WBPP and rejected images question

Roger Smith

Hi Adam,

I ran WBPP successfully, with my RGB subs with a reasonable master color image result, I have some questions.

1. If I don't select any WBPP preset, ie Fastest, Faster or Maximum quality, then is WBPP going to select one for me?

2. I selected Maximum, and I had Drizzle Integration unchecked. WBPP created drizzle files. So what is the point of unchecking it?

3. I was monitoring the Process Console when WBPP was processing and I noticed it said some images were rejected. So I next watched your WBPP Pipeline Processing Container Video...

In the logs directory I opened the long .log file which has every step done for every frame throughout the pipeline. Here I found the End of Demosaicing followed by "Perform Image Measurements" and then Subframe Selector: Global Context shows up. After that was the list of Rejected Images where any image below .400 was rejected. What is ths paramter callled?

4. Then I used the Script Editor to open the Process Container for the pipeline. I double clicked Subframe selector. It was empty. So I executed it and found every frame in there. To me this does not make sense because I processed R, G, B separately and the RGB frames. My feeling is that each color frame will be much weaker, ie different than a RGB frame. Please give me your opinion on this.

Second, I could find no place inside Subframe selector to set any rejection limit. So that magic .400 was not showing. FYI, I have been a NSG guy, and not used subframe selector since early days of NSG. I really don't like it very much, and not so long ago you were pointing out it's limitations.

5. Finally in Image Integration (double clicked to open it from the Script Editor listing). I found the .400 minimum weight there. I used PSF Signal Weight. But back in WBPP setup there is no place to set the limit inside Image Integration. For future I will rdo Image Integration so have more granular control.  My question is..How can WBPP list the rejected frames before Image Integration is started?

I appreciate your comments/answers. If you have any questions, please let me know. 

Thanks,

     Roger

  

BlockHead

Hi Roger,

1. Maximum quality is the default. Really though, these are just presets for the number of things you turn off in WBPP to speed things up. These presets came before Fastintegration. If you are concerned about anything other than Maximum Quality.. you should just use FastIntegration. 

2. By unchecking Drizzle integration...it literally will not integrate the data using this method. The Drizzle files are small files (not data files) that contain registration, normalization, and rejection information for Drizzle integration. They are always created. PI developers consider the memory usage so tiny... they are just created whether they are used or not.

3. It appears you changed the minimum weight parameter? The default is 0.01 or something. You perhaps did this based on my recommendation? Of course only do this if you understand how and why. The minimum weighting parameter is in the ImageIntegration process- though it comes into play early on. This was always a tricky parameter...

4. I am not certain I understand the first part of the question here. 

5. I disagree and I think you are mistaken. In WBPP if you to under ImageIntegration you can adjust the minimum weight setting. As far as knowing the frames before ImageInteration... that is tough. You see PSF SIgnal Weight does calculate the weights early on. This is NOT true for PSF Scale SNR. So what do you do as a developer? Show in some and not in other cases? It gets complicated.

-the Blockhead

Roger Smith

Hi Adam,

Thank you for your replies/advice.

1. I understand this better thanks.

2. Yes drizzle files are small and a small addition to overhead. But if it won't integrate w/o drizzle, then seems the choice to not use it should be removed. I want to go back to your video again where you discussed checking Enable Drizzle for undersampled data.

3. and 5. I agree you.  I just went back to WBPP under the Lights Tab/ Image Integration, and did find the Minimum Weight setting input box. So I was wrong when I said it was not there. Sadly I do not recall if I typed in .400.

4. Let me rephrase my question. Do you think it proper to have the individual channel frames (R, G, & B) evaluated together with the RGB frames in subframe selector? They will all be evaluated against each other. 

Maybe it is okay, becuase I had 24 rejected RGB frames, and 29, 18, & 24 of the individual R, and G and B frames respectively. That is an average of 23 frames each individual color. 

I have a great respect and appreciation for Pixinsight and the developers. And the same for you. So thank you for sharing your advice and expertise.

Roger

BlockHead

Hi Roger,

4. I guess the confusing part of the question for me is the fact I do not think there is a difference. When you give SFS a color image...I think it is analyzing the three channels individually. So the results should be the same as giving SFS individual channels (roughly). But I am speculating a little here..I have not looked at this in a while.

-the Blockhead