Why CFA Drizzle Integration to be best for OSC / DSLR data ?

[alexxw24]

Hi everyone,

e.g. Juan mentioning CFA DrizzleIntegration as the recommended way for integrating DSLR/OSC data some time ago (see underneath).
I'm usually working with a mono-CCD but have some DSLR widefields from earlier this year to process ahead of me.

I want to understand the background, why CFA Drizzle is supposed to be better / recommended, well dithered and larger number of sub frames assumed.

The positive effect on undersamples subs I am aware off but what makes it with the bayer matrix, better color interpolation?

Thanks for your thoughts,
Alex

I am under the impression that AHD is superior to VNG.

Why? Yes, we can implement more CFA interpolation algorithms in the Debayer tool. However, this is not a priority task for the following reason: In current versions of PixInsight, CFA interpolation with the Debayer tool is intended exclusively for generation of temporary working images, required as input for StarAlignment. The only recommended way to generate integrated color images from CFA raw data, be it data acquired with digital cameras or single-shot CCD cameras, is DrizzleIntegration (i.e., CFA drizzle, AKA Bayer drizzle). The only exception may be if you have just a marginal number of raw frames, where drizzle may not be applicable.
....

 

[acmalko]

Because with BayerDrizzle, there is no interpolation.
BayerDrizzle will use subs before debayer. We debayer them only for alignment.

I usually use it with scale=1 and Drop Shrink=1

 

[alexxw24]

Okay, thanks.

Yes, had best results with scale 1 and drop shrink 1 as well.

 

[mjunius]

Hi all,
After mostly mono imaging for quite a while, I'm currently processing some new data from the ASI 294 MC Pro. And thought that it might be a good thing to adapt my workflow to Bayer drizzle. Alas, this yields far inferior results, checkerboard like patterns show up around bright stars, see attachment (only Autocolor script + Arcsinh after integration). Any idea what I'm doing wrong? I used scale 1 here and the defaults (right), versus standard integration (left). But also Drop shrink 1.0 and Gaussian didn't yield any better result, same at scale 2.
My workflow: ImageCalibration -> CosmeticCorrection -> Debayer -> SubframeSelector -> StarAlignment -> ImageIntegration -> DrizzleIntegration with all the drizzle and CFA options (hopefully) properly set.
Thanks.
Martin

 

[John_Gill]

for the pink stars try this script "Repaired HSV Separation" enable "V - no repair" then use ChannelCombination ---> HSV ---> H, Sv, Unrepaired _V images and apply.

 

[mjunius]

Thx, in fact I'm using this script (a personally modified version of it) already later in the workflow, but the pink cores aren't the problem. I'm a bit at a loss regarding the checkerboard pattern with DrizzleIntegration.

 

[rbarbout]

Thx, in fact I'm using this script (a personally modified version of it) already later in the workflow, but the pink cores aren't the problem. I'm a bit at a loss regarding the checkerboard pattern with DrizzleIntegration.

How many subs do you have ? Do you dither ?
Bayer Drizzle needs (among other things) a lot of subs and dithering.

 

[mjunius]

How many subs do you have ? Do you dither ?
Bayer Drizzle needs (among other things) a lot of subs and dithering.

Yes, I do dither, and typically have ~40 subs in the recent cases. Meanwhile I switched to using WBPP with separate processing (alignment, integration) of the R, G, B channels which doesn't create these artifacts and results in better image overall quality, in particular stars and their colors.

 

[rbarbout]

Yes, I do dither, and typically have ~40 subs in the recent cases. Meanwhile I switched to using WBPP with separate processing (alignment, integration) of the R, G, B channels which doesn't create these artifacts and results in better image overall quality, in particular stars and their colors.

So all is OK then !
You can still look in the WBPP logs directory and open the process containers to see what is the difference with you manual worflow.

 

[mjunius]

Yes, it is. ;-) But of course the separate R/G/B alignment/integration is a different approach than Bayer Drizzle, and these can't be combined in WBPP. Thus I'm still at a loss where the artifacts originally come from.

 

[jkovach]

for the pink stars try this script "Repaired HSV Separation" enable "V - no repair" then use ChannelCombination ---> HSV ---> H, Sv, Unrepaired _V images and apply.

When should this step be performed?

 

[John_Gill]

Sometime in the RGB image the star cores turn a bit pink. This process fixes the issue and returns better star colors. I normally doi this shortly after ChannelCombination or LRGBcombination.