WBPP Flat Issues

Comments

• BlockHead

  September 2021

  DLSRs are tough. They do some weird things compared to astroCCD/CMOS cameras. 

  It seems like the calibration of the flats is causing an issue- the problem is with the flats or the bias. 

  
  

  The part of the equation you didn't include was your bias. So... make the master bias and master dark available to check out. 

  
  

  My silly guess (and probably wrong) is that the bias isn't really a bias.

  
  

  -the Blockhead
• Tim Dowling

  September 2021

  Master bias and dark added to the dropbox link. On the chance the bias I was using was the culprit, I created a new one, and tried again, but got the same result. The bias at the link is a stack of 150+ frames/ISO 3200/@1/8000". I through in an uncalibrated light as well.

  
  

  FWIW, I haven't changed my workflow. The only thing that has changed is WBPP 2.2. That being said, I'm still fairly confident I am the main variable here. 

  
  

  Tim
• Tim Dowling

  September 2021

  I've done a little more investigation, and here's what I've found out:

  
  

  1. light-dark in PixleMath looks fine

  2. ((light-dark)/flat)*med(flat) looks bad where flat = a calibrated flat, and dark = calibrated dark

  3. ((light-dark)/flat)*med(flat) looks bad where flat = a single, raw, uncalibrated flat looks normal, and dark = calibrated dark

  4. ((light-dark)/flat)*med(flat) looks good where all frames are uncalibrated - i.e. no bias involved
  

  5. And finally, (light-(dark-bias))/(flat-bias))*med(flat-bias) looks bad where all frames are uncalibrated

  
  

  Since the only thing influencing the calibrated flat is the bias, the issue must be with the bias. I used the defaults in WBPP when setting up the bias tab. I also created a master bias using ImageIntegration. Still no luck. So that leaves me with some pretty convincing evidence that my biases are crap.

  
  

  If my bias isn't really a bias, then what is it? I haven't had issues with this before. Perhaps I changed a setting on my camera without realizing it. Going to have to do some digging there. Is there anything you can tell by looking at the bias frame that would give a clue to what's going on?

  
  

  Thanks,

  Tim
• BlockHead

  September 2021

  That was the direction I was leading without actually investing...something about the bias.

  Typically the issue (with DSLR users) is that they take biases and darks in lighted conditions (daytime)...but be done in complete darkness.

  
  

  There are also some sensors for which do you do not use a bias to calibrate the flats at all. This is an exception with some DSLR sensors. You simply use the flats as is.

  
  

  This information is spelled out in Bernd's guide to Calibration...the best information available that covers a wide range of sensors/configurations.

  
  

  It is so good I have it available and I just updated it yesterday (he had some updates).

  
  

  https://www.adamblockstudios.com/categories/Fundamentals
  

  (last link on the page)

  
  

  
  

  -the Blockhead
• Tim Dowling

  October 2021

  Do despite more or less definitively ID-ing the bias as the culprit, I'm still puzzled. Something else has to be going on here. I have successfully calibrated images in the past from the Ra using a bias. And I do take my biases in the dark most of the time. If I don't, I wrap the camera in foil to prevent light leak. To confirm re-confirm the bias issue, I re-ran some old data:

  
  

  Run 1 - Old: lights, flats, darks, bias = no issue

  Run 2 - Old: lights, flats, darks, New: bias = no issue

  Run 3 - Old: lights, darks, New: flats, bias = fail

  Run 4 - Old: lights, darks, bias, New: flats = fail

  
  

  So this seems to indicate that the flat is the actual problem. However, I'm not sure how to troubleshoot this beyond taking new flats... which I've done. I took them manually - histogram right in the middle. Still failed! And really, all the math in PM I ran doesn't support the flats being the issue. So confusing. Any suggestions? 
• BlockHead

  October 2021

  Ok... so, you will need to organize your files.

  You need to make 11 images of each file type available to download. 

  I warn you...If I figure it out... I might make a video tutorial on the issue! (with your permission(

  
  

  But before this- You should try not calibrating the flats. So you just have raw flats loaded, darks that might the lights and lights. 

  
  

  If you read Bernd's document- there are some cameras that do some electronic stuff that messes up what would be a "bias".

  
  

  -the Blockhead
• Tim Dowling

  October 2021

  Haha! Sounds like a fair trade! I'll be happy to accept 10% of the royalties. :) 

  
  

  I did a run without calibrating the flats with the new data. It worked just fine, but there was a pretty high amount of noise in the final integration compared to what I normally see. I'm guessing that was from the bias that wasn't subtracted from the flats. 

  
  

  I did read it - actually read it before you suggested. Interesting stuff. But again, what's not adding up is that my biases used to work, and now they don't. 

  
  

  New dropbox link here:

  
  

  https://www.dropbox.com/sh/67u14gpyxb9l7oj/AAA71p5I2tOTKhoT_yc0lYnqa?dl=0
  

  
  

  The only thing I can't provide is the raw bias files for the Rosette data. I only have the master I created back then. 

  
  

  Thanks!

  Tim