Please let me know if anyone sees any errors or has any additional advice they think helpful.
I will update this post as needed.
For an index of similar notes on the other StarTools modules see StarTools Main Window Use.
Flux Module
Purpose:
- To de-noise, sharpen or augment DSO detail
For a general overview see Flux: Automated Astronomical Feature Recognition and Manipulation.
The Fractal Flux module uses an algorithm which pin-points features in the image by looking for natural recurring fractal patterns that make up a DSO, such as gas flows and filaments. Once the algorithm has determined where these features are, it then is able to modify or augment them.
Useful Sources
There are a number of useful links in StarTools Links and Tutorials.
The notes below relate to StarTools version 1.5, 1.6, and 1.7
When to use:
- The Flux module works best on images of gas clouds, flows, knots, shock waves - where self-similarity at different scales is high (see below).
- The Flux module will not work well on images dominated by stars (e.g.wide field images where any gas clouds occupy a small area).
- Use after extracting the most detail (i.e. Decon, Sharp Contrast, HDR) and before using the Color module.
- The noise reduction capabilities of the Flux module can be used in place of the Denoise module where Tracking is not used (for example where an image is not linear to begin with and StarTools is being used just to touch up an image).
- It could be a decent alternative to the Denoise module after working in the Layer module with Tracking off. In particular, for fine noise due to the way the self-similarity works.
- Only use once.
{Compose}-AutoDev-{Lens}-Bin-Crop-Wipe-AutoDev (or FilmDev)-{Contrast/HDR/Sharp/Decon/Flux}-Color-{Shrink/Filter/Entropy/SuperStr}-Track/NR-{Layer/Heal/Repair/Synth/Stereo 3D}
Key: {...} optional modules
Example Workflow (v1.8):
{Compose}-AutoDev-{Lens}-Bin-Crop-Wipe-AutoDev (or FilmDev)-{Contrast/HDR/Sharp/SVDecon}-Color-{Shrink/Filter/Entropy/SuperStr/NBAccent/}-Track/NR(Unified-Denoise)-{Flux/Repair/Heal/Layer/Synth/Stereo 3D}
Key: {...} optional modules
Method:
This is a way of using the module which should give good results in most cases:
- Set a mask if an appropriate one isn't already loaded - An inverted star mask is recommended with this module - Auto, Stars, Do, Invert, Keep.
- Select the Wavelet Library size - after a while you will be presented with Screen 2.
- Select the desired preset - normally choose Sharpen or Noise - avoid using Detail except in special circumstances.
- Keep the result when done.
- Any change may be subtle - Use Before/After button to see the change.
- If the 'Sharpen' preset is chosen the image will look sharper in the areas where there is detail.
- If the 'Noise' preset is chosen the noise in the image will be reduced.
Background Noise Reduction
This approach was described by elpajare here
Fractal flux can help remove background noise. Using the Noise preset and appropriately adjusting the settings can help a lot. This is an example of the settings used for an open star cluster and the result with a good contrast to see the details:
- Click the 'Noise' Preset
- Adjust the Positive flux as needed. This increases or decreases the bright zone. If we lower it, the stars are rounded a little bit.
- Adjust the Negative flux as needed. This increases or decreases dark areas. If we increase it, the stars are compressed a little and become rounder.
- Adjust the Filter amount. Set the intensity of the filter, lowering it until the brightness doesn't make the star's core too big.
- Adjust the Filter radius. This determines the radius up to which the filter will act. The larger the radius, the larger the affected radius of the star.
- Adjust the Filter fuzz. This eliminates noise according to the size of the pixel we set.
Screen 1
Wavelet Library
This specifies the amount of 'knowledge' the fractal flux module starts off with. The smaller the amount of knowledge the more flux you might get.
- Small. The smaller the amount of knowledge the more flux you might get.
- Large. A larger wavelet library will make the module better at relating detail to other detail at larger scales (essentially making it more precise), but will also make the effect a lot less "dramatic", especially when noise is involved.
- Default is Small.
Sharpen preset
This button optimises the settings for sharpening of the image.
Detail preset
This button optimises the settings for adding natural-looking detail to the image.
- This adds artificial detail but...
- It can distort stars so using a star mask to mask out stars is advisable.
- The detail added is an 'educated guess'. Adding this detail reduces the documentary value of the image.
This button optimises the settings for removing noise from the image.
Mask
For general instructions on using masks see Mask.
- Recommended to use an inverted star mask with this module - Auto, Stars, Do, Invert, Keep.
This selects the filtering algorithm to use. Choose from:
- Add Detail - Used to add detail to the image.
- Modulate Unsharp Mask - the presence of flux turns on and off an Unsharp Mask sharpening algorithm. Effectively performs a localized sharpening of only ‘interesting’ features, while leaving others (including noise) alone.
- Modulate Gaussian
- Modulate Median
- Modulate MOMH - Mean of Median Half - 25% upper and lower outlier rejection, taking mean of remaining values - Used to reduce noise.
- Modulate Undo Buffer- Use when you want to combine an image with another image using the flux to modulate.
- Default is Modulate Unsharp Mask [Sharpen], Modulate MOMH [Noise], Add Detail [Detail]
- Use the default setting unless you understand and need the effects the other settings give.
Controls the amount any positive flux is amplified
- Default is 500%, 25% [Sharpen], 600% [Noise], Maximum is 1000%
Controls the amount any negative flux is amplified:
- Default is 0%, 600% [Noise], Maximum is 1000%
Sets the minimum allowable size of any flux concentrations.
- Increasing this value allows only the bigger clumps of flux through.
- Decreasing this value allows more and more smaller, detailed flux to pass through.
- Default is 1.0 pixels, 2.5 pixels [Sharpen], Range is 1.0 to 40.0 pixels.
Sets the power that should be applied to a pixel in the luminance mask if one is set in Brightness Mask Mode.
- Default is 1.00, 3.00 [Noise], Maximum is 5.00.
Sets the mode of the optional masking using the luminance information. Options are:
- Off
- Where fg is dark, use bg
- Where fg is light, use bg
- Where fg is light & dark, use bg
- Where fg is grey, use bg
- Default is 'Off', 'Where fg is light, use bg' [Noise]
Sets the strength of the Unsharp Mask sharpening.
- Default is 200%. 100% [Noise], Maximum is 1000%.
Specifies the maximum size of any details to be sharpened by the Unsharp Mask algorithm (if used) or the filter width for the other algorithms.
- Default is 1.5 pixels, 12.5 pixels [Noise], Range is 1.0 to 40.0 pixels.
- The larger the radius the larger the details that are sharpened.
Specifies the kernel radius of an optional Gaussian blur, to be applied to the flux before it is passed to the Unsharp Mask routine for modulation.
- Default is 1.0 pixels, Range is 1.0 to 40.0 pixels.
- Increasing this value effectively 'smears out' the flux over a larger area so the detail around the flux is also sharpened.
- As the flux is spread out its intensity becomes lower. You may compensate for this by increasing the Positive Flux and Negative Flux amplifiers, or by increasing the Filter Amount (sharpening strength) parameter.
If a mask is used, Mask Fuzz controls the blending of the transition between masked and non-masked parts of the image.
- Default is 1.0 pixels, Range is 1.0 to 40.0 pixels
The Flux module uses fractal analysis to look for structures that are similar at different scales and identifies these as detail. It builds up a map of areas of which are more likely to be detail and which are less likely to be detail. It can use this map to identify areas to target for augmenting detail, sharpening or noise reduction.
The Flux module works best on images of gas clouds, flows, knots, shock waves - where this self-similarity is high.
The Flux module will not work well on wide field images where any gas clouds occupy a small area on the image which is dominated by stars. Stars are point lights and don't look consistent across different scales.
The Flux module can augment detail by analysing the images larger scale structures and triying to apply what it has learnt to the smaller structures. This works best where the similarity at different scales is high - as in gas clouds.