Rumors about new photo noise reduction software?

[CarlTN]

Hello to everyone.  Recently there was a similar topic about noise reduction, specifically what is the best currently available? 

My question is, does anyone have any inside information, or have you heard any interesting rumors, about near-future releases, perhaps new versions of current software, specifically "third party", and non-Adobe-based?

Also, in case CanonRumors is not the best place to ask, what photo or software-related websites might be on the cutting edge of such rumored information?

It just seems like the late spring is when much of the software gets updated, so I thought I would ask.

[RGF]

I would expect the major plug-in providers (Nik, OnOne, Topaz) as well as ninja noise to continually update their products.  I have no knowledge if they will make major improvements or minor tweaks.  Time will tell.

[Mt Spokane Photography]

Noise reduction software all works the same way, some of it has better algorithms to decide which pixels to smear, but that's how its done, smearing pixels to reduce detail and eliminate the grainy look.
 
Its best to apply NR selectively to just the areas that benefit.  Photoshop is great for that, but even Lightroom allows layers to be added with NR in the areas needed.
 
You can use Photoshop and a NR software plugin to get the best of both worlds.
 
Don't expect any new revolutionary technology in software, that has to be done at the camera body level.  NR makers put out a lot of hype, but most of them also have limited trials.  Unfortunately, the last time I tried, the trials were 8 bit only which made them worthless to me.  That might have changed.
 
 

[CarlTN]

MtSpokane, thanks for your words of wisdom.  It turns out you're not telling me anything I don't already know.  I disagree with your sentiment that they all work the same way.  The object of noise reduction, is not to blur detail.  It is to preserve detail and eliminate noise.  They certainly do not all work the same way.  Some are better with chrominance, others with luminance.  If all they did was blur detail, then we could get the same results by simply applying softening to the image.  The results are not the same at all, though.  So for you to assume there will be no further advancement in noise reduction algorithms, or approaches, is rather absurd and wrong-minded.  No offense.  You might as well just suggest that deleting the image would be the simplest and best way to clean it.  There's always a better mousetrap.  As for changing to a better performing camera, I'm already planning to add a terrific one to my tool collection.  But as a fan of low-light photography, and shooting at ISO higher than the base level...new noise reduction is especially interesting to me.  Since noise reduction does not interest you, perhaps you need not even waste your time commenting.

[Pi]

He was basically right. There is no difference between noise and detail once the image is captured. Noise IS detail. Yes, all NR algorithms blur detail but they do it in a different way, hence different algorithms. NR can be very effective under the assumption that the image has a certain structure, like well-defined curves, etc. A well designed NR in that case can do much better job than Gaussian blur, for example. But it will destroy texture, etc., creating the well known cartoonish look.

MtSpokane, thanks for your words of wisdom.  It turns out you're not telling me anything I don't already know.  I disagree with your sentiment that they all work the same way.  The object of noise reduction, is not to blur detail.  It is to preserve detail and eliminate noise.  They certainly do not all work the same way.  Some are better with chrominance, others with luminance.  If all they did was blur detail, then we could get the same results by simply applying softening to the image.  The results are not the same at all, though.  So for you to assume there will be no further advancement in noise reduction algorithms, or approaches, is rather absurd and wrong-minded.  No offense.  You might as well just suggest that deleting the image would be the simplest and best way to clean it.  There's always a better mousetrap.  As for changing to a better performing camera, I'm already planning to add a terrific one to my tool collection.  But as a fan of low-light photography, and shooting at ISO higher than the base level...new noise reduction is especially interesting to me.  Since noise reduction does not interest you, perhaps you need not even waste your time commenting.

[D.Fordice]

You're all partially right.  Current software is not very advanced.  However, there are detectable differences between some types of detail and noise, but it is very hard to distinguish with current software.  The way it could be done by more complex software is by better "recognizing" patterns (lines, areas, and edges) and then using complex algorithms to preserve these patterns while eliminating the noise within them.  Secondarily, the algorithms can be "trained" to recognize the types and patterns of noise generated by particular cameras at particular ISO settings in order to more effectively eliminate that noise.  This software could also have advanced scene recognition capabilities, which enable it to better apply the right type of sharpening to the right type of areas (for example, if the software recognized a face, it could make the irises and eyelashes really sharp while evening out the skin tones (colors as well)).  These are just a few examples of what could be done, but there is a whole lot more.

[Pi]

You're all partially right.  Current software is not very advanced.  However, there are detectable differences between some types of detail and noise...

That is the whole point. What if your detail is not of the "some types" - like random structures in landscapes, etc.

As I said, NR can be very effective under the assumptions that we know more or less what the original looks like. Like the faces you mention, or detecting sky and blurring what does not look like stars. But what if the sky turns out to be a blue mural with its natural texture?

[TrumpetPower!]

Noise reduction is never going to be perfect. A close-up shot of some rough granite, or even a photo of the TV tuned to a station that's gone off the air...I can't imagine any general-purpose noise reduction software that wouldn't completely obliterate those types of things.

But the good news is that noise is so well controlled in modern DSLRs. Even at insane ISOs like 25,600 on the 5DIII, if you knock down the chroma noise, what's left isn't at all objectionable. Since you're basically shooting in light that's too dark to see in by that point, the grain you get just puts things in a very artistically-suitable context. It's like the cricket chirp comedic sound effect when a character is left dumbfounded and silence resounds, except it's not corny.

And in cases where you actually don't want noise but you do need to shoot in pitch black, such as astrophotography, there are techniques (a tracking mount and / or stacking multiple exposures) to once again eliminate the noise.

Were I in the business of selling noise reduction software, I'd be putting my effort into diversification with an eye towards a long-term exit strategy. The days of standalone noise reduction applications are numbered, and the number isn't very large.

Cheers,

b&

[CarlTN]

Interesting replies.  Stand alone noise reduction will eventually go away, huh?  Nah, don't think so.  I was under the impression that some of the NR techniques already employ methods to recognize noise patterns, specific to the camera and the ISO level.  And if long exposure noise reduction can be applied this way, it seems to me that "gain" noise reduction has further room for improvement.

Whether or not luminance noise, or grain, is objectionable, pretty much always has to do with just how much cropping you're doing.  If there is no cropping and the image isn't viewed very large, whether on screen or in a print, then certainly it is not objectionable.  But if you're cropping in on some wildlife at dusk, the grain is more objectionable.

[TrumpetPower!]

Interesting replies.  Stand alone noise reduction will eventually go away, huh?  Nah, don't think so.  I was under the impression that some of the NR techniques already employ methods to recognize noise patterns, specific to the camera and the ISO level.  And if long exposure noise reduction can be applied this way, it seems to me that "gain" noise reduction has further room for improvement.

What makes you think that won't all be done onboard the camera and / or in the raw development software?

Whether or not luminance noise, or grain, is objectionable, pretty much always has to do with just how much cropping you're doing.  If there is no cropping and the image isn't viewed very large, whether on screen or in a print, then certainly it is not objectionable.  But if you're cropping in on some wildlife at dusk, the grain is more objectionable.

Today?

Sure.

But, again, what makes you think tomorrow's cameras will be as noisy and that their noise will be as objectionable? Especially since the trend has strongly been in the opposite direction, with no end in sight?

Not that long ago there used to be a huge market in image enlargement software, but that's largely been rendered obsolete by increasing camera resolution and the quality of enlargement that Photoshop and the printer drivers deliver out of the box.

The exact same thing is already happening to noise reduction software.

Cheers,

b&

[brett b]

MtSpokane, thanks for your words of wisdom.  It turns out you're not telling me anything I don't already know.  I disagree with your sentiment that they all work the same way.  The object of noise reduction, is not to blur detail.  It is to preserve detail and eliminate noise.  They certainly do not all work the same way.  Some are better with chrominance, others with luminance.  If all they did was blur detail, then we could get the same results by simply applying softening to the image.  The results are not the same at all, though.  So for you to assume there will be no further advancement in noise reduction algorithms, or approaches, is rather absurd and wrong-minded.  No offense.  You might as well just suggest that deleting the image would be the simplest and best way to clean it.  There's always a better mousetrap.  As for changing to a better performing camera, I'm already planning to add a terrific one to my tool collection.  But as a fan of low-light photography, and shooting at ISO higher than the base level...new noise reduction is especially interesting to me.  Since noise reduction does not interest you, perhaps you need not even waste your time commenting.

Wow. What in Mt. Spokane's reply deserved this kind of response?

[Mt Spokane Photography]

MtSpokane, thanks for your words of wisdom.  It turns out you're not telling me anything I don't already know.  I disagree with your sentiment that they all work the same way.  The object of noise reduction, is not to blur detail.  It is to preserve detail and eliminate noise.  They certainly do not all work the same way.  Some are better with chrominance, others with luminance.  If all they did was blur detail, then we could get the same results by simply applying softening to the image.  The results are not the same at all, though.  So for you to assume there will be no further advancement in noise reduction algorithms, or approaches, is rather absurd and wrong-minded.  No offense.  You might as well just suggest that deleting the image would be the simplest and best way to clean it.  There's always a better mousetrap.  As for changing to a better performing camera, I'm already planning to add a terrific one to my tool collection.  But as a fan of low-light photography, and shooting at ISO higher than the base level...new noise reduction is especially interesting to me.  Since noise reduction does not interest you, perhaps you need not even waste your time commenting.

You can believe anything you want, that does not change the fact that NR blurs detail, its just a matter of which detail gets blurred. 
 
Apparently you missed the part where I said that  "some of it has better algorithms to decide which pixels to smear, but that's how its done, smearing pixels to reduce detail and eliminate the grainy look."
 
Undoubtedly, there will be more and better algorithms as more powerful computers are able to handle it, but you can also apply what there is today selectively to make the best use of what is out there. 
 
NASA has super computers that are able to do a much better job of noise removal than we can.  I believe that our computers are the limiting factor, someone editing 500 images may not want to wait if NR takes a excessive time.  With my low light D800 images, I had to wait far too long for NR to complete with a ISO 12800 image, I just could not spend the time for a large shoot.
 
I also do a lot of low light photography and use NR, so I'm quite interested in NR, and have tried them all.  I have also read all their hype, and find it to be just that.  You have to try it, and when they limit you to 8 bit trials, the results are not usable for me.
 
In general, I have found that the different software packages do a good job with certain types of images, and poor with others. 
 
Good Luck!
.
 
 
 

[Mt Spokane Photography]

MtSpokane, thanks for your words of wisdom.  It turns out you're not telling me anything I don't already know.  I disagree with your sentiment that they all work the same way.  The object of noise reduction, is not to blur detail.  It is to preserve detail and eliminate noise.  They certainly do not all work the same way.  Some are better with chrominance, others with luminance.  If all they did was blur detail, then we could get the same results by simply applying softening to the image.  The results are not the same at all, though.  So for you to assume there will be no further advancement in noise reduction algorithms, or approaches, is rather absurd and wrong-minded.  No offense.  You might as well just suggest that deleting the image would be the simplest and best way to clean it.  There's always a better mousetrap.  As for changing to a better performing camera, I'm already planning to add a terrific one to my tool collection.  But as a fan of low-light photography, and shooting at ISO higher than the base level...new noise reduction is especially interesting to me.  Since noise reduction does not interest you, perhaps you need not even waste your time commenting.

I apologize if I hit your trigger, I did not mean to.
 
As you said, NR software will get updated every year. Undoubtedly, there will be more and better algorithms as more powerful computers are able to handle it, but you can also apply what there is today selectively to make the best use of what is out there. 
 
As far as what is best, I find that the different packages are better for a certain type of image, so it really depends on the subject as to what is best, and I've not been to define that.
 
NASA has super computers that are able to do a much better job of noise removal than we can.  I believe that our computers are the limiting factor, someone editing 500 images may not want to wait if NR takes a excessive time.  With my low light D800 images, I had to wait far too long for NR to complete with a ISO 12800 image, I just could not spend the time for a large shoot.
 
I also do a lot of low light photography and use NR, so I'm quite interested in NR, and have tried them all.  I have also read all their hype, and find it to be just that.  You have to try it, and when they limit you to 8 bit trials, the results are not usable for me.
 

 
Good Luck!
.

[CarlTN]

MtSpokane, thanks for your words of wisdom.  It turns out you're not telling me anything I don't already know.  I disagree with your sentiment that they all work the same way.  The object of noise reduction, is not to blur detail.  It is to preserve detail and eliminate noise.  They certainly do not all work the same way.  Some are better with chrominance, others with luminance.  If all they did was blur detail, then we could get the same results by simply applying softening to the image.  The results are not the same at all, though.  So for you to assume there will be no further advancement in noise reduction algorithms, or approaches, is rather absurd and wrong-minded.  No offense.  You might as well just suggest that deleting the image would be the simplest and best way to clean it.  There's always a better mousetrap.  As for changing to a better performing camera, I'm already planning to add a terrific one to my tool collection.  But as a fan of low-light photography, and shooting at ISO higher than the base level...new noise reduction is especially interesting to me.  Since noise reduction does not interest you, perhaps you need not even waste your time commenting.

I apologize if I hit your trigger, I did not mean to.
 
As you said, NR software will get updated every year. Undoubtedly, there will be more and better algorithms as more powerful computers are able to handle it, but you can also apply what there is today selectively to make the best use of what is out there. 
 
As far as what is best, I find that the different packages are better for a certain type of image, so it really depends on the subject as to what is best, and I've not been to define that.
 
NASA has super computers that are able to do a much better job of noise removal than we can.  I believe that our computers are the limiting factor, someone editing 500 images may not want to wait if NR takes a excessive time.  With my low light D800 images, I had to wait far too long for NR to complete with a ISO 12800 image, I just could not spend the time for a large shoot.
 
I also do a lot of low light photography and use NR, so I'm quite interested in NR, and have tried them all.  I have also read all their hype, and find it to be just that.  You have to try it, and when they limit you to 8 bit trials, the results are not usable for me.
 

 
Good Luck!
.

Thanks, and good luck to you too, but it sounds like you don't need any.  Not sure why you wanted to quote yourself and post again, but I guess you like having a lot of posts.

What is your primary business, if I may ask?  You have hundreds of images shot at ISO 12,800?  Seems like a lot for a photo shoot.  And where are these NASA supercomputers located?  I was under the impression that NASA has a lot of budget cutting going on, as per the current administration's wishes.

A university here, has had a computer lab which NASA has supposedly subcontracted over the last few years...I think they used it to process and model some of the data after the Columbia accident in 2003.  Don't think they've used them for image processing and noise reduction, but it's possible. 

I assume the noise processing you're referring to, done by NASA, has to do with deep space astronomy?  Or perhaps also to do with images shot by the Mars Rovers?  Care to elaborate?

[chauncey]

I have no experience with NR software except for PS CS6 which I'm satisfied with.
I've found that the key is to use various selection methods to "chop up" that image and use different amounts of NR in different parts of the image.