Question about drizzle bayern matrix while using HA-OIII filter

Why do you use scale 2x ?

Normally, we must use scale 1x with bayer-drizzle with droplet between 1 to 2.5 (generally I take 1.5). You will obtain a Ha, OIII, or color HaOIII with native resolution of your sensor without artefacts. Droplet of 1 can favorise artefact and noise. 1.5 to 2.0 is better.

@philippe-bernhard well someone told me for undersample that scale must be 2 and droplet 1or maybe 0.8 for better detail.

I got undersample resolution 3.

Stefan

I can confirm that a scale of 2 and a droplet size of 0.8 is a good starting point for bayern/X trans drizzle with RGB data from a OSC camera, provided you have lots of frames and meet the usual requirements for drizzle (undersampled data and frequent dithers). However, it sounds like you are using a dual narrowband filter to extract Ha and Oiii from an RGB dataset. In that case your starting point (after extract) is narrowband data rather than bayered RGB, so I don't think that the normal method for bayer drizzling RGB data will work well.

Someone can correct me if I'm wrong, but I would think that having extracted NB data from the OSC camera it should be drizzled as if it was from a mono camera.

I would add a note of caution. NB data from a colour camera will always be lower quality than from a mono camera, so I would be cautious about deciding to drizzle at all, unless you have hundreds of frames, maybe not even then. But there is no harm in experimenting and seeing what it looks like.

JC

@stefanastro i would agree with Phillippe here.

Scale 1x and drop size 1.2 is my personal favourite, (my camera’s best balance between sharpness and drizzle injected noise), for full fov shot but with at least 80 frames.I use a 2600mc osc camera. If your doing a tight crop on a small object then use drizzle 2x with drop 1.2 and use the integration inbuilt crop mode, not a crop after integration).

Bayer drizzle has a DIFFERENT OBJECTIVE to standard nasa drizzle as its really just a debayer mechanism to restore full resolution from an osc camera primarily, rather than a restoration of resolution from undersampled data. You can use bayer drizzle on well sampled and over sampled data just as effectively. ( I do!). Its designed to combat the loss of resolution from using a bayer matrix where usually red and blue channels are only 1 out of every 4 pixels and green is 2 out of every four pixels. Because of this, drop size gives  a different end result than the same drop size when used in standard drizzle and a mono camera. (Where every pixel contributes). You are ‘filling in’ totally missing whole pixels in each colour rather than ‘smoothing the edges’ of undersampled data as with standard drizzle. Too small a drop size in bayer drizzle produces abberations where the same drop size in std drizzle does not. Smaller drop sizes in bayer drizzle need vastly lots more frames to counteract abberations than std drizzle. (More examples to fill in the whole pixel missing data). Too few frames overall in bayer drizzle also produces aberrations, usually a ‘basket weave’ effect on the background. I find at least 80 frames and much better 120+ frames gives the best result @ drop 1.2-1.5. AND the frames need to be dithered well, ideally every frame, and at least 4 pixels dither to combat the bayer matrix, the difference in position of each pixel in each dithered frame is CRUCIAL in calculating the true colour channel each pixel finally represents in the integration. Bayer drizzle is about tracking the pixel colour of each frame to restore missing resolution (std drizzle is tracking luminosity of the pixel). Good dither ensures this. I dither every frame, and 10 pixels in a spiral pattern, minimum 80 frames. This ensures proper true colour is extracted for each final pixel in the imtegration.

Set the tab 0 algorithm to ha mono and then bayer drizzle integration, 1x and drop 1.2.

then a second integration with tab 0 on oiii mono and bayer drizzle.

this will give you a ha and oiii bayer drizzle full resolution MONO frames to combine in rgb combine to produce a colour pic.

if you want to do just one integration and get duo colour immediately, just  use the haoiii colour algorithm in tab 0 and bayer drizzle.

APP is intelligent enough todo all correct extractions in the background for a duo filter.

Dont forget, APP does not know about any chemicals , so when you say ha, app is thinking extract red channel only, and when you say oii, its thinking extract both  green and blue channels only, and merge 50/50. No software knows chemical, only the channel to extract. The filter needs to separate the chemicals cleanly into channels beforehand. A duo filter passes only ha into the red channel, and splits oiii into both green and blue channels.  A quad filter cannot be seperated correctly chemically, as both ha and sii are passed into the red channel and oiii and hbeta into green and blue. So for a true SHO rendition with an OSC camera, you need a seperate sii only filter, and extract the red channel (in APP use same ha extract, but rename file sii) AND a duo filter, and extract ha (true ha) and oiii green and blue channels.

sorry I went off topic a bit, but may be useful info for someone lol!

I always drizzle 1.5X with droplet size of 0.75

/stefan

@andybooth so I can better do mono HA and drizzle, not HA-OIII extract HA and drizzle like you told.

Thanks I am going to try.

I did drizzle with 60x300 L-quad enhance a broad and filter and therefore I did scale 2 and drop 0.9 and no problems. Even after the intergrate I did in tab 2 split channels. No problems with the pixels

@stefanastro i am away on holiday with no pc, so I cannot remember exactly the exact titles of the options in tab 0,

but you are selecting the extraction of ha and  separately oiii, from a duo filter, with a osc camera and creating mono files.

if you hover the mouse cursor over the options it should be clear I think.  Do not pick options designed for use with a mono camera.

if you are unclear, then please post a pic of the options and I can advise.

andy

@digitaliz-se interesting, i always get artifacts on such low drop size on bayer drizzle.

what is your pixel image scale, for information and understanding? Thx

Andy

Thanks for all that information - it was very interesting. I would never have considered using bayer drizzle on oversampled data, but I get your idea that its really doing two different things at the same time. Maybe I have been too conservative in experimenting with this tool.

JC