27 hours on M27

[ollypenrice]

19 minutes ago, Laurin Dave said:

 Also when I look at your processed images I see no signs of the Oiii being saturated.   

HTH 

Dave

 

I had another look at Dave's posted OIII stretch and it's certainly saturated in the same place, Dave. The screen grab goes off the scale in Curves if you mouse over it. (I've found that screen grabs are quite reliable where the histogram is concerned. I've tried uploading and screen-grabbing an image of my own to compare the grab with the original and found them equivalent.)

However, I'm puzzled by why this should have happened, especially in the light of the comparison you make with your own linear data. Maybe it has had a stretch somewhere? Or is this some CMOS mystery like gain? Does the FITS header tell you what the gain was? I'm just wondering if it might accidentally have been set too high.

Olly

[DaveS]

Hi and thanks.

I have my cousin and her hubby coming for a socially distanced meet while it's still legal, so it might be this afternoon before I can have a bash at the data.

[Laurin Dave]

14 minutes ago, ollypenrice said:

I had another look at Dave's posted OIII stretch and it's certainly saturated in the same place, Dave. The screen grab goes off the scale in Curves if you mouse over it.

Maybe it has had a stretch somewhere? Or is this some CMOS mystery like gain? Does the FITS header tell you what the gain was? I'm just wondering if it might accidentally have been set too high.

Olly

Agreed Olly, the shot Dave posted of the Oiii stack is saturated in the core but his original NHO seems to me not to be..  I also wonder about the gain used, maybe Dave used high gain which compresses the dynamic range ..  I've only ever used what's known as Unity Gain.  Some folk recommend using higher gain for narrowband with the ASI1600 but maybe as M27 is so bright it has caused an issue.  FITS header will display the gain... need a FITS for that though

Dave

[ollypenrice]

47 minutes ago, Laurin Dave said:

Agreed Olly, the shot Dave posted of the Oiii stack is saturated in the core but his original NHO seems to me not to be..  I also wonder about the gain used, maybe Dave used high gain which compresses the dynamic range ..  I've only ever used what's known as Unity Gain.  Some folk recommend using higher gain for narrowband with the ASI1600 but maybe as M27 is so bright it has caused an issue.  FITS header will display the gain... need a FITS for that though

Dave

True. I think the N and H must be showing through the core which is saturated in O. That wouldn't work if the O were used as luminance but it will be getting only about a one third weighting if combined as a colour channel. I reckon even just one non-saturated O core would save the day.

Olly

Edited October 6, 2020 by ollypenrice
clarification

[DaveS]

OK, @ollypenrice and @Laurin Dave, I think the problem occured when I saved the files as 16 bit TIFFs. The originals are 32 bit FITS, so when I saved as TIFFs the high values were truncated.

Here are the original 32 bit FITS if your software can handle them.

9 Hours H.fit

9 Hours N.fit

9 Hours O.fit

I can do a Sigma Average stack which will produce 16 bit FITS which won't truncate.

[ollypenrice]

19 minutes ago, DaveS said:

OK, @ollypenrice and @Laurin Dave, I think the problem occured when I saved the files as 16 bit TIFFs. The originals are 32 bit FITS, so when I saved as TIFFs the high values were truncated.

Here are the original 32 bit FITS if your software can handle them.

9 Hours H.fit 62.53 MB · 1 download

9 Hours N.fit 62.53 MB · 0 downloads

9 Hours O.fit 62.53 MB · 1 download

I can do a Sigma Average stack which will produce 16 bit FITS which won't truncate.

I'm speaking sweetly to my internet connection and giving it all the encouragement I can. I may be some time...

🤣lly

[DaveS]

Ha yes, sorry they are a bit big.

When I opened the FITS file in AA7, I could see structure that was blocked up in the TIFF file.

[ollypenrice]

13 minutes ago, DaveS said:

Ha yes, sorry they are a bit big.

When I opened the FITS file in AA7, I could see structure that was blocked up in the TIFF file.

Well, it downloaded in no time (amazing!) but I have the same trouble as you with the compression. It looks lovely in AstroArt at 32 bit but saving as 16 bit white clips it.  The 32 bit TIFF won't open in PI for me. This is beyond me. (Not difficult with IT issues...)

Can anyone help? So frustrating, the stack's lovely.

Olly

Edited October 6, 2020 by ollypenrice

[Laurin Dave]

Similar issues here Dave and Olly..  They opened in Pi as a grey image all at 0.5,...  Used APP to convert it to a 32 bit TIFF which Pi could then open properly.

here's the TIFFs I created...  hopefully they'll open properly for you Olly

Dave

9_Hours_O__1_-St_16bit.tiff9_Hours_H__2_-St_16bit.tiff

[DaveS]

4 hours ago, Laurin Dave said:

Agreed Olly, the shot Dave posted of the Oiii stack is saturated in the core but his original NHO seems to me not to be..  I also wonder about the gain used, maybe Dave used high gain which compresses the dynamic range ..  I've only ever used what's known as Unity Gain.  Some folk recommend using higher gain for narrowband with the ASI1600 but maybe as M27 is so bright it has caused an issue.  FITS header will display the gain... need a FITS for that though

Dave

Just to say I used zero gain, AKA the Max DR setting in the ZWO ASCOM driver.

[tomato]

Hi Dave, I loaded the TIFF files from @Laurin Dave into StarTools v 1.7 assigning Ha to green and OIII to blue.  I followed the default workflow but struggled to retain central detail while trying to bring out the outer shells. I think I have pushed the contrast and detail enhancement too far as a result.

Edited October 6, 2020 by tomato
File type corrected

[ollypenrice]

29 minutes ago, tomato said:

Hi Dave, I loaded the TIFF files from @Laurin Dave into StarTools v 1.7 assigning Ha to green and OIII to blue.  I followed the default workflow but struggled to retain central detail while trying to bring out the outer shells. I think I have pushed the contrast and detail enhancement too far as a result.

Piece of work! You're making me nervous but I now have workable OIII. Great job there...

Olly

[ollypenrice]

2 hours ago, Laurin Dave said:

Similar issues here Dave and Olly..  They opened in Pi as a grey image all at 0.5,...  Used APP to convert it to a 32 bit TIFF which Pi could then open properly.

here's the TIFFs I created...  hopefully they'll open properly for you Olly

Dave

9_Hours_O__1_-St_16bit.tiff 31.29 MB · 9 downloads 9_Hours_H__2_-St_16bit.tiff 31.29 MB · 11 downloads

Thanks Dave. Looked promising to start with (I only downloaded the OIII) but, while it isn't white clipped any more, it still isn't a happy file. DBE rejected everything I tried. Thinking that I didn't need anything but the nebula itself I took the file straight into Photoshop, aiming to dump everything but the nebula in the final blend. However, it won't stretch and falls apart when I try do stretch it, viz:

It's odd that I had no issues with the Ha file, which behaved normally, but I can get no sense out of the OIII. There has to be a clue in there. It can't be to do with the brightness at capture, it has to be to with filing/formats/other IT stuff I don't understand!

Olly

 

[DaveS]

This is a native 16 bit FITS from a Sigma Average stretch. Might have fewer problems.

9 Hours O Sigma Average.fit

 

[Laurin Dave]

52 minutes ago, ollypenrice said:

Thanks Dave. Looked promising to start with (I only downloaded the OIII) but, while it isn't white clipped any more, it still isn't a happy file. DBE rejected everything I tried. Thinking that I didn't need anything but the nebula itself I took the file straight into Photoshop, aiming to dump everything but the nebula in the final blend. However, it won't stretch and falls apart when I try do stretch it, viz:

It's odd that I had no issues with the Ha file, which behaved normally, but I can get no sense out of the OIII. There has to be a clue in there. It can't be to do with the brightness at capture, it has to be to with filing/formats/other IT stuff I don't understand!

Olly

 

Same here Olly..  its got me stumped, when I stretch it the Histogram becomes discontinuous ie vertical lines as if its low bit depth...   same with your most recent posted file Dave..   Could you post a single calibrated sub for each channel - it might shed some light on what's going on.

Time for Bake Off  

Dave

Edited October 6, 2020 by Laurin Dave

[DaveS]

Here you are

Single O Sub Cal.fit

Single H Sub Cal.fit

As it happens I've noticed the low end of the histogram "combing". I ought to check the settings, that I wasn't inadvertently capturing 8 bit data, but I haven't rebooted the platform computer since a power cut a couple of days ago. Will do it tomorrow and check the settings.

[DaveS]

When I look at the [OIII] Sigma Average stack in AA the histogram says 12 bit, but looking at the Sigma Add stack it says 17 bit, hence why the FITS is 32 bit rather than 16 bit.

My head hurts.

My next big purchase will be a QHY 268, which does, at least, have a decent 16 bit ADC and 14+ stop DR.

But this is no nearer solving the problem or producing a believable image.

 

Edit: The single calibrated sub says 9 bit on the histo. Now my head really *does* hurt.

Edited October 6, 2020 by DaveS

[Laurin Dave]

Can't be certain Dave as I've no way of measuring it but a single sub of yours looks more granular (so I assume lower bit depth) than one of mine (which are at most 12 bit)..   I'd check your capture settings and also whether the same issue affects your flats and darks

Dave

[x6gas]

Hmmm.  This is an interesting one.  I should say that I am stuck with AstroArt 5 so things may be different with more modern releases, but:

If you click on View>Range>CCD range then you get the burnt out image that Dave posted originally.  This is strange as whenever I've done this with my data it gives the complete CCD range (as you'd expect from the name of the function) but that's not the case here - you get a very different result if you select View>Range>Min->Max... so saving that gives this (which behaves OK when I stretch it in PS):

9 Hours O.tif

so I wonder if AstroArt is somehow misreading or misinterpreting the bit depth of the camera?

Ian

[andrew s]

4 hours ago, DaveS said:

Just to say I used zero gain, AKA the Max DR setting in the ZWO ASCOM driver.

At this gain you get 5 electrons per ADU.  This level of quantisation noise may be your issue.  If your nowhere near saturation your throwing bits away. I think you would be better off taking more shorter exposures, to avoid saturation,  at say 1 electron per adu (gain 139).

Regards Andrew 

[Laurin Dave]

Dave..  further to Ian’s comment probably worth posting uncalibrated subs

[DaveS]

This is the histogram I get of the 9 hour unstretched stack

[steppenwolf]

What is even more strange is that using Laurin Dave's O TIF, I am not getting this issue in Photoshop unless I am not trying hard enough?

[DaveS]

1 minute ago, Laurin Dave said:

Dave..  further to Ian’s comment probably worth posting uncalibrated subs

OK, single sub? or stack?

[DaveS]

OK, single uncalibrated Oxygen sub.

Dumbbell-Oxygen-0018.fits