Crazy idea or not HOO approaches Soul neb

[Mike in Rancho]

Seemed more than just noise, Freddy, it had a pattern. Unless it's just me?

Were you random dithering?

Maybe something with the pre-processing? I'm not sure how mono DSLR files come out, I assume something like CR2? DSS may insist that it be debayered still? I'd have to look at the options.

[fmeireso]

Mike,

Mono DSLR is Canon so yes CR2 files.
But you have to run DCraw on it , otherwise DSS does see it as colored files. DCraw is a program that converts to tiff and other stuff so DSS then sees it as mono files.
PI does exactly thesame! But the program is integrated. The Elf and me sorted this out....

I did test shots wih the mono in daylight, i could not see anything wrong, pictures were clear and very sharp...they came out pretty good.

and yes i have dithered. I don't really see walking noise in the stack so i guess it has done his job...
Could you explain a bit more what you mean about pattern?
Myself i see more a bit of rough outlook of the Ha stack, The Elf are way more i call it refined

Yesterday i denoised it twice in GIMP, it did get a much 'softer' outlook without really loosing detail. Could also be that a few subs were of less good quality, i stacked them all though...the subs are subs of 10 minuts.

[admin]

Mike,

Mono DSLR is Canon so yes CR2 files.
But you have to run DCraw on it , otherwise DSS does see it as colored files. DCraw is a program that converts to tiff and other stuff so DSS then sees it as mono files.
PI does exactly thesame! But the program is integrated. The Elf and me sorted this out....

I did test shots wih the mono in daylight, i could not see anything wrong, pictures were clear and very sharp...they came out pretty good.

and yes i have dithered. I don't really see walking noise in the stack so i guess it has done his job...
Could you explain a bit more what you mean about pattern?
Myself i see more a bit of rough outlook of the Ha stack, The Elf are way more i call it refined

Yesterday i denoised it twice in GIMP, it did get a much 'softer' outlook without really loosing detail. Could also be that a few subs were of less good quality, i stacked them all though...the subs are subs of 10 minuts.

Thanks for sharing those stacks Freddy.

I think your #1 priority should be to figure out why both stacks are exhibiting rather severe pattern noise. As Mike noted, the Ha stack particularly, has some sort of odd hatch pattern going on. Try dithering a lot more aggressively. Are you using a spiral pattern?

There is indeed no distinct walking noise visible, but I can spot some diagonal dark streaking/smudging (and even a satellite trail?);

trail.jpg (87.14 KiB) Viewed 4974 times

At the same time noise is not single-pixel and bleeds/clumps into multiple pixels;

Selection_725.jpg (317.42 KiB) Viewed 4974 times

All this "detail" that is smaller than a single star should not exist. E.g. all you should ideally be seeing, is random, single pixel noise. Not pixels that are noise that look almost the same as their neighboringing noisy pixel.

Parameter [Luminance, Color] set to [L + Synthetic L From RGB, RGB] (still unsure about this setting)

This should indeed just be L, RGB. You want to use the detail of the Ha, but only the color of the OSC stack. Not the Detail of the Ha and OSC and the color of the OSC stack.

It's definitely not the easiest data to process with the severe pattern noise, and I think you've done well with the dataset as-is, and staying true to the data.

Be careful when applying noise reduction routines that perform edge enhancement and/or structure modification/"repair". They may work reasonably well for terrestrial scenes with well defined edges, but in DSO "edges" are never truly edges and almost always have micro-detail

The result of using them on DSOs is usually an "oil painting" look, like one of the PI/PS image in the CN thread, where stars - always the sharpest objects in the scene because they are infinite point lights - seem like smudges, while the edges are sharper than the stars. That's obviously not how optics work...

Hope this helps!

[Mike in Rancho]

Hi Freddy,

Well Ivo already covered it (and better than I could), but I also excerpted from your Ha data last night (but no cool arrows), and also one of my recent sets for comparison - though probably will take two posts.

Spiral vs random dithering is indeed something to look into. I used spiral, thinking I was keeping the image from shifting away from me, until I realized it was imprinting a hatch pattern into my images that just would not process out. In fact, on my West Veil, with the first 3 hours being spiral dithered, it took an extra 6 hours of random dithering to mostly wash that pattern out.

I'm not sure yours has the obvious criss-cross pattern that my spiral dithering did, but you do have what seems to be like shoe tread-print marks throughout the image. Not all uniform...but, you can see the same patterning in very different areas of the image (structural, empty, etc.), which of course nature would not do. And similarly to my spiral dithered stack, yours is also pretty much impossible to process out.

So something is amiss, especially as you say Elf gets very clean subs out of his mono modded camera -- which is the reason you did it of course. You should expect the same. If it's not in the acquisition, then maybe something in the CR2 to TIFF conversions? Have you compared a CR2 sub to its corresponding TIFF? Just open them as layers in Gimp and toggle between the two? Or perhaps DSS settings but at the moment I can't think of what might cause that.

Anyway here's your Ha stack, just opened as Open (I think Open is ok for grayscale mono? ST seems to know there's no color). Screenshot so not so great and then jpg compressed, but at 66% I think the ST scale is. You have to get the scale right and then lean back from the screen a little so as not to be mesmerized by all the stuff...and you should start seeing these patterns throughout.

Freddy Ha autodev scaled.jpg (380.89 KiB) Viewed 4956 times

[Mike in Rancho]

I know this isn't a perfect comparison, since it isn't mono. But, from my recent Pacman data, also DSLR, and taken at 720mm, albeit demosaicked. I composited the R channel only, so this is my "Ha."

Now, I wouldn't call it perfect of course, but for the most part the noisy stuff is kind of uniformly distributed, and any little areas that may show some patterning are not repeated all over the place, like your Ha shows.

My Ha extract autodev scaled.jpg (263.4 KiB) Viewed 4956 times

We'll get to the bottom of it Freddy! You should be getting the excellent results you expect from that mono mod.

[fmeireso]

Hi Ivo and Mike,

I really , really ,appreciate this feed back, cause if nobody takes a look with the knowledge to evaluate i will not know it. I am not yet experienced enough to analyse or evalate images unless it is something obvious.

@Ivo and Mike , you think that dithering might be the big issue ?

On the 76 mm i use APT setting on 6 and in PHD2 scale setting 3.
That is 18 x 0,95

0,95 is the result of my guide scale/image scale

So 18x 0,95 is roughly 18 pixels, if this calc is right.
What do you suggest? How many pixel should i try. It's true i heard stories of people with DSLR going to 30 pixels. God knows how that will affect the settle time...

Dithering is set to random, i never used spiral...
Mono stack dithered every sub of 10 minutes , OSC every 2 frames, of 2 minutes

but for nowi can't think of anything else? If it is dithering it will be easy to solve. It might be just a case of more agressive dithering to apply...

Also true i was not topicky on my subs. I only had 20 of them Ha subs, I just stacked them all, and yes forgot the one witht he satelite trail...

Still all things considered, i am happy witht the result. I know it can work and the Ha just boots up the nebulae. Off course much room for improvement but as this was my second attempt, personaly i
found it not too bad,cause the difference with or without Ha is pretty big..

Glad to hear youre comments,

[fmeireso]

I think your #1 priority should be to figure out why both stacks are exhibiting rather severe pattern noise. As Mike noted, the Ha stack particularly, has some sort of odd hatch pattern going on. Try dithering a lot more aggressively. Are you using a spiral pattern?

There is indeed no distinct walking noise visible, but I can spot some diagonal dark streaking/smudging (and even a satellite trail?);

trail.jpg

At the same time noise is not single-pixel and bleeds/clumps into multiple pixels;

Selection_725.jpg

All this "detail" that is smaller than a single star should not exist. E.g. all you should ideally be seeing, is random, single pixel noise. Not pixels that are noise that look almost the same as their neighboringing noisy pixel.

i think i see this in former stacks of mine too even in another scope. I see pixels that colorwise (greyscale) are thesame as their neigbourhood pixel, 3 to 4 or even 5.
So every pixels should look different unless there stars or real data?

[Mike in Rancho]

I do pretty much the same, Freddy. For the L-eNhance I dither every frame, which is usually 6 minutes, and full RGB I tend to do every 3 frames.

I am going to have to look up APT, as I do not know how it commands the dithering. I had assumed all third-party connections to PHD2 used a similar control method. In BYN, the dithering "aggressiveness" settings run from 1 to 5. Those settings mean certain pixel amounts (and it is not obvious what those are). In turn, PHD2 has a scalar that can then be applied to the amount of pixels that the 1 to 5 choice represents. But you say your option is 6. Well, that's not 1 to 5 lol. Hmmm.

That, of course, is the amount that the guidescope will dither, and has to be converted to the imaging scale. If we know your cameras and focal lengths for both, we can look those up and calculate the ratio in arc-seconds.

Good that you chose random, though that makes it more difficult to figure out. And I presume PHD2 reported decent guiding in the graph? The only other thing I've got is the subs and that conversion...

[admin]

Really appreciate @Mike in Rancho helping use out here. I'm too much of an "armchair" astrophotographer with too little real world experience with specific gear. I can only "work backwards" from what I see in a dataset.

The thing is, the satellite trail should have been stacked out by the stacking/rejection algorithm. 20 frames is *just* about enough to start using Kappa/Sigma rejection, but you could also use Median stacking.

I wouldn't think settling time would change too much whether dithering a little or a lot. I would think the amount of introduced vibrations should be roughly the same, while the slewing shouldn't take long either. But again, I'm too much of an armchair astrophotographer to be sure.

[Mike in Rancho]

I just like investigating and figuring things out.

Now, I just need to do the same when it comes to luminance and color spaces. Ugh.

Freddy, I think you noted that this was taken with a 76EDPH and .8 reducer. Some more googling indicates your mono mod is a 600D/T3i. So astronomy tools says that is 2.66"/pixel. Likewise, more google (all of CN stuff, of course) leads me to believe you are guiding with an ASI290 and 60/280 scope. That would be 2.67"/pixel. So, basically identical movement will be achieved pixel to pixel.

From looking up APT's instructions, when dithering via PHD2 it seems to have the same third-party instructions that BYN does. Meaning it sends PHD2 a number from 1 to 5. 1 = 0.5 pixels. 2 = 1 pixel. 3 = 2 pixels. 4 = 3 pixels. and 5 = 5 pixels. You can then scale that in PHD2.

You said your scalar is 3, but...there is no 6! So I'm not sure exactly how much you have been dithering. Can you check your APT settings and let us know what it says? And as a check on things working, have you watched subs come in to your screen and noticed the image shift due to the dithers?