NGC7000 nice or over processed noise?

[TiffsAndAstro]

im a bit surprised with this result as i really quite like it. but is it as nice as i think it is, or have i just over processed noise? 43 x 120sec subs, crap gear

[carastro]

If it was noise I think you'd see that in the dark areas.  I am not so familiar with RGB images of this area, what camera are you using,  

so what is this 

Quote

crap gear

Lol 

[TiffsAndAstro]

stock canon 600d and my "non premium " crap gear is in my signature. I think this is the first 'proper' image ive managed and less than two hours

it does have a bit of an impressionist painting look to it, which i rather like, but im not sure if that's 'good' or not

i think i can improve it a bit more, i haven't bothered putting it through basic stuff in gimp yet

its ngc7000 and some of the pelican nebula. couldn't get them both in same frame without rotating my ota in its rings which im trying to avoid.

Edited June 3 by TiffsAndAstro

[Clarkey]

I would say for less than 2 hours with a stock 600D, this is pretty good. The star shapes are good and although there is some noise it is not excessive. Longer integration time will allow you to pull out more detail without the noise taking over.

I started with a 600D and it is not a bad camera. The AZ Gti has its limits, but for a 72ed it is fine.

Personally, I would be happy with this result.

[Albir phil]

6 minutes ago, TiffsAndAstro said:

stock canon 600d and my "non premium " crap gear is in my signature. I think this is the first 'proper' image ive managed and less than two hours

it does have a bit of an impressionist painting look to it, which i rather like, but im not sure if that's 'good' or not

i think i can improve it a bit more, i haven't bothered putting it through basic stuff in gimp yet

its ngc7000 and some of the pelican nebula. couldn't get them both in same frame without rotating my ota in its rights which im trying to avoid.

Like I always say if you are happy with your image that's all that matters, after all other people will have different opinions on any image, take in what they say but if you like it that's ok🤗🤗

[TiffsAndAstro]

4 minutes ago, Clarkey said:

I would say for less than 2 hours with a stock 600D, this is pretty good. The star shapes are good and although there is some noise it is not excessive. Longer integration time will allow you to pull out more detail without the noise taking over.

I started with a 600D and it is not a bad camera. The AZ Gti has its limits, but for a 72ed it is fine.

Personally, I would be happy with this result.

very kind of you to say. the mount isn't the az version its the  eq mount version and i agree with you, im really happy with this, bordering on amazed with myself.  hence why im double checking i haven't just over stretch it. tbh i think it could be stretched a little more which im playing with atm, but don't want to over do it

[TiffsAndAstro]

1 minute ago, Albir phil said:

Like I always say if you are happy with your image that's all that matters, after all other people will have different opinions on any image, take in what they say but if you like it that's ok🤗🤗

im actually borderline smug atm. 

i've been avoiding this area of the sky, elephants trunk nebula, ngc7000, pelican nebula as i assumed i'd get nothing from a stock dslr (and i've looked at a lot of people's stock dslr images on astrobin etc) but it seems (shockingly) i was completely wrong

Edited June 2 by TiffsAndAstro

[Albir phil]

Just now, TiffsAndAstro said:

im actually borderline smug atm. 

Nowt wrong with that👍

[TiffsAndAstro]

Just now, Albir phil said:

Nowt wrong with that👍

i think i need to have a lie down 

[Elp]

Looks quite nice. You'll know about the over processing by looking at your image histogram, black/shadows shouldn't be clipped (LHS of the graph, stars or highlights also shouldn't be clipped (RHS), the peaks shouldn't flatline anywhere off the graph and excess noise will show up as jagged peaks and troughs along the histogram. A "true" colour corrected image will also have the peaks of r, g and b near enough in the same position, colour calibration routine normally does this automatically or you can do it manually.

[TiffsAndAstro]

14 minutes ago, Elp said:

Looks quite nice. You'll know about the over processing by looking at your image histogram, black/shadows shouldn't be clipped (LHS of the graph, stars or highlights also shouldn't be clipped (RHS), the peaks shouldn't flatline anywhere off the graph and excess noise will show up as jagged peaks and troughs along the histogram. A "true" colour corrected image will also have the peaks of r, g and b near enough in the same position, colour calibration routine normally does this automatically or you can do it manually.

no data should be clipped in the above. siril has a handy display of how much is going to be cliped in  its ghs/istrograms and im usually very careful to not do that.

i photometric colour calibrate after i send it through graxpert (tiny crop, gradient and noise) after stacking. my thinking at the moment with a stock dslr is to try and just show what's there. 

i have a t least one dust spot my flats couldn't handle and i could paint/clone it out, but i don't think that sort of thing helps experienced people like yourselves constructively criticise and help

i still think sensor is a tad too close to the flattener and there is some tilt (?) on the left hand side. im hoping i can reduce that a bit next session by not using the focus lock screw thing quite so tightly. 

Edited June 2 by TiffsAndAstro

[Elp]

I clone stamp all the time, starnet leaves hideous star residue

[TiffsAndAstro]

4 minutes ago, Elp said:

I clone stamp all the time, starnet leaves hideous star residue

yeah im not quite up to that in gimp yet. but ill get there eventually. at the moment i just want to put lens flare on everything.

[carastro]

Yes I thought it was an unmodified DSLR.  

Well in that case it's a good result with the kit you have.  You sound as though you are keen enough to get into this in a big way and when you can I should consider upgrading both the camera and the mount,  I have an ED72 and whilst it is on the cheap end of the market, have got some pretty good resuts with it,  

Like your sense of humour BTW not only with your "crap gear" but some comments you made on another thread we were both on.   

Edited June 3 by carastro

[vlaiv]

To my eyes it looks very nice at 50%-67% zoom level.

At 100% zoom level - it shows what could be called over processed noise pattern (a bit more artificial and stronger noise levels than what I would call "natural" noise").

See this bit:

Very nice looking image, tight stars, and although there is noise in the image - it is close to "natural" levels - or unobtrusive.

However when we look at image like this:

noise levels start to bother me.

(btw these crops should be viewed at 100% to convey what I wanted to say - I'm aware that different devices apply different level of scale to images  - so view them at 100% if possible - btw 100% does not mean stretched across the screen, it means 1 image pixel maps to 1 device pixel).

[TiffsAndAstro]

1 hour ago, carastro said:

Yes I thought it was an unmodified DSLR.  

Well in that case it's a good result with the kit you have.  You sound as though you are keen enough to get into this in a big way and when you can I should consider upgrading both the camera and the mount,  I have an ED72 and whilst it is on the cheap end of the market, have got some pretty good resuts with it,  

Like your sense of humour BTW not only with your "crap gear" but some comments you made on another thread we were both on.   

Ty so much I feared this hobby could be a little humourless and very serious but with the weather here, I think humour is essential:(

Camera is #1 priority. Imx585 was top of my list but could be a long way off.

[david_taurus83]

Great result! I would keep at it and add more, maybe 5/6 hours worth. I took this on a single session with a redcat and my 6D a few days before or after the summer solstice so don’t listen to any naysayers that say you can’t image in summer time.

[TiffsAndAstro]

8 minutes ago, david_taurus83 said:

Great result! I would keep at it and add more, maybe 5/6 hours worth. I took this on a single session with a redcat and my 6D a few days before or after the summer solstice so don’t listen to any naysayers that say you can’t image in summer time.

That's really nice. Can clearly see the teeth and jaw of the alien queens skull:)

Is your 6d stock? 6d is full frame?

 

[TiffsAndAstro]

1 hour ago, vlaiv said:

To my eyes it looks very nice at 50%-67% zoom level.

At 100% zoom level - it shows what could be called over processed noise pattern (a bit more artificial and stronger noise levels than what I would call "natural" noise").

See this bit:

Very nice looking image, tight stars, and although there is noise in the image - it is close to "natural" levels - or unobtrusive.

However when we look at image like this:

noise levels start to bother me.

(btw these crops should be viewed at 100% to convey what I wanted to say - I'm aware that different devices apply different level of scale to images  - so view them at 100% if possible - btw 100% does not mean stretched across the screen, it means 1 image pixel maps to 1 device pixel).

Are you talking about the blobbyness or paint stipple effect? Because that's what I was afraid of. I'm messing with some filters in gimp which reduces it a bit, I think.

[vlaiv]

1 minute ago, TiffsAndAstro said:

Are you talking about the blobbyness or paint stipple effect? Because that's what I was afraid of. I'm messing with some filters in gimp which reduces it a bit, I think.

I'm not really sure how to put it.

Normal/natural noise follows certain distribution, or has certain range of intensities across different size scales. Most importantly it starts at single pixel level, so smallest noise grain is single pixel.

Here are examples.

left most image is just gray image with natural noise. Next in line is "denoised" image with some blurring effect. You can clearly see it has been blurred a bit - it starts loosing that finest level of grain, but larger components remain. Third image in line is another attempt to "control" the noise by using clipping on the low side, and last image in the row is combination of the two techniques - some blur (or sometimes people use median filter) and stretch that tries to mask the noise.

To my eye - first one looks the most natural and other show processing features in the noise. I actually don't object to image having a bit of natural noise, it does not bother me as much as poor noise handling - or even worse (and it happens a lot these days with AI based tools) - "plastic look" of the image where denoising was taken overboard and image looks very unnaturally smooth.

[david_taurus83]

14 minutes ago, TiffsAndAstro said:

That's really nice. Can clearly see the teeth and jaw of the alien queens skull:)

Is your 6d stock? 6d is full frame?

 

Thanks! The 6D is full frame yes and it has been modified.

[TiffsAndAstro]

16 minutes ago, vlaiv said:

I'm not really sure how to put it.

Normal/natural noise follows certain distribution, or has certain range of intensities across different size scales. Most importantly it starts at single pixel level, so smallest noise grain is single pixel.

Here are examples.

left most image is just gray image with natural noise. Next in line is "denoised" image with some blurring effect. You can clearly see it has been blurred a bit - it starts loosing that finest level of grain, but larger components remain. Third image in line is another attempt to "control" the noise by using clipping on the low side, and last image in the row is combination of the two techniques - some blur (or sometimes people use median filter) and stretch that tries to mask the noise.

To my eye - first one looks the most natural and other show processing features in the noise. I actually don't object to image having a bit of natural noise, it does not bother me as much as poor noise handling - or even worse (and it happens a lot these days with AI based tools) - "plastic look" of the image where denoising was taken overboard and image looks very unnaturally smooth.

i think you put this very very well those images show it pretty clearly and i only find the first one acceptable. my image shows some of the third and fouth images i think, so going forwards ill try and minimise getting those effects. not sure how yet but i hope ill learn. i've not really used the median filter in siril yet, so i guess ill start there.

im going to post process this image again from a point right after stacking. i'll probably end up with an almost completely different result

as for adding more hours to this, im not sure. the  composition isn't great,  plus i want to crop out about the left hand third because the star shapes are not nice. also, after this result i think i can take some decent images of other Ha targets in the same sort of area of the sky, heart nebula, elephant's trunk etc.

[alacant]

16 hours ago, TiffsAndAstro said:

over processed noise

Hi
Lovely shot. Especially given the short session. Oh and of course the crap gear (CG). Yes, I know exactly what you mean!

An easy way to avoid noise is to bin the data [1] before you process. Don't then though enlarge the image to the point where the noise and pixels become annoying.
There's also tilt you may want to correct. Hardly surprising bearing in mind the CG😉

Cheers and HTH

[1]
Siril > image processing > geometry > binning

[vlaiv]

54 minutes ago, TiffsAndAstro said:

i've not really used the median filter in siril yet, so i guess ill start there.

I probably was not clear in what I've said - but I mentioned median filter as one of best examples of filters that skews the noise the wrong way (although people used to use and recommend it quite a lot back in the day) - so don't use it

I think that best method to reduce noise is to blend denoised image with original image. This is best done using luminance of the image itself as mask layer. This produces "natural" looking result because:

a) it keeps some percentage of original noise - thus making it look natural

b) it reduces noise based on luminosity of image - which again is mimicking natural looking image that has more noise in darker areas and less in bright areas.

 

[TiffsAndAstro]

12 minutes ago, vlaiv said:

I probably was not clear in what I've said - but I mentioned median filter as one of best examples of filters that skews the noise the wrong way (although people used to use and recommend it quite a lot back in the day) - so don't use it

I think that best method to reduce noise is to blend denoised image with original image. This is best done using luminance of the image itself as mask layer. This produces "natural" looking result because:

a) it keeps some percentage of original noise - thus making it look natural

b) it reduces noise based on luminosity of image - which again is mimicking natural looking image that has more noise in darker areas and less in bright areas.

 

I believe you but I'll have to research this. I've not done any blending and I think reason I avoided median filter is I didn't like it's results. I put this down to not understanding it.