes - as most Sony sensors do - but it's slight. Reminds a lot of the Panasonic sensor in the ASI1600, in that it's just a bit of a glow in the corners/down the sides.

 

This is a small stack of shots taken at unity gain, 17x 30 second subs, 16-bit mode, stretched using a STF in PixInsight which makes it look a lot worse than it is. I'd imagine it'd calibrate out quite nicely. Just be aware that these shots were taken at a room temperature of 22*C.

Attached Thumbnails

  • DARK-462-unity-17x30s.jpg

  • eros312 and calypsob like this

#48 troyt

troyt

    Apollo

  • -----
  • Posts: 1,127
  • Joined: 19 Nov 2008
  • Loc: Australia

Posted 03 August 2020 - 09:16 PM

I was comparing the dynamic range of the two camera's and this is what I found.

Formulas used and for number checking.

 

calculated range = well dept/noise

 

LOG(calculated range)*20 = dynamic range in db

 

ASI462mc                                                                  ASI224mc

12000 / 0.5 = 24000                                                  19200 / 0.8 = 24000

 LOG(24000)*20 = 87.6 db                                        LOG(24000)*20 = 87.6 db

 

So it appears the two camera's have the same dynamic range!

 

Sony's internal measurement of signal to noise ratio which they call SNR1s  supprised me a little bit for the asi462 the SNR1s value is 0.18lx and for the asi224 the value is 0.13lx 

for the imx224 sensor the value is really hard to find online now but this is what I remember.

If a smaller value indicates better picture quality at low illumination then Sony's SNR1s score has the asi224 slightly ahead. 

 

The ASI462mc has better noise, frame rate(@640 x 480 not by much so no real world difference just a number) and HCG mode which means you can crank up the gain and not loose much if anything in doing so. 

The ASI224mc has better SNR1s value, known QE performance and proven track record in planetary imaging. 

 

Draw your own conclusions lol.gif but I thought it was interesting, tried to be unbiased and hopefully it will be useful for somebody. smile.gif  wishing to purchase either camera. waytogo.gif


Edited by troyt, 04 August 2020 - 01:26 AM.


#49 Dunkstar

Dunkstar

    Viking 1

  • -----
  • Posts: 680
  • Joined: 26 Mar 2012
  • Loc: Brisbane, Australia

Posted 03 August 2020 - 10:14 PM

That the dynamic range is achieved with 60% smaller pixel area is where the progress is demonstrated. Does the SNR1s take into account pixel size?

 

The 224 is the benchmark camera, no question about that.


  • troyt likes this

#50 troyt

troyt

    Apollo

  • -----
  • Posts: 1,127
  • Joined: 19 Nov 2008
  • Loc: Australia

Posted 04 August 2020 - 01:51 AM

I guess if considering both a imx224 and imx462 based camera I would ignore dynamic range and frame rate and make the decisions on other factors

because there doesn't seem to be much difference in these - unless the person thinks every frame counts. wink.gif

 

From memory Dunk Sony does make some effort to reduce the effects of different pixel size before testing but there still may be a differences after this - how much not sure. But Sony does seem happy

to display and compare there own sensor this way. I think it's a good guide at least. 

 

At the end of the day that is the big question how much real world difference is there?

I don't know and I think there is an effort here to try and sort that out. smile.gif


Edited by troyt, 04 August 2020 - 03:36 AM.



CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics