First M42 - Orion. Strange noise in image

[joecoyle]

Hi

Below you can see my first LRGB image of M42. 10x2min each of LRGB for 80 minutes of exposure. 

Two things to note, 1) I know it's quite over processed, but that was to highlight the noise issue. 2) the nebula has filled my entire sensor. 

My issue surrounds all the pixels of noise. What could have caused them? It spoils for me what was in my opinion a good first attempt.

I processed in PixInsight, following Kayron's @Gib007 excellent tutorial on combining monochrome images, either i did something wrong in PI or my camera had a wobble! 

Thanks

Joe

[Oddsocks]

As a last stage in PI, when the image is pretty much as you want it, open the ACDNR tool from the Noise Reduction modules and apply it to the image, this will remove the white speckle noise.

The default module settings should be good, for more help see the PI website for the ACDNR tutorial, or other on-line help, Harry has a good video tutorial of ACDNR on his web site in the newbie's section using the ACDNR tool earlier in the processing stream using masks, for your image where the noise is most obvious in the bright area try applying to the whole image without a mask.

http://www.harrysastroshed.com/pixinsight/pixinsight%20video%20html/pixinsighthomenewbie.html

 

Edited February 24, 2016 by Oddsocks
Added process details

[Oddsocks]

Realised I did not answer fully your question.....

The random pixel noise is an inherent part of any image captured by a CCD or CMOS detector.

When you calibrate your light frames by subtracting darks and bias most of the noise in the image is removed but any that remains will be enhanced along with the image once you start applying stretches and curves. 

More data for each channel will help reduce the background noise and any that remains can be removed using the noise reductions tools in PI.

In your posted composite image it is not possible to say where the noise originated, you would need to look at and assess the luminance, colour channels, darks and bias individually to make a diagnosis.

If your un-calibrated luminance and colour channel light frames are relatively noise free but the noise appears after calibration then try using the bias frames only to calibrate the darks and then calibrate the lights only with the darks leaving the bias out.

HTH.

[ollypenrice]

One thing that isn't helping is imperfect registration of the colour channels. Your stars have red left hand edges and green right hand ones.

If my regular stacking software (AstroArt) doesn't get it perfectly correct I align the colour channels in Registar, which I find infallible. I presume you aligned in PI? While it is very capable it is also possible to make a bad choice of parameters, I thnk. (I don't stack in PI.)

Olly

[joecoyle]

Hi Olly

Yeah I aligned in PI. I just left all the parameters at the default. I haven't got around to experimenting with them all yet. I did wonder if my alignment was off slightly. I shall try out Registar and see what difference that makes. 

 

ACDNR fixed the noise issue which was good. Potentially my own stupid fault as looking back, I didn't have the CCD cooler on! DOH!! 

 

Next job is to try and merge two images. One of the centre so it's not so blown out and one of the nebulosity. Any ideas appreciated. 

 

Thanks

Joe

[ollypenrice]

58 minutes ago, joecoyle said:

Hi Olly

 

Next job is to try and merge two images. One of the centre so it's not so blown out and one of the nebulosity. Any ideas appreciated. 

 

Thanks

Joe

I follow this method. It's the best I know of. http://www.astropix.com/HTML/J_DIGIT/LAYMASK.HTM

PI have an HDR Wavelets demo on M42 but I'm afraid I don't like the result at all. It has that 'human brain' look which afflicts so many HDR processed images.

Olly

[pipnina]

47 minutes ago, ollypenrice said:

I follow this method. It's the best I know of. http://www.astropix.com/HTML/J_DIGIT/LAYMASK.HTM

PI have an HDR Wavelets demo on M42 but I'm afraid I don't like the result at all. It has that 'human brain' look which afflicts so many HDR processed images.

Olly

I think the reason that "brain effect" occurs is because they don't darken the more highly exposed images and so they occupy the same brightness ranges as the lower exposed images designed to capture brighter detail...

I dabbled in manual HDR a few times, and often I would give up on an image because of this effect. But sometimes it would come out perfectly... Examples:

These were taken some years back when I had a lot less experience with photography so the angles/DOF etc isn't there but the processing & final image quality is pretty good! But these 3 images were pot-luck, HDR is quite difficult to do right but pays off if it works IMO.

    ~pip

[joecoyle]

Hi Olly

Does that tutorial work for CCD images as well?

I'm curious about registar... do you use it regularly to register images? I downloaded the trial, but since the trial doesnt let you save images, I cant actually see if those red/green edges from my M42 have gone. And i really dont want to shell out £180 for Registar if it doesnt make a massive difference! What are your opinions of it?

Thanks

Joe

Edited February 26, 2016 by joecoyle
Extra question

[andrewluck]

There's some confusion crept in here. PI has two processes; HDRMultiscaleTransform(HDRMT), which compresses the dynamic range in an image and HDRComposition, which combines images of different exposure lengths to produce an HDR image.

I've found that HDRComposition produces excellent results with very little effort. The default parameters usually work fine. HDRMT is an acquired taste and requires some effort to get the best results from it. It is very effective in what it does though.

I don't see the colour fringing in the stars on the right of the image so I suspect that this isn't due to the registration.

Andrew

[ollypenrice]

On 26/2/2016 at 18:46, joecoyle said:

Hi Olly

Does that tutorial work for CCD images as well?

I'm curious about registar... do you use it regularly to register images? I downloaded the trial, but since the trial doesnt let you save images, I cant actually see if those red/green edges from my M42 have gone. And i really dont want to shell out £180 for Registar if it doesnt make a massive difference! What are your opinions of it?

Thanks

Joe

Yes, the Lodigruss tutorial works fine with CCD. I Made 3 RGB exposure sets (about 11 seconds, 50 seconds and 600 seconds from memory) and combined them that way. Ditto for luminance. I layer masked them into a single RGB and a single L and comined them. As well as being an effective technique I find it a particularly enjoyable one to use. I was going through it with a guest recently and he felt the same.

I don't use Registar as a stacking tool because it doesn't facilitate calibration, but I use it endlessly to prepare mosaics, align problem colour channels, combine data from different cameras and scopes, combine different focal lengths etc.

If you like you can send me your image and I'll split the channels in Ps and align them all in Registar then send them back to you to see what you think. I can't promise any kind of result but I think it will probably help.

Olly

[Stub Mandrel]

Registax has a very effective RGB align function, even if you don't use it for anything else. Just darg and drop a tiff or jpeg onto it. It also has a manual mode.

I tried it on your image, but the stars at the right hand side need a greater shift to those on the left, not sure what you can do about that?

 

[Aussie Dave]

On 26/2/2016 at 01:46, joecoyle said:

Hi Olly

Does that tutorial work for CCD images as well?

I'm curious about registar... do you use it regularly to register images? I downloaded the trial, but since the trial doesnt let you save images, I cant actually see if those red/green edges from my M42 have gone. And i really dont want to shell out £180 for Registar if it doesnt make a massive difference! What are your opinions of it?

Thanks

Joe

A little tip Joe. If you have a large monitor, set it to maximum resolution and with your work finished in Registar take a screenshot/capture, the paste in your image editor, then crop. For larger works if you want full resolution results, take screenshots of the full res image on each area, overlap the screen shots, then stitch together.