New Module: SuperBias

[Carlos Milovic]

This process takes a master bias frames, and makes operation between pixels to emulate a master bias of several hundred of frames. Differences between the super bias and the master bias of a huge set is well below 1ADU acording our tests.

The idea behind this was born by some experiments conducted by Vicent Peris. He wandered what could be the result of a insanely huge amount of individual bias frames integrated. He used some 2000 images and found that for his ccd camera, vertical variations inter pixel were only noticeable at large scales. Locally, only column patters were visible. I repeated the experiment with some 400 bias with a Canon 300D (CMOS sensor) and the same was found. So, we deviced an algorithm that takes a master bias of a smaller set, and tried to reproduce the behaviour of those huge data sets.

I'll give an example of this process tomorrow.

Downloads:
Windows 64bits: http://pteam.pixinsight.com/pcldev/carlos/x64/SuperBias-pxm.dll
Linux 64bits: http://pteam.pixinsight.com/pcldev/carlos/x64/SuperBias-pxm.so

[Nocturnal]

That sounds interesting but I'm a little concerned about messing with master bias files. If I don't like the quality of my master bias I can take more of them, right? It seems that fixed patterns in bias frames depends heavily on both the sensor and the circuitry used.

[Carlos Milovic]

Yeah, you can always take more subs. The deal here, is that with 20 or 40, you'll end with a result that will look (and have an maximum error below 1 ADU) almost the same as a 400 or 2000 subs master bias. It is a lot that you are saving there (disk space and time).
The real issue is the question if such improvement really makes any difference to the final images... In any case, since the SuperBias procedure is almost instant, I'm always using it. My latest 10 images used them.

[Geoff]

 Hi Carlos
Not sure how to proceed after downloading the file.  Presumably I have to move it to the appropriate folder.  Help appreciated.
Geoff

[Alejandro Tombolini]

Hi Geoff, save the file in bin folder and from PI install it

Process>Modules>Install Modules>Search

But!, I think that this module do not works in 1.8RC3 yet. 

[Geoff]

Thanks Alejandro
That works for me in 1.7.  I haven't installed 1.8 yet.
Geoff

[georg.viehoever]

Are there any physical considerations that are behind the process? It is difficult for me to understand that it is possible to generate a high quality masterBias from just a few biases? Sound like mathematical voodoo ...
Georg

[cs_pixinsight]

Just curious.   What determines when a module and/or script is released with PI versus having to download it manually?

Craig

[Alejandro Tombolini]

Just curious.   What determines when a module and/or script is released with PI versus having to download it manually?

mmmh, it is not easy. I have a file with last update of most of the modules and scripts and if one is not present in the oficial distribution list, then I install it. But it depends on the vertion. For example I have right now 52 modules in win7 and 42 in Linux. I am missing Carlos' modules and other than are not available for 1.8 yet.   

[Carlos Milovic]

Hi Georg

The idea behind this is that for certain cameras, vertical patterns are dominant over differences between pixels in the same column, so we may assume that the only high frequency components that are relevant are those in the x axis. So, we just have to average each column, to achieve the mean value. Since we have hundreds of pixels in the same column, it is almost the same as averaging hundreds of frames, in a individual pixel basis. Medium and large scale components are preserved with ATW.

[Carlos Milovic]

Hi Craig

Basically, that depends on Juan. If he has the time to look at a code, and it is considered as a valuable contribution, it is included in the standard repository. In the case of my processes, he just have put them on the qeue list, untill he have more time to look at them. We have plans to include most of them in the standard package during the 1.8 cycle.

[jdupton]

Is this module still available? 

I played with it on PixInsight v1.7 but found that now that I have upgraded to the latest v1.8 it will not install.  I went back the original directory (shown in the first post above) to see if a new version was available.  The file link now gives a "404 Not Found" error and the whole directory gives a "403 Forbidden" error.

I'd like to play with this some more.  It seemed to have a lot going for it.  Is there now a version of SuperBias for v1.8 and if so where can it be found?

John

[Andres.Pozo]

Hi Craig

Basically, that depends on Juan. If he has the time to look at a code, and it is considered as a valuable contribution, it is included in the standard repository. In the case of my processes, he just have put them on the qeue list, untill he have more time to look at them. We have plans to include most of them in the standard package during the 1.8 cycle.

This is why I write scripts and not modules. Modules can be much faster but the maintenance is too involved: you have to recompile the module for each platform for each version of PI.

[cbradshaw]

I would also like to try this module.  With my DSLR, I want to minimize shutter actuation.

Regards,

Charlie