Hi Scott,
I ran an experiment recently with combining multiple exposures of M42. I was curious to see the difference between the resultant combination of the same data using layer masking verses using weighted layer averaging. My take on the outcome was pretty much what I suspected in that the layer masking technique gave a better result because it preserved more of the low level data present in the longer exposures.
Here's the workflow I used for the weighted average experiment (see attached sample image: M42_Composite_Flat_Weighted_Layers_Average.jpg).
As mentioned above, I've used masks before when combining the trapezium of M42, but tried a different method this time to see if it would reduce the (sometimes) noticeable transitions between the different exposures. This alternative hybrid technique goes back to the way stacking was done years ago using Photoshop before auto stacking software was available. The workflow I used was the following:
- Take a series of exposures over the desired range. For that M42 image I took 20 shots each at 30 sec, 15 sec, 7.5 sec, 4 sec, 2 sec, 1 sec, and 0.5 sec.
- Align & stack each set of exposures using automated software. For that M42 image I used Images Plus software for auto alignment and stacking.
- I saved each image in the set to uncompressed TIFF (32bit float) and opened each image as a layer in Photoshop.
- The Photoshop layers were then organized (ordered) with the longest exposure time image as the bottom layer to the shortest exposure time as the top layer.
- The Photoshop layers were then manually aligned to one another using the stars in each image as reference points.
- The Opacity of each Photoshop layer was then set as follows:
- 0.5 sec layer, Opacity = 14%
- 1 sec layer, Opacity = 17%
- 2 sec layer, Opacity = 20%
- 4 sec layer, Opacity = 25%
- 7.5 sec layer, Opacity = 33%
- 15 sec layer, Opacity = 50%
- 30 sec layer, Opacity = 100%
- (Note: The Opacity setting is 1 divided by the number of total number of layers in the stack at that point in the stack, starting from the bottom.)
- The Photoshop layers were then flattened into one layer and Levels and Curves were used to stretch the histogram of the composite image.
I've also attached a sample image from the same data but this time using the using layer masking technique (see: M42_Composite, Flat, Layer Masking, w-Txt.jpg). Note: I tend to use a very soft brush for the masks and sometimes I reduce the opacity of the shorter exposure layers to soften the blending transitions.
Please excuse the elongated stars in the sample images. The M42 data was taken while testing a new planetary camera (ZWO ASI485MC) to see how it would do with a bright DSO. The mount was only roughly polar aligned and I wasn't tracking.
Clear skies,
Gary