Please, please, please...isn't there a way to have a width of fewer than 60 pixels still work in the noise analysis process? Sometimes the worst noise in my images is in the shadow areas where I have pulled the detail out using Shadow/Highlights in Photoshop CS. I seldom have imges at all with large enough areas ofno detail to staisfiy NI. But what about a 30 x 120 pixel area, for example Or, the possiblility of making several simultaneous little rectangular selections for NI to determine noise?
I tried Noise Ninja, and while the increased speed and ease is impressive, what really impressed me was the ability to use smaller selections to determine noise levels. I feel loyal to NI, but would like to see a change in this area.
Smaller noise-sampling rectangles?
WalterK,
We could reduce the size requrements for analysis area but this can become detrimental to the profile quality and filtration results when used by beginners. Would you be ready to accept lower filtration quality? Would you continue to use the software if by mistake you, as a beginner, selected too small area and got inaccurate profile? - Maybe it is easier to use the calibration target in case there is not enough space in the working image?
WalterK, you are comparing NI with its clones. What do you consider more difficult with Neat Image? It is processing speed or something else?
With respect to profiling using smaller areas, as far as I know, you anyway need to use a calibration target, which provides large featureless areas, so this seems to be no different.
BTW, we have been always working on improvements. You can expect further speed up in Neat Image processing and, what we consider the most important, the filtration quality improvements.
Vlad
We could reduce the size requrements for analysis area but this can become detrimental to the profile quality and filtration results when used by beginners. Would you be ready to accept lower filtration quality? Would you continue to use the software if by mistake you, as a beginner, selected too small area and got inaccurate profile? - Maybe it is easier to use the calibration target in case there is not enough space in the working image?
WalterK, you are comparing NI with its clones. What do you consider more difficult with Neat Image? It is processing speed or something else?
With respect to profiling using smaller areas, as far as I know, you anyway need to use a calibration target, which provides large featureless areas, so this seems to be no different.
BTW, we have been always working on improvements. You can expect further speed up in Neat Image processing and, what we consider the most important, the filtration quality improvements.
Vlad
WalterK,
A true word is spoken in jest and in fact we are considering something like this. - Not exam of course, but some kind of advanced mode with more control provided to an experienced user than to a beginner.
On the other hand, there is still a quality limit - the smaller area is used for analysis the less accurate becomes filtration in low and mid frequency ranges. - You need to sample a large area to measure low frequency noise, there is mathematically not enough information to measure it using a small area.
Reducing both high frequency noise and noise in lower frequencies is one of the stronger points of Neat Image. It would be a pity to drop this and degrade to the level of average noise reducers that deal with something like 8x8 pixel areas.
Still, we are considering possible options.
Vlad
A true word is spoken in jest and in fact we are considering something like this. - Not exam of course, but some kind of advanced mode with more control provided to an experienced user than to a beginner.
On the other hand, there is still a quality limit - the smaller area is used for analysis the less accurate becomes filtration in low and mid frequency ranges. - You need to sample a large area to measure low frequency noise, there is mathematically not enough information to measure it using a small area.
Reducing both high frequency noise and noise in lower frequencies is one of the stronger points of Neat Image. It would be a pity to drop this and degrade to the level of average noise reducers that deal with something like 8x8 pixel areas.
Still, we are considering possible options.
Vlad
This explains something I noticed. Once I broke the sampling rule on purpose to see if it would solve a problem with some unusual noise that was just in one part of the image. And Neat Image solved the problem. Another time I used no manual sample, because all I needed was some smoother raw material for something else. The result was not as good as using a manual sample, but it was pretty impressive.NITeam wrote:They have no time even to read the user guide and the software have to work even in such conditions.
I just checked, and it looks to me like Noise Ninja does NOT allow for a smaller primary sampling area. Its fixed-size sampling area appears to be the same size as Neat Image's minimum. Neat Image also allows the user to go smaller than the minimum size, but warns that the sample is too small to get a good sample. And, of course, Neat Image can also go larger if there is enough room.
The fine tune sample is a different story. Noise Ninja has a fixed size that is smaller than Neat Image's minimum. I'm not sure that this is as important, though, considering that someone can do an auto-fine tune. But I guess it would it would offer some advantage if it could be smaller and this worked just as well.
The fine tune sample is a different story. Noise Ninja has a fixed size that is smaller than Neat Image's minimum. I'm not sure that this is as important, though, considering that someone can do an auto-fine tune. But I guess it would it would offer some advantage if it could be smaller and this worked just as well.
The fine tune sample is a different story. Noise Ninja has a fixed size that is smaller than Neat Image's minimum. I'm not sure that this is as important, though, considering that someone can do an auto-fine tune. But I guess it would it would offer some advantage if it could be smaller and this worked just as well.
That's my point...If there were some way to place several smaller squares simultaneously, and have NeatImage consider them as one, then many images would be able to get a real FineTune, where the noise matters, but in smaller, detail-less areas. The AutofineTune function in NI, as far as I can see, has to guess about what is noise and what is detail.
WalterK
Ah, I see what you are saying now. But I should point out that Neat Image can already do a 30 by 120 selection for the fine tuning sample. Even 30 by 30.
OTOH, I don't want to step on your request. If you really wanted it to go down as far as 15x15 for fine tuning, then I agree that would make things easier on some images, if auto fine tune doesn't seem to be working well enough.
OTOH, I don't want to step on your request. If you really wanted it to go down as far as 15x15 for fine tuning, then I agree that would make things easier on some images, if auto fine tune doesn't seem to be working well enough.