Recently this YouTube video was posted in a forum thread on AstroBin where it’s suggested not to use dark frames for DSLR deep-sky astrophotography!
Since this would be really a huge time saver, I took the last weeks data of my Propeller Nebula session to create different stacking combinations to find out, if there is a reason for using darks. At least for this session data.
- You’ve to find out by yourself if dark frames improves your integrated image or not. It seems to be different for every camera sensor. Also different session circumstances like ambient temperature and exposure time will pretty sure have an impact to the result.
I tested all stacking combinations that are feasible for Lights, Darks, Bias and Flats. Even senseless combinations, only to see the result.
Captured with my modified Canon EOS70D, Canon EF 400mm f/5.6 lens and MGEN II guider (Dithering enabled).
Ambient temperature during shooting 13°C at session start, 11°C at the end.
All versions are calibrated, registered and debayered with WBPP script, then integrated with ImageIntegration process, AutomaticBackgroundExtractor and stretched with STF in PixInsight 1.8.8-8.
47 lights (180s), RAW temp. readout: 22-24°C
24 darks (180s), 20-23°C
39 bias (1/8000s), constantly 23°C
43 flats (1/100s), constantly 26°C
For a deeper insight, I uploaded all 8 versions in full resolution on AstroBin:
Relevant for me are only the versions with flat frames (revisions F, G and H on AstroBin). The question is, gives a Lights/Flats/Bias integration at least the same result as the full integration with Darks? That would save much time in the field since creating only flat and bias frames doesn’t need much time.
But surprisingly I don’t benefit from Bias frames. Darks are useful in my case! Even if I use the MGEN II dithering function, which is always on anyway.
This is a 100% crop of the integrated and ABE/stretched images: Thanks to the dark frames the background seems smoother (good visible if you look at the center):
The same noise reduction is visible in Lights-Flats-Bias integration against full integration:
Surprisingly the Lights-Flats-Darks only integration seems to be identical to the full integration including Bias:
No more Bias Frames?
So, this result indicates nothing more than I should stop taking bias frames! Dark frames really seems to have a positive impact to my integrated images. But since bias frames doesn’t need much time to take, I’ll continue with it. If I need them, at least I got them.
Please experiment with your own data to find out if Darks are important in your final integration. The results above belongs to my modified EOS 70D. Another camera will result in probably other results, even if it’s a 70D too!
