RawMagick's MSU noise reduction???

Joined
Jan 7, 2006
Messages
955
Location
Detroit
I gave a pointer to some lady on that other forum, for trial downloads of raw converters, NX and RawMagick Lite. While on the RawMagick site, I noticed some new stuff, the MSU noise reduction plug-in and BlendMagick.

So, I went back to RML site a little while ago and bought the MSU plug-in, but don't know much about it. Can't believe I haven't heard of it before. :confused: I gotta get out from under my rock more often, I guess...

Anyone here using it?

BlendMagick, if it does what the hints say, would be very cool also. Sometimes I just fight with the colors, rather than correcting them. :Sick:
 
Joined
Apr 24, 2005
Messages
6,099
Location
Denver, CO
I use MSU when needed.

Iliah would be the expert here, but I think his summer work schedule has started and his time on here may be sporadic for a while.

I will try to help.
 
Joined
Jan 7, 2006
Messages
955
Location
Detroit
Thanks, Charles. I appreciate the offer. I was just curious as to what others with more experience have found to be useful settings with the plug-in. Of course, I assume that much depends on the image and amount of noise. :smile:
 
Joined
Apr 24, 2005
Messages
6,099
Location
Denver, CO
Like you said, it depends on the image and the intended output.

I usually do 2 passes. First pass to remove chroma noise, next one to reduce luma noise. If the image is going to be printed I will be leave some noise as it helps reduce banding in inkjet prints. Controlled noise is usefull.
 
Joined
Jan 7, 2006
Messages
955
Location
Detroit
That's interesting. I don't recall having seen a 2 pass technique like that before. I'll have to give that a look. Thanks. :smile:
 
Joined
Jan 7, 2006
Messages
955
Location
Detroit
Let me know what you think of it. :smile:
Will do, Charles. I have a bunch of ISO1600 wedding and ISO3200 sports shots to process, so while I'm doing that, I'll give it a play in noiseware, to see what it does with that program. :wink:
 
Joined
Apr 24, 2005
Messages
6,099
Location
Denver, CO
I no longer use those noise removal apps. MSU only.

You can also save an image from photoshop as a tiff. Open that in RML and move to phase II and use MSU and other filters. :wink:
 
Joined
Aug 13, 2006
Messages
3,235
Location
Berlin, Old Europe
My approach is usually to get rid of chroma noise as much as possible, then decide whether I also want to fight luma noise as well.
To do so I usually use the demosaicing settings in RML (chroma noise slider) which yields already very good results when max'ed (though takes long to process).

My current experience is that if processed that way, usually MSU will not reduce chroma noise more than that.

For luma noise I found that I live with it most of the time or reduce it only very slightly, because I don't want to loose details and can't stand that posterized look of de-noised images.
Usually, when scaled down for web viewing or printing, the remaining noise can't be seen anyway.

Interesting question from my perspective then would be: Which approach is better: Handling chroma noise in RML demosaicing or afterwards in MSU?
 
Joined
Apr 24, 2005
Messages
6,099
Location
Denver, CO
His question was about MSU so I did not go into noise reduction in phase I. You pose an interesting question. One that I have not tested nor have I asked Iliah about it. Next time I am in contact with him I will ask.:wink:

My approach is usually to get rid of chroma noise as much as possible, then decide whether I also want to fight luma noise as well.
To do so I usually use the demosaicing settings in RML (chroma noise slider) which yields already very good results when max'ed (though takes long to process).

My current experience is that if processed that way, usually MSU will not reduce chroma noise more than that.

For luma noise I found that I live with it most of the time or reduce it only very slightly, because I don't want to loose details and can't stand that posterized look of de-noised images.
Usually, when scaled down for web viewing or printing, the remaining noise can't be seen anyway.

Interesting question from my perspective then would be: Which approach is better: Handling chroma noise in RML demosaicing or afterwards in MSU?
 
Joined
Jan 29, 2005
Messages
2,761
Location
nowhere
Noise "reduction" in demosaicing works better with some cameras, and depends on ISO. I use simple resolution test to compare noise "reduction" in Phase I to MSU NR in Phase II.

Noise reduction needs depend on the print size. To get an idea of the amount of noise reduction I need I examine the image at 66% or 100% print size (66% if I use LCD monitor, 100% if it is CRT), not at 100% screen view.
 
Joined
Jan 29, 2005
Messages
2,761
Location
nowhere
> I usually do 2 passes. First pass to remove chroma noise, next one to reduce luma noise.

Same here. First chroma, then Luma.


> If the image is going to be printed I will be leave some noise as it helps reduce banding in inkjet prints

Digital has very high noise levels in highlights. It is easier to hide noise behind noise, so sometimes I need to add noise after noise removal, to turn blotches into grain. One of good ways to get rid of unpleasant noise is pixel-level sharpening of luminosity channel.
 
Joined
Aug 13, 2006
Messages
3,235
Location
Berlin, Old Europe
Noise "reduction" in demosaicing works better with some cameras, and depends on ISO. I use simple resolution test to compare noise "reduction" in Phase I to MSU NR in Phase II.
Maybe I'll do some comparison for my D200 in the next days, using even simpler "see what looks better" comparison. :wink:
There's one parameter I'm a bit unsure about, which is the sharpness slider in phase 1. It looks it turns noise blotches (low) into finer grained noise (high), and I don't know which kind of noise appearance is easier to handle for MSU. Any recommendations?

Noise reduction needs depend on the print size. To get an idea of the amount of noise reduction I need I examine the image at 66% or 100% print size (66% if I use LCD monitor, 100% if it is CRT), not at 100% screen view.
That's roughly what I do.
 
Joined
Jan 29, 2005
Messages
2,761
Location
nowhere
> There's one parameter I'm a bit unsure about, which is the sharpness slider in phase 1.

I keep it at 3/4 for shots with for Kodak SLR/n and D2X taken at base ISO and at highest for shots above base ISO. D200 seems less noisy, and 3/4 setting makes sense to me up to ISO 400 equivalent. For D2H and D2Hs I use full at ISO 800 equivalent and above. Grainy noise prints better, while blotches are more difficult to deal with in post-processing noise reduction, including noise reduction in Phase II.
 
Joined
Mar 18, 2005
Messages
22,474
Location
Richmond, Va
Iliah
I went to the Rawmagic site to purchase the product and start to learn it but it appears to be PC only at this time. I will send you a PM regarding this.
 
Joined
Jan 7, 2006
Messages
955
Location
Detroit
I no longer use those noise removal apps. MSU only.

You can also save an image from photoshop as a tiff. Open that in RML and move to phase II and use MSU and other filters. :wink:
Interesting. Dunno if that would work for me. The vast majority of my high ISO stuff is high volume sports shooting. I may give it a go on some of my wedding shots though.

Thanks again for the info. :biggrin:
 
Joined
Apr 24, 2005
Messages
6,099
Location
Denver, CO
>Digital has very high noise levels in highlights. It is easier to hide noise behind noise, so sometimes I need to add noise after noise removal, to turn blotches into grain. One of good ways to get rid of unpleasant noise is pixel-level sharpening of luminosity channel.
I usually add monochromatic noise in PS and paint it in where I need it. Will look into the pixel level sharpening.
 
Joined
Aug 13, 2006
Messages
3,235
Location
Berlin, Old Europe
Here's my unscientific comparison of phase I vs phase II chroma noise reduction:

This mediocre shot has to serve as test object:
Subscribe to see EXIF info for this image (if available)


I processed this photo four times with phase 1 "colour noise" and "sharpness" sliders varied each to min/max, then I took each result and tried to reduce chroma noise in MSU further. (I also tried luma NR with value 1 or 2; more would have destroyed too much detail.)
Here are the results (100% and 50% crops from below her right ear):

Subscribe to see EXIF info for this image (if available)


Subscribe to see EXIF info for this image (if available)


My quick conclusion (at least for my D200):

In case of chroma noise, you should fight it in demosaicing first as this seems to be the best basis for further processing. IF needed you can then further reduce chroma noise in MSU, but only slightly, as the input is already nearly chroma noise free.

Increasing the sharpness value wasn't of much help for MSU as it seems, so for this kind of noise amount, a high colour noise slider value and a quite low sharpness slider value would be my choice. Then see if further processing in MSU is required.
 
Links on this page may be to our affiliates. Sales through affiliate links may benefit this site.
Nikon Cafe is a fan site and not associated with Nikon Corporation.
Forum post reactions by Twemoji: https://github.com/twitter/twemoji
Copyright © 2005-2019 Amin Forums, LLC
Top Bottom