After a couple of cloudy weeks, I had the chance of taking some new images, this time of the flaming star nebula IC 405. The images are not processed, yet, but after some updates to PixInsight I again gave the Weighted Batch Processing Script a chance and stacked the images as OSC to compare results and processing time with SiriL.
My PC has 16 GB RAM and an AMD FX 8320 eight core @ 3.5 GHz.
The images were taken with my ZWO ASI 294 MC pro, 4144x2822 px, size about 22.3 MB, stored as FITS files.
The set consists of
- 40 bias frames
- 18 dark frames
- 15 flat frames
- 41 light frames
Stacking parameters in PixInsight and in Siril were set identical or at least as close as possible. For the image calibration and stacking of the lights, the following parameters were set:
- subframe weighting with PSF signal weight,
- local normalization,
- image integration with average / linear fit clipping.
Results
Processing time PixInsight: 40 min 7 s
Processing time SiriL: 9 min 55 s
Some actions in detail:
| Action | PixInsight | Siril |
|---|---|---|
| Integration bias frames | 54 s | 21 s |
| Integration dark frames | 27 s | 10 s |
| Calibration flat frames | 12 s | 14 s |
| Integration flat frames | 18 s | 5 s |
| Calibration light frames | 39 s | 1 min 19 s |
| Debayer | 2 min 36 s | (integrated in calibration) |
| Registration | 4 min 6 s | 5 min 24 s |
| Local normalization | 21 min (with reference gen.) | 1 min 44 s |
| Integration light frames | 7 min 59 s | 2 min 16 s |
Sure, this is only one example, one measurement point - and the parameters might not be exactly the same.
What about the image results? I extracted the red channel for better visualization of the Ha signal:
Resulting image from PixInsight
Resulting image from SiriL
Even a subtraction of the images hardly reveals any difference. It is less than 0.1% of the average signal values.
This shows why I am using SiriL for stacking, not PixInsight. Typically, I use one of the processing scripts and repeat the image integration with modified parameters on the results light sequences. Even with this additional integration pass, I am much faster than with the Weigthed Batch Processing Script.
The PixInsight solution is only used e.g. when I have several sets from different nights with different flats. But this could also easily be scripted in SiriL.
But after stacking - PixInsight is the tool of choice. SiriL is only used for stacking.
Add comment
Comments