DeBayer Process Issue

James Alley

If I let WBPP do the DeBayering it works fine but if I try to use the DeBayer process on a registered gray image file using any of the settings for pattern (I tried them all), they always are turning out a mix of purple and green splotches.  I went through each one with each making the splotches appear in different areas but always purple and green.  The reason I am trying to run the process by itself is because I was getting a grid pattern in stacked image when it was stretched so I decided to backup and remove the ones WBPP had added and do it manually.  

Thank you

-Jim

BlockHead

Did you do an unlinked AutoSTF?

Can you post a screenshot or two?

Sometimes it is difficult in instances where there isn't much information in the image.

(compared to like say M42)

-the Blockhead

James Alley

I had erased and restarted.  Every time I run the Weighted script I am getting different results with the debayering even though I'm setting it what my camera (Sony A7RIV) pattern is and not using auto.  I tried auto once with the same result.  Most of the time (I ran 4 times today) it is red but this last time it's green.  Each time I erase the output folder and start from scratch.  This time the pattern was at it's worst in the example attached. I had used StarXterminator with the defaults and did a comet align process.  Then I did an integration and this is the result.  My guide camera that night had died and I was only using 10 second subs and only got 162 before the clouds rolled in.

James Alley

James Alley

Sorry about black image.. Hopefully uploading it correctly now.

BlockHead

Hi James,

This is a processed image (with StarRemoval)... so we should not be looking at this to figure out the debayer question. In addition, this image is created by the integration of star aligned images (I assume) from which you used SXT to remove stars. The quilted background is due to SXT on the very nosiest/faintest pixels of the data which you are displaying. This indicates you do not have much signal..but it would be better if the image was integrated on a comet aligned result... not a comet rejected image.You may have tried to align on the coment in CometAlignment..but I you didn't quite succeed.  Although this is a comet rejected image, the colors look basically correct to me here (given the small amount of information).

Can you please ask your question again given my understanding above?

Given the same settings in WBPP you will get the same answer each time. Remember to delete the files, clear WBPP (reset) AND clear the cache. Comet processing is tricky- but your issue is more fundamental and likely the steps of FasTTrack training are where we need to be absolutely solid. You need to look at the calibrated data and make certain everything is OK. You should then be able to look at the debayered data (or take a calibrated image and debayer it yourself like I show in FastTrack Training) to find the correct pattern (likely RGGB). You have not indicated you are not being "tricked" by a linked and unlinked STF. This is also part of the mix I cannot assume is not an issue. 

Thanks,

-the Blockhead

James Alley

Hi Adam,

I actually made it green but this attachment shows what happens, it's the masterLight, this is straight from WBPP with either auto/VNG debayering or RGGB.  I tried it manually as well and couldn't get the colors to look right.  I'm wondering if something in the settings has be changed and it's causing this?  I haven't had this issue before with a couple other projects.

Thank you!

-Jim

BlockHead

Hi James,

Again, I am not certain I understand. If this is straight from WBPP- then this is a fine looking image that has not yet been color calibrated. There is a mostly red bias which is OK. What does an unlinked STF look like for this data? Can you provide the image you are having trouble with by uploading to a server to be downloaded?

-the Blockhead

James Alley

I finally figured it out and the results here.  The problem is with StarXterminator.  It's creating the grid artifacts.  I then used StarNet 2 with much cleaner results.  I found some bugs in PixInsight along the way that was causing me fits lol.