Blown Out M42 Core - Need Advice on Sigma High & Low Values

[jwaters125]

Age old problem...  The core of my wide-field Orion image is blown out.  I have 52 Light subs at 80 seconds each with my Canon 6D and 200mm lens.  It was my intention to take subs at 30, 50 and also 110 seconds to better show the color and detail but the weather has been poor at best.  I also wanted to take many more subs at each length.

The steps I used are as follows.
- Batch PreProcessing - Default Values Sigma High and Low values, Winsorized Sigma Clipping
- Image Integration - Default Sigma Values, Stacking Average, Winsorized Sigma Clipping
- Screen Transfer Function
- Dynamic Batch Preprocessing
- Photometric Color Calibration
- Multiscale Median Transform
- Histogram Transformation
- HDR Multiscale Transform
- Masks - To bring out details
...etc

The problem I have is that a single Canon 6D CR2 image shows more M42 color detail than the PI stacked, stretched and processed image. 

Would changing the Sigma High / Low during BPP and/or ImageIntegration retain more of the color and detail in the core of M42?

JPEG Image - https://flic.kr/p/QJ8x2M

[aworonow]

M42 is a prime (and rare) example of an astronomical target whose brightness range is greater than the dynamic range covered by our cameras. The best images of M42 require a short and a long (relative, speaking) to cover all that range. Then the HDR capabilities in PI are used to blend the two separate stacks together.

Alex

[jwaters125]

Thanks Alex - Yep, I understand the dynamic range requirement of M42 is a challenge.  I was hoping that changing the Sigma High and Low values would bring out or retain more of the colors that the RAW CR2 images show.  IMHO stretching (STF) destroys some data.

This year's weather has been poor here in Arizona.  Hopefully I will be able to go out in March and collect the rest of my data.

[pfile]

maybe try MaskedStretch for the initial stretch?

rob

[jwaters125]

I will have to look into that.

What about stretching less with STF - (33 and 66%) then 100%, processing each image set then adding them together with HDRComposition?

[pfile]

you should use HDRComposition on linear images, so don't stretch before running that. maybe do DBE and then HDRC and then start processing the HDR image.

rob

[jwaters125]

Not sure I have any other options.  I only have one data set.  Stretching (STF and HT) looks like it destroys some color detail in relation to my CR2 RAW's.  For now I would like to just regain some of that data.

[pfile]

oh i see what you meant, well, it might work, but i'm not sure HDRComposition was designed for nonlinear images.

there are tools out there that can do HDR composites of stretched data. enfuse comes to mind.

rob

[ngc1535]

The problem I have is that a single Canon 6D CR2 image shows more M42 color detail than the PI stacked, stretched and processed image. 

JPEG Image - https://flic.kr/p/QJ8x2M

I think something that is being neglected is the above statement. If the "detail" or information is contained within a single subframe- then it should be possible to combine the information without extraordinary blending. Now, if that single frame is *not* good enough- then acquisition strategies likely need to be adjusted.

It seems to me that HRDMT can go a long way towards helping in this particular situation.

-adam

[jwaters125]

Yes - HDRMultiscaleTransform does recover the inner structure of the core but the colors are muted or bleached after stretching.  The RAW CR2 images have more color on the outer sections of the M42 core.  The very core is blown out which I would expect. 

Canon CR2 RAW converted to JPEG
https://flic.kr/p/2eRDLV7

Final PI Processed Image
https://flic.kr/p/QJ8x2M

After ImageIntegration there's a lot of information being rejected in 'rejection_high' and some in 'rejection_low' within the M42 core.  This may be core structure and color detail.

[msmythers]

One thing is that color like brightness can be stretched and using the right tools can help. In areas where you don't have saturated pixels your color information is still there.

Here I used your processed image and applied ColorSaturation through a mask being careful of the background and stars. I used SCNR to remove the green tinting in some areas. This all done to a jpg.


Mike

[msmythers]

Also reducing the luminance shows the process image to look more like the single CR2 image that shows little brightness but has color.


Mike

[jwaters125]

Many thanks Mike...