Using the defect map

[Geoff]

I have no idea how to use this feature. My naive assumption is that one of its uses would be to build a map of hot/cold pixels, which can saved and used in image calibration.  Can someone clarify the use of this tool?
Geoff

[Carlos Milovic]

Hi Geoff

Juan already have the documentation of this process, and should release it soon (I hope so...).

In the meanwhile: This process uses a hot/cold/bad pixel map that you created by other means (for example, simple binarization of a master dark frame). The idea is that it uses only information from neighbouring pixels to replace the bad pixel. You select the neighbourhood shape, but the size is automatically calculated, depending on the number of bad pixels surrounding the current pixel to fix.

[Harry page]

Hi

I do show how to use it here http://www.harrysastroshed.com/Calibration.html 

Harry

[Geoff]

This process uses a hot/cold/bad pixel map that you created by other means (for example, simple binarization of a master dark frame). The idea is that it uses only information from neighbouring pixels to replace the bad pixel. You select the neighbourhood shape, but the size is automatically calculated, depending on the number of bad pixels surrounding the current pixel to fix.

Thanks Carlos.  I have now tried this but the process seems incredibly slow.  I must be doing something wrong.  I binarized a dark frame, then loaded the binarized frame into the DefectMap panel and dragged the new instance triangle to a light frame. I thought it had hung up, showing 0% in the Process Console for a long time, but then it moved on to 1% and after I'd had a shower it had got to 3%, so I aborted.

Hi

I do show how to use it here http://www.harrysastroshed.com/Calibration.html 

Harry

Thanks Harry.  Yes I've used the CosmeticCorrection tool following your video and it works very well.  I was exploring the DefectMap (Process > ImageCalibration > DefectMap) as an additional arrow to my bow.
Geoff

[Geoff]

I have now tried this but the process seems incredibly slow.  I must be doing something wrong.  I binarized a dark frame, then loaded the binarized frame into the DefectMap panel and dragged the new instance triangle to a light frame. I thought it had hung up, showing 0% in the Process Console for a long time, but then it moved on to 1% and after I'd had a shower it had got to 3%, so I aborted.

OK I've figured it out.  I was doing something wrong. (The error often lies between the back of the chair and the keyboard.)  After binarizing the dark frame, you have to invert it and use the inverted image as the defect map.

[Carlos Milovic]

Sorry, this happens quite often with this tool... I've been wondering if I should change the policy of its internal workings to make it a bit more intuitive  (anyway, it was explained in the documentation (Juan....)).

[LukeB]

Guys,
I just want to thank you all for these VERY helpful processing tips.

I have applied them to a couple of the images that were causing me much dismay and thanks to you guys I feel that I have made some progress at last.

Harry, your tutorials are just fantastic and I would have been far behind without them.

Cheers,
Luke

[mads0100]

Just wanted to spell this out for future readers of this post since there is no documentation for the DefectMap:

1. Create a master dark
2. Use the Binarize tool.  You'll need to play around with this using the slider (and lower than the slider will go) to get the value low enough to highlight your hot/cold pixels.
3. Invert the image.
4. Save as your DefectMap.
5. Open your un-calibrated images
6. Use the DefectMap tool. This will allow you to select your DefectMap view.  I don't know what the various options mean nor is there any tool tips for the various options.  Hopefully someone else can fill this in for us. 

Now, apply the 'triangle' to each image.  You'll notice the noise is significantly reduced similar to calibrating with darks.  In addition, it will correct defects such as hot/cold pixels.

I hope this will help someone else.  It really saved the data I was gathering as just using darks wasn't enough. 

[cfranks]

I just tried the above sequence and get an 'invalid memory read operation' error.  It is such a simple process that Mads0100 kindly listed that I can't see what I might have done wrong.
Ripley RC3, Win 7  64 bit, 16GB RAM

Charles

[Geoff]

I just tried the above sequence and get an 'invalid memory read operation' error.  It is such a simple process that Mads0100 kindly listed that I can't see what I might have done wrong.
Ripley RC3, Win 7  64 bit, 16GB RAM

Charles

I've also been getting that lately, but it's inconsistent. I've been waiting till I load  v1.8 before raising bug reports

[Carlos Milovic]

Does this error happen with the average operator? There was a bug (that i fixed, but i'm not sure that Juan updated the repository with the new code).

[derrickf]

Does this error happen with the average operator? There was a bug (that i fixed, but i'm not sure that Juan updated the repository with the new code).

In my case Mean and Gaussian operators work without errors, the other three operators (Median, Minimum and Maximum) throw the error reported above.

HTH

Derrick

[Carlos Milovic]

Ok, so it could be the same bug I already fixed. I think we have to wait until RC4 is launched to put pressure on Juan to update this.

[Geoff]

Does this error happen with the average operator? There was a bug (that i fixed, but i'm not sure that Juan updated the repository with the new code).

In my case Mean and Gaussian operators work without errors, the other three operators (Median, Minimum and Maximum) throw the error reported above.

HTH

Derrick

+1

[cfranks]

Thanks for the info, I will wait for RC4.  I tried 'Mean' for the Operation and had no failure so I just played around a bit with it.

Charles