New version of iWE Raw Editor (free replacement for DXO and Denoise AI)

[ProDude]

Well I am not stupid enough to use my real name in a forum like you, shows how much you know about security

ROFLMAO you really nailed that one Mud Flap.

You are not the sharpest tool in the shed are you How long do you think it would take to ID you with the host of information provided . maybe 1 min letting your EGO get the better of you . James is a very very common name in Scotland Stirling just very common good luck with that

But he's a PRO !!!!

A pro bull-----er more like ;-)

You clowns (of which proliferate on this forum in general) can sling all the mud you want. As long as I"m getting paid to do what I do last I checked that can be considered professional. I fully admit I'm blunt and don't tend to soften edges around here. No doubt it rubs some the wrong way. I'm used to it. Would rather say it like it IS then add mush.

--
Name the gear and I've probably owned it and used it.

 

[James Stirling]

Well I am not stupid enough to use my real name in a forum like you, shows how much you know about security

ROFLMAO you really nailed that one Mud Flap.

You are not the sharpest tool in the shed are you How long do you think it would take to ID you with the host of information provided . maybe 1 min letting your EGO get the better of you . James is a very very common name in Scotland Stirling just very common good luck with that

I'm not exactly trying to hide in the shadows now am I. LOL

No but you seemed quite amused by Mud's security comment about using your own name . When you have links that identify , who you are and where you come from , just add bank details and make a lot of Nigerians happy

--
Jim Stirling:
“It is one thing to show a man that he is in error, and another to put him in possession of truth.” Locke
Feel free to tinker with any photos I post

 

[James Stirling]

Well I am not stupid enough to use my real name in a forum like you, shows how much you know about security

ROFLMAO you really nailed that one Mud Flap.

You are not the sharpest tool in the shed are you How long do you think it would take to ID you with the host of information provided . maybe 1 min letting your EGO get the better of you . James is a very very common name in Scotland Stirling just very common good luck with that

But he's a PRO !!!!

A pro bull-----er more like ;-)

You clowns (of which proliferate on this forum in general) can sling all the mud you want. As long as I"m getting paid to do what I do last I checked that can be considered professional. I fully admit I'm blunt and don't tend to soften edges around here. No doubt it rubs some the wrong way. I'm used to it. Would rather say it like it IS then add mush.

The pro thing is a silly label there are dozens of professional { paid for their photography } folk on the forum. Having done it for 20yrs myself "pro" is no guarantee of good work or knowledge for that matter { not aiming at you at all } . The biggest reasons for folk failing to make money in photography are poor people skills and poor money management. The term pro has also been devalued with "pro" being added to all kinds of trivia. I am very happy to have made my main income elsewhere and can now just enjoy photography for my own pleasure.



--
Jim Stirling:
“It is one thing to show a man that he is in error, and another to put him in possession of truth.” Locke
Feel free to tinker with any photos I post

 

[ProDude]

, who you are and where you come from , just add bank details and make a lot of Nigerians happy

That's not even funny. Haven't we all gotten email from those SOB's about transferring funds to them. Interestingly enough I was actually on Dpreview forums (Canon and Nikon and Olympus) since 1999 back when Phil Askey owned it. My username got messed up and I had to switch out of it in 2018. So, I've been around a lot longer then what is indicated.

 

[ProDude]

The pro thing is a silly label there are dozens of professional { paid for their photography } folk on the forum. Having done it for 20yrs myself "pro" is no guarantee of good work or knowledge for that matter { not aiming at you at all } . The biggest reasons for folk failing to make money in photography are poor people skills and poor money management. The term pro has also been devalued with "pro" being added to all kinds of trivia. I am very happy to have made my main income elsewhere and can now just enjoy photography for my own pleasure.

Yeah I get it. The dictionary states "Professional" as one that earns an income from it. so I suppose I may have fit the bill. Since 2000 all my work has come from referrals. I never paid for advertising in the past. I just came up with the nick name on a lark. Guess in some ways perhaps I should have thought more about it. Around here it's just putting a target on oneself.

 

[James Stirling]

The pro thing is a silly label there are dozens of professional { paid for their photography } folk on the forum. Having done it for 20yrs myself "pro" is no guarantee of good work or knowledge for that matter { not aiming at you at all } . The biggest reasons for folk failing to make money in photography are poor people skills and poor money management. The term pro has also been devalued with "pro" being added to all kinds of trivia. I am very happy to have made my main income elsewhere and can now just enjoy photography for my own pleasure.

Yeah I get it. The dictionary states "Professional" as one that earns an income from it. so I suppose I may have fit the bill. Since 2000 all my work has come from referrals. I never paid for advertising in the past. I just came up with the nick name on a lark. Guess in some ways perhaps I should have thought more about it. Around here it's just putting a target on oneself.

I should have been a bit politer myself I drift between the frank and rude myself . Photography was always just a profitable side line to my main interests and I don't miss the hassle of "pro" shooting one little bit

 

[James Stirling]

, who you are and where you come from , just add bank details and make a lot of Nigerians happy

That's not even funny. Haven't we all gotten email from those SOB's about transferring funds to them. Interestingly enough I was actually on Dpreview forums (Canon and Nikon and Olympus) since 1999 back when Phil Askey owned it. My username got messed up and I had to switch out of it in 2018. So, I've been around a lot longer then what is indicated.

I know not everyone is computer literate but it is hard to believe folk fall for the Nigerian prince skit. Though I suppose if you send out enough emails they will get a bite otherwise they wouldn't be doing it . No doubt using computers donated to them

--


Jim Stirling:
“It is one thing to show a man that he is in error, and another to put him in possession of truth.” Locke
Feel free to tinker with any photos I post

 

[ProDude]

Photography was always just a profitable side line to my main interests and I don't miss the hassle of "pro" shooting one little bit

I hear ya. I no longer do large weddings. At 68 I don't have the energy (or the back) to tolerate a 8-10 hour day of all that fun. Those days are over for me. I prefer a home shoot for a realtor or family portrait shot during the holidays and such. An easy buck. I have a Canon Pro-1000 printer so can offer top quality large prints if called for in house.

 

[ProDude]

, who you are and where you come from , just add bank details and make a lot of Nigerians happy

That's not even funny. Haven't we all gotten email from those SOB's about transferring funds to them. Interestingly enough I was actually on Dpreview forums (Canon and Nikon and Olympus) since 1999 back when Phil Askey owned it. My username got messed up and I had to switch out of it in 2018. So, I've been around a lot longer then what is indicated.

I know not everyone is computer literate but it is hard to believe folk fall for the Nigerian prince skit. Though I suppose if you send out enough emails they will get a bite otherwise they wouldn't be doing it . No doubt using computers donated to them

I love the ones that say it's a lawyer representing a poor soul that is about to pass away and they want to donate a couple million to you. Just forward your information to them and they will transfer the funds. LOL. Yeah one way, OUT of your account.

 

[James Stirling]

Photography was always just a profitable side line to my main interests and I don't miss the hassle of "pro" shooting one little bit

I hear ya. I no longer do large weddings. At 68 I don't have the energy (or the back) to tolerate a 8-10 hour day of all that fun. Those days are over for me. I prefer a home shoot for a realtor or family portrait shot during the holidays and such. An easy buck. I have a Canon Pro-1000 printer so can offer top quality large prints if called for in house.

I stopped about 5yrs ago now. Awhile back I did my nieces wedding { I had promised her years before } .Unfortunately it landed shortly after I had a knee replacement and I spent a painful day hobbling about using a monopod as a crutch I mentioned in another post recently how it was my first wedding using mirrroless { and only no more } and I forgot that I was shooting in silent mode and for a few heart stopping moments thought I had missed a vital part of the ceremony , not fun as I was looking at my smiling sister at here daughters wedding and imagining her chasing me about with the cake knife

--
Jim Stirling:
“It is one thing to show a man that he is in error, and another to put him in possession of truth.” Locke
Feel free to tinker with any photos I post

 

[Searching]

Replacement for DXO seems misleading.

 

[Serguei Palto]

Replacement for DXO seems misleading.

I have replaced. I have DXO PL4, but now never use it. I have Topaz Denoise AI, but now never use it except for demonstration. The problem of DXO and Topaz - AI artifacts, which I can not accept. Sure, many of users can even not see them. Below are just simple illustrations with about one stop underexposed image (#1, #2) taken at ISO 5000 (so one can think on it as taken at ISO of about 10000) . In #3, #4 the crop examples of my original post image, where #3 is processed in raw model of Topaz Denoise AI.

DXO tries to enhance edges, which results in apparent, but nonexistent details. Topaz has the same problem.

In some special cases iWE is capable to extract details totally hidden by noise, when DXO Deep Prime and Topaz fail (If you are interested then I can show you such examples).

IWE, of course, is not yet perfect, but it is very young.


#1. DXO DeepPrime . Feather artifacts are well visible.


#2. Iwe-processed with just predefined filters. No artifacts well visible in #1.


#3. Topaz Denoise AI. Strange AI-artifacts are well visible.


#4, iWE processed. No artifacts as in #3 and more natural details compared to #3

 

[James Stirling]

Replacement for DXO seems misleading.

It may not have as simple an interface as DXO but it gives very good results for those prepared to do a little more work.

 

[Searching]

You can replace whatever you want, but you can't convince the public that it's replacement for a world class software like DXO.

Give it up with your misinformation. Typical.

 

[Serguei Palto]

You can replace whatever you want, but you can't convince the public that it's replacement for a world class software like DXO.

Give it up with your misinformation. Typical.

To judge whether I'm giving false information and whether iWE can replace DXO, you need to first carefully study my examples (I showed you obvious examples where iWE works better than DXO, since it does not generate specific AI artifacts). Secondly, you need to study iWE before judging its capabilities. You didn't do the first or the second. However, without understanding (essentially, not owning the subject), you are trying to condemn me for false information.

I am not convinced that I will be able to convince you of anything. However, I believe that you are not the only one reading this post. And for those who thoughtful and have eyes, I will continue.

If the person reading this post does not belong to the category of people who are technical or working in science, then this paragraph can be skipped by going to thefollowing paragraph, which gives another example of a comparison between DXO and iWE. Well, for those who continue to read this paragraph, I want to give a brief information about the basics that are laid down in iWE. At the heart of iWE I have tried to put my more than 40 years of experience in solid state physics, optics and Fourier spectroscopy, where I have constantly had to solve the problems of filtering signals in order to improve the signal-to-noise ratio. iWE (image waves editor) uses spectral or wave methods. These differ from the spectral methods traditionally used in image processing, which are based on wavelets. In iWE, for example, the image ifrom 20 Mpx sensor is converted into 10 million waves, each of which has its own amplitude, spatial frequency and phase. Each of these waves includes two contributions: a useful signal and noise. The noise wave has its own amplitude and phase. iWE's algorithms allow to determine not only the proportion of noise in each of the waves based on the spectral properties of photon noise, but also to judge the phase shifts between the signal and noise components. This allows, for example, to increase the sharpness by amplifying the high-frequency components taking into account the proportion of noise that is present in them. If, for example, the algorithm detects that the spectral component is exclusively noise, then it is simply removed from the spectrum. All this allows simultaneously increase sharpness and, at the same time, suppress noise. After the wave components are edited in the spectral domain, the waves are transformed into the ordinary space, where the value of each pixel is formed anew as a result of interference of all remaining waves. Below, I give an example of how it works, allowing to suppress noise while increasing sharpness.

And so, let's look at another example where let's compare the work of DXO DeepPrime and iWE. Now as a test image, let's take an image at low ISO 200. The image #1 and, for ease of comparison, its crop #2 show the standard processing result without applying any filters related to sharpening or noise reduction. This result is obtained in iWE using the standard AHD debayerization algorithm, recorded in the raw file of the camera's color matrix and using the standard (sRGB) gamma curve. This standard image can be considered as a reference image for the subsequent evaluation of both DXO DeepPrime and iWE spectral filters.

The #3, 4, respectively, full-frame and crop images of the same raw file processed by DXO DeepPrime. The Lens sharpness option is ON in DXO to get the maximum effect of sharpening and detailing. If you compare image #3 and #1, then undoubtedly image #3 looks much more attractive than image #1, where there is noise, and detail, at least at first glance, may seem below.

Let's ask a question. Can iWE make something similar to image #3 from image #1? The answer to this question is illustrated by image #5,6. As you can see, image #5,6 is in no way inferior in sharpness and detail to DXO image #3,4 – iWE did a very good job of both sharpening and suppressing noise. Moreover, with careful examination, you can find that image #5 contains more fine detail, and elements subject to sharpening look more natural than the DXO version. The sensation of artificial and excessive underlining of contours, as is the case with DXO, does not occur.

An even greater positive difference between iWE and DXO is evident in image #7,8, where iWE uses the wave debayerization algorithm I've already described (see the reference in my original post). The higher detail level in iWE image is obvious. For my taste, in this case, sharpening is even excessive, which just demonstrates the potential of iWE sharpening/denoise algorithms.

IWE is a very young program, it is not yet ideal for the average user, because it requires knowledge. But in the hands of a person who understands, it provides an excellent result and, again, very easily competes in terms of processing quality with DXO and others.




#1. Standard raw output (no denoise, no sharpening)




#2. Crop of #1.




#3. DXO DeepPrime processed.




#4. Crop of #3.




#5. iWE output. The same iWE preprocessing as in #1 plus iWE spectral denoise and sharpening filters




#6. Crop of #5




#7. iWE wave demosaicing plus wave denoise and sharpening




#8. Crop of #7

 

[MichaelAwkward]

You can replace whatever you want, but you can't convince the public that it's replacement for a world class software like DXO.

Give it up with your misinformation. Typical.

To judge whether I'm giving false information and whether iWE can replace DXO, you need to first carefully study my examples (I showed you obvious examples where iWE works better than DXO, since it does not generate specific AI artifacts). Secondly, you need to study iWE before judging its capabilities. You didn't do the first or the second. However, without understanding (essentially, not owning the subject), you are trying to condemn me for false information.

I am not convinced that I will be able to convince you of anything. However, I believe that you are not the only one reading this post. And for those who thoughtful and have eyes, I will continue.

If the person reading this post does not belong to the category of people who are technical or working in science, then this paragraph can be skipped by going to thefollowing paragraph, which gives another example of a comparison between DXO and iWE. Well, for those who continue to read this paragraph, I want to give a brief information about the basics that are laid down in iWE. At the heart of iWE I have tried to put my more than 40 years of experience in solid state physics, optics and Fourier spectroscopy, where I have constantly had to solve the problems of filtering signals in order to improve the signal-to-noise ratio. iWE (image waves editor) uses spectral or wave methods. These differ from the spectral methods traditionally used in image processing, which are based on wavelets. In iWE, for example, the image ifrom 20 Mpx sensor is converted into 10 million waves, each of which has its own amplitude, spatial frequency and phase. Each of these waves includes two contributions: a useful signal and noise. The noise wave has its own amplitude and phase. iWE's algorithms allow to determine not only the proportion of noise in each of the waves based on the spectral properties of photon noise, but also to judge the phase shifts between the signal and noise components. This allows, for example, to increase the sharpness by amplifying the high-frequency components taking into account the proportion of noise that is present in them. If, for example, the algorithm detects that the spectral component is exclusively noise, then it is simply removed from the spectrum. All this allows simultaneously increase sharpness and, at the same time, suppress noise. After the wave components are edited in the spectral domain, the waves are transformed into the ordinary space, where the value of each pixel is formed anew as a result of interference of all remaining waves. Below, I give an example of how it works, allowing to suppress noise while increasing sharpness.

And so, let's look at another example where let's compare the work of DXO DeepPrime and iWE. Now as a test image, let's take an image at low ISO 200. The image #1 and, for ease of comparison, its crop #2 show the standard processing result without applying any filters related to sharpening or noise reduction. This result is obtained in iWE using the standard AHD debayerization algorithm, recorded in the raw file of the camera's color matrix and using the standard (sRGB) gamma curve. This standard image can be considered as a reference image for the subsequent evaluation of both DXO DeepPrime and iWE spectral filters.

The #3, 4, respectively, full-frame and crop images of the same raw file processed by DXO DeepPrime. The Lens sharpness option is ON in DXO to get the maximum effect of sharpening and detailing. If you compare image #3 and #1, then undoubtedly image #3 looks much more attractive than image #1, where there is noise, and detail, at least at first glance, may seem below.

Let's ask a question. Can iWE make something similar to image #3 from image #1? The answer to this question is illustrated by image #5,6. As you can see, image #5,6 is in no way inferior in sharpness and detail to DXO image #3,4 – iWE did a very good job of both sharpening and suppressing noise. Moreover, with careful examination, you can find that image #5 contains more fine detail, and elements subject to sharpening look more natural than the DXO version. The sensation of artificial and excessive underlining of contours, as is the case with DXO, does not occur.

An even greater positive difference between iWE and DXO is evident in image #7,8, where iWE uses the wave debayerization algorithm I've already described (see the reference in my original post). The higher detail level in iWE image is obvious. For my taste, in this case, sharpening is even excessive, which just demonstrates the potential of iWE sharpening/denoise algorithms.

IWE is a very young program, it is not yet ideal for the average user, because it requires knowledge. But in the hands of a person who understands, it provides an excellent result and, again, very easily competes in terms of processing quality with DXO and others.


#1. Standard raw output (no denoise, no sharpening)


#2. Crop of #1.


#3. DXO DeepPrime processed.


#4. Crop of #3.


#5. iWE output. The same iWE preprocessing as in #1 plus iWE spectral denoise and sharpening filters


#6. Crop of #5


#7. iWE wave demosaicing plus wave denoise and sharpening


#8. Crop of #

This is very interesting, and thank you for posting your work. Are you able to provide a settings file for the iWE image that we can load into the program? I tried all of the 1-9 profiles and tweaked the settings, but nothing compares to DXO.

So to conclude, I can get results close to DXO using iWE interpolation but I have to use Topaz for denoise.

Mike

 

[MichaelAwkward]

I made a OM1 color profile for iWE so we have better color than the programs default matrix.

ISO 1250 image 100% crops.

iWE (Interpolation & Sharpening) & Topaz (Denoising)

DXO Photolab 5 with DeepPRIME & Lens Sharpness

Mike

 

[Serguei Palto]

You can replace whatever you want, but you can't convince the public that it's replacement for a world class software like DXO.

Give it up with your misinformation. Typical.

To judge whether I'm giving false information and whether iWE can replace DXO, you need to first carefully study my examples (I showed you obvious examples where iWE works better than DXO, since it does not generate specific AI artifacts). Secondly, you need to study iWE before judging its capabilities. You didn't do the first or the second. However, without understanding (essentially, not owning the subject), you are trying to condemn me for false information.

I am not convinced that I will be able to convince you of anything. However, I believe that you are not the only one reading this post. And for those who thoughtful and have eyes, I will continue.

If the person reading this post does not belong to the category of people who are technical or working in science, then this paragraph can be skipped by going to thefollowing paragraph, which gives another example of a comparison between DXO and iWE. Well, for those who continue to read this paragraph, I want to give a brief information about the basics that are laid down in iWE. At the heart of iWE I have tried to put my more than 40 years of experience in solid state physics, optics and Fourier spectroscopy, where I have constantly had to solve the problems of filtering signals in order to improve the signal-to-noise ratio. iWE (image waves editor) uses spectral or wave methods. These differ from the spectral methods traditionally used in image processing, which are based on wavelets. In iWE, for example, the image ifrom 20 Mpx sensor is converted into 10 million waves, each of which has its own amplitude, spatial frequency and phase. Each of these waves includes two contributions: a useful signal and noise. The noise wave has its own amplitude and phase. iWE's algorithms allow to determine not only the proportion of noise in each of the waves based on the spectral properties of photon noise, but also to judge the phase shifts between the signal and noise components. This allows, for example, to increase the sharpness by amplifying the high-frequency components taking into account the proportion of noise that is present in them. If, for example, the algorithm detects that the spectral component is exclusively noise, then it is simply removed from the spectrum. All this allows simultaneously increase sharpness and, at the same time, suppress noise. After the wave components are edited in the spectral domain, the waves are transformed into the ordinary space, where the value of each pixel is formed anew as a result of interference of all remaining waves. Below, I give an example of how it works, allowing to suppress noise while increasing sharpness.

And so, let's look at another example where let's compare the work of DXO DeepPrime and iWE. Now as a test image, let's take an image at low ISO 200. The image #1 and, for ease of comparison, its crop #2 show the standard processing result without applying any filters related to sharpening or noise reduction. This result is obtained in iWE using the standard AHD debayerization algorithm, recorded in the raw file of the camera's color matrix and using the standard (sRGB) gamma curve. This standard image can be considered as a reference image for the subsequent evaluation of both DXO DeepPrime and iWE spectral filters.

The #3, 4, respectively, full-frame and crop images of the same raw file processed by DXO DeepPrime. The Lens sharpness option is ON in DXO to get the maximum effect of sharpening and detailing. If you compare image #3 and #1, then undoubtedly image #3 looks much more attractive than image #1, where there is noise, and detail, at least at first glance, may seem below.

Let's ask a question. Can iWE make something similar to image #3 from image #1? The answer to this question is illustrated by image #5,6. As you can see, image #5,6 is in no way inferior in sharpness and detail to DXO image #3,4 – iWE did a very good job of both sharpening and suppressing noise. Moreover, with careful examination, you can find that image #5 contains more fine detail, and elements subject to sharpening look more natural than the DXO version. The sensation of artificial and excessive underlining of contours, as is the case with DXO, does not occur.

An even greater positive difference between iWE and DXO is evident in image #7,8, where iWE uses the wave debayerization algorithm I've already described (see the reference in my original post). The higher detail level in iWE image is obvious. For my taste, in this case, sharpening is even excessive, which just demonstrates the potential of iWE sharpening/denoise algorithms.

IWE is a very young program, it is not yet ideal for the average user, because it requires knowledge. But in the hands of a person who understands, it provides an excellent result and, again, very easily competes in terms of processing quality with DXO and others.


#1. Standard raw output (no denoise, no sharpening)


#2. Crop of #1.


#3. DXO DeepPrime processed.


#4. Crop of #3.


#5. iWE output. The same iWE preprocessing as in #1 plus iWE spectral denoise and sharpening filters


#6. Crop of #5


#7. iWE wave demosaicing plus wave denoise and sharpening


#8. Crop of #

This is very interesting, and thank you for posting your work. Are you able to provide a settings file for the iWE image that we can load into the program? I tried all of the 1-9 profiles and tweaked the settings, but nothing compares to DXO.

So to conclude, I can get results close to DXO using iWE interpolation but I have to use Topaz for denoise.

Mike

I'm running a newer, but not yet sufficiently tested, version of iWE with new functionality. Therefore, the version that you have installed will not be able to read the profile file that my new version generates. It is possible that this new version will already be available soon at the link you know. It will have a larger number accordingly.

So I'll try to describe the steps you should do in your version of iWE if you want to replicate the result I showed with sparrow at low ISO. I think the information below will be useful for understanding the overall processing approach in iWE.

1. Load the RAW file in the standard way, so the file will undergo normal preprocessing, as was shown in image #1 of my latest post above.
2. Go to the "Detail" tab and turn on the preset sharpening filter #2.

At this stage, you will significantly increase the detail. But if you want even more detail, as was required in my example, then go to the "Spectral S/D" tab and enable the "Spectral Sharpening" option. Thus, you add another spectral sharpening filter, which you can also adjust to your taste, without yet paying attention to the increased noise.

Since you have only turned on the sharpening filters, your noise level may rise noticeably. Therefore, to suppress noise, you need to turn on the appropriate filters.

1. For photos at relatively low ISO, the filter on the "Space S/D" tab, which is called "Deep Diff." is very useful. This filter is very effective for getting a nice background. If the initial background noise is weak, then in most cases the default values work very well, but if you want to increase the effect of noise suppression, then you need to start setting this filter by increasing the value of the slider D2 and only then raise the value of D1.
2. Another universal and very important filter that I used in my examples together with "Deep.Diff." is spectral "Amplitude denoising". Access to this filter is available both in the main panel and in the quick options panel ("Spectral A/C-D" tab). If you work with images at low ISO, then the value of "Threshold" should be raised approximately to the average value or even more (the higher this value, the higher the priority we give detail), and then you need to adjust the value "Strength".

The described filters are usually enough to get a pleasant background and a high degree of detail. It should be remembered that pre-installed filters are only a kind of basis that can be modified by adding the others. In case of rather high ISO “Dual scan” can be useful to get more pleasant background. Also in more complicated cases the spectral “Profiled denoising” can be used.

An even greater degree of detail on low-ISO files can be obtained by downloading raw data without LibRaw preprocessing. To do this, you must enable "Enable Wave Demosaicing" on the "RAW" tab before loading the data. The downloaded image will look black and white if the "Smart Box" is turned off. In this case, for the correct color reproduction on the same tab, it is important to choose the right type of Bayer matrix from 4 possible.

 

[MichaelAwkward]

You can replace whatever you want, but you can't convince the public that it's replacement for a world class software like DXO.

Give it up with your misinformation. Typical.

To judge whether I'm giving false information and whether iWE can replace DXO, you need to first carefully study my examples (I showed you obvious examples where iWE works better than DXO, since it does not generate specific AI artifacts). Secondly, you need to study iWE before judging its capabilities. You didn't do the first or the second. However, without understanding (essentially, not owning the subject), you are trying to condemn me for false information.

I am not convinced that I will be able to convince you of anything. However, I believe that you are not the only one reading this post. And for those who thoughtful and have eyes, I will continue.

If the person reading this post does not belong to the category of people who are technical or working in science, then this paragraph can be skipped by going to thefollowing paragraph, which gives another example of a comparison between DXO and iWE. Well, for those who continue to read this paragraph, I want to give a brief information about the basics that are laid down in iWE. At the heart of iWE I have tried to put my more than 40 years of experience in solid state physics, optics and Fourier spectroscopy, where I have constantly had to solve the problems of filtering signals in order to improve the signal-to-noise ratio. iWE (image waves editor) uses spectral or wave methods. These differ from the spectral methods traditionally used in image processing, which are based on wavelets. In iWE, for example, the image ifrom 20 Mpx sensor is converted into 10 million waves, each of which has its own amplitude, spatial frequency and phase. Each of these waves includes two contributions: a useful signal and noise. The noise wave has its own amplitude and phase. iWE's algorithms allow to determine not only the proportion of noise in each of the waves based on the spectral properties of photon noise, but also to judge the phase shifts between the signal and noise components. This allows, for example, to increase the sharpness by amplifying the high-frequency components taking into account the proportion of noise that is present in them. If, for example, the algorithm detects that the spectral component is exclusively noise, then it is simply removed from the spectrum. All this allows simultaneously increase sharpness and, at the same time, suppress noise. After the wave components are edited in the spectral domain, the waves are transformed into the ordinary space, where the value of each pixel is formed anew as a result of interference of all remaining waves. Below, I give an example of how it works, allowing to suppress noise while increasing sharpness.

And so, let's look at another example where let's compare the work of DXO DeepPrime and iWE. Now as a test image, let's take an image at low ISO 200. The image #1 and, for ease of comparison, its crop #2 show the standard processing result without applying any filters related to sharpening or noise reduction. This result is obtained in iWE using the standard AHD debayerization algorithm, recorded in the raw file of the camera's color matrix and using the standard (sRGB) gamma curve. This standard image can be considered as a reference image for the subsequent evaluation of both DXO DeepPrime and iWE spectral filters.

The #3, 4, respectively, full-frame and crop images of the same raw file processed by DXO DeepPrime. The Lens sharpness option is ON in DXO to get the maximum effect of sharpening and detailing. If you compare image #3 and #1, then undoubtedly image #3 looks much more attractive than image #1, where there is noise, and detail, at least at first glance, may seem below.

Let's ask a question. Can iWE make something similar to image #3 from image #1? The answer to this question is illustrated by image #5,6. As you can see, image #5,6 is in no way inferior in sharpness and detail to DXO image #3,4 – iWE did a very good job of both sharpening and suppressing noise. Moreover, with careful examination, you can find that image #5 contains more fine detail, and elements subject to sharpening look more natural than the DXO version. The sensation of artificial and excessive underlining of contours, as is the case with DXO, does not occur.

An even greater positive difference between iWE and DXO is evident in image #7,8, where iWE uses the wave debayerization algorithm I've already described (see the reference in my original post). The higher detail level in iWE image is obvious. For my taste, in this case, sharpening is even excessive, which just demonstrates the potential of iWE sharpening/denoise algorithms.

IWE is a very young program, it is not yet ideal for the average user, because it requires knowledge. But in the hands of a person who understands, it provides an excellent result and, again, very easily competes in terms of processing quality with DXO and others.


#1. Standard raw output (no denoise, no sharpening)


#2. Crop of #1.


#3. DXO DeepPrime processed.


#4. Crop of #3.


#5. iWE output. The same iWE preprocessing as in #1 plus iWE spectral denoise and sharpening filters


#6. Crop of #5


#7. iWE wave demosaicing plus wave denoise and sharpening


#8. Crop of #

This is very interesting, and thank you for posting your work. Are you able to provide a settings file for the iWE image that we can load into the program? I tried all of the 1-9 profiles and tweaked the settings, but nothing compares to DXO.

So to conclude, I can get results close to DXO using iWE interpolation but I have to use Topaz for denoise.

Mike

I'm running a newer, but not yet sufficiently tested, version of iWE with new functionality. Therefore, the version that you have installed will not be able to read the profile file that my new version generates. It is possible that this new version will already be available soon at the link you know. It will have a larger number accordingly.

So I'll try to describe the steps you should do in your version of iWE if you want to replicate the result I showed with sparrow at low ISO. I think the information below will be useful for understanding the overall processing approach in iWE.

1. Load the RAW file in the standard way, so the file will undergo normal preprocessing, as was shown in image #1 of my latest post above.
2. Go to the "Detail" tab and turn on the preset sharpening filter #2.

At this stage, you will significantly increase the detail. But if you want even more detail, as was required in my example, then go to the "Spectral S/D" tab and enable the "Spectral Sharpening" option. Thus, you add another spectral sharpening filter, which you can also adjust to your taste, without yet paying attention to the increased noise.

Since you have only turned on the sharpening filters, your noise level may rise noticeably. Therefore, to suppress noise, you need to turn on the appropriate filters.

1. For photos at relatively low ISO, the filter on the "Space S/D" tab, which is called "Deep Diff." is very useful. This filter is very effective for getting a nice background. If the initial background noise is weak, then in most cases the default values work very well, but if you want to increase the effect of noise suppression, then you need to start setting this filter by increasing the value of the slider D2 and only then raise the value of D1.
2. Another universal and very important filter that I used in my examples together with "Deep.Diff." is spectral "Amplitude denoising". Access to this filter is available both in the main panel and in the quick options panel ("Spectral A/C-D" tab). If you work with images at low ISO, then the value of "Threshold" should be raised approximately to the average value or even more (the higher this value, the higher the priority we give detail), and then you need to adjust the value "Strength".

The described filters are usually enough to get a pleasant background and a high degree of detail. It should be remembered that pre-installed filters are only a kind of basis that can be modified by adding the others. In case of rather high ISO “Dual scan” can be useful to get more pleasant background. Also in more complicated cases the spectral “Profiled denoising” can be used.

An even greater degree of detail on low-ISO files can be obtained by downloading raw data without LibRaw preprocessing. To do this, you must enable "Enable Wave Demosaicing" on the "RAW" tab before loading the data. The downloaded image will look black and white if the "Smart Box" is turned off. In this case, for the correct color reproduction on the same tab, it is important to choose the right type of Bayer matrix from 4 possible.

Thank you, I am using iWE-Wave with low ISO images with good results. The issue is high ISO 1600+ images, whatever I try I cannot get a good result. Some guide on how to zone in on a the correct parameters would be welcome.

Could you PM me your email address please?

Mike.

 

[Serguei Palto]

You can replace whatever you want, but you can't convince the public that it's replacement for a world class software like DXO.

Give it up with your misinformation. Typical.

To judge whether I'm giving false information and whether iWE can replace DXO, you need to first carefully study my examples (I showed you obvious examples where iWE works better than DXO, since it does not generate specific AI artifacts). Secondly, you need to study iWE before judging its capabilities. You didn't do the first or the second. However, without understanding (essentially, not owning the subject), you are trying to condemn me for false information.

I am not convinced that I will be able to convince you of anything. However, I believe that you are not the only one reading this post. And for those who thoughtful and have eyes, I will continue.

If the person reading this post does not belong to the category of people who are technical or working in science, then this paragraph can be skipped by going to thefollowing paragraph, which gives another example of a comparison between DXO and iWE. Well, for those who continue to read this paragraph, I want to give a brief information about the basics that are laid down in iWE. At the heart of iWE I have tried to put my more than 40 years of experience in solid state physics, optics and Fourier spectroscopy, where I have constantly had to solve the problems of filtering signals in order to improve the signal-to-noise ratio. iWE (image waves editor) uses spectral or wave methods. These differ from the spectral methods traditionally used in image processing, which are based on wavelets. In iWE, for example, the image ifrom 20 Mpx sensor is converted into 10 million waves, each of which has its own amplitude, spatial frequency and phase. Each of these waves includes two contributions: a useful signal and noise. The noise wave has its own amplitude and phase. iWE's algorithms allow to determine not only the proportion of noise in each of the waves based on the spectral properties of photon noise, but also to judge the phase shifts between the signal and noise components. This allows, for example, to increase the sharpness by amplifying the high-frequency components taking into account the proportion of noise that is present in them. If, for example, the algorithm detects that the spectral component is exclusively noise, then it is simply removed from the spectrum. All this allows simultaneously increase sharpness and, at the same time, suppress noise. After the wave components are edited in the spectral domain, the waves are transformed into the ordinary space, where the value of each pixel is formed anew as a result of interference of all remaining waves. Below, I give an example of how it works, allowing to suppress noise while increasing sharpness.

And so, let's look at another example where let's compare the work of DXO DeepPrime and iWE. Now as a test image, let's take an image at low ISO 200. The image #1 and, for ease of comparison, its crop #2 show the standard processing result without applying any filters related to sharpening or noise reduction. This result is obtained in iWE using the standard AHD debayerization algorithm, recorded in the raw file of the camera's color matrix and using the standard (sRGB) gamma curve. This standard image can be considered as a reference image for the subsequent evaluation of both DXO DeepPrime and iWE spectral filters.

The #3, 4, respectively, full-frame and crop images of the same raw file processed by DXO DeepPrime. The Lens sharpness option is ON in DXO to get the maximum effect of sharpening and detailing. If you compare image #3 and #1, then undoubtedly image #3 looks much more attractive than image #1, where there is noise, and detail, at least at first glance, may seem below.

Let's ask a question. Can iWE make something similar to image #3 from image #1? The answer to this question is illustrated by image #5,6. As you can see, image #5,6 is in no way inferior in sharpness and detail to DXO image #3,4 – iWE did a very good job of both sharpening and suppressing noise. Moreover, with careful examination, you can find that image #5 contains more fine detail, and elements subject to sharpening look more natural than the DXO version. The sensation of artificial and excessive underlining of contours, as is the case with DXO, does not occur.

An even greater positive difference between iWE and DXO is evident in image #7,8, where iWE uses the wave debayerization algorithm I've already described (see the reference in my original post). The higher detail level in iWE image is obvious. For my taste, in this case, sharpening is even excessive, which just demonstrates the potential of iWE sharpening/denoise algorithms.

IWE is a very young program, it is not yet ideal for the average user, because it requires knowledge. But in the hands of a person who understands, it provides an excellent result and, again, very easily competes in terms of processing quality with DXO and others.


#1. Standard raw output (no denoise, no sharpening)


#2. Crop of #1.


#3. DXO DeepPrime processed.


#4. Crop of #3.


#5. iWE output. The same iWE preprocessing as in #1 plus iWE spectral denoise and sharpening filters


#6. Crop of #5


#7. iWE wave demosaicing plus wave denoise and sharpening


#8. Crop of #

This is very interesting, and thank you for posting your work. Are you able to provide a settings file for the iWE image that we can load into the program? I tried all of the 1-9 profiles and tweaked the settings, but nothing compares to DXO.

So to conclude, I can get results close to DXO using iWE interpolation but I have to use Topaz for denoise.

Mike

I'm running a newer, but not yet sufficiently tested, version of iWE with new functionality. Therefore, the version that you have installed will not be able to read the profile file that my new version generates. It is possible that this new version will already be available soon at the link you know. It will have a larger number accordingly.

So I'll try to describe the steps you should do in your version of iWE if you want to replicate the result I showed with sparrow at low ISO. I think the information below will be useful for understanding the overall processing approach in iWE.

1. Load the RAW file in the standard way, so the file will undergo normal preprocessing, as was shown in image #1 of my latest post above.
2. Go to the "Detail" tab and turn on the preset sharpening filter #2.

At this stage, you will significantly increase the detail. But if you want even more detail, as was required in my example, then go to the "Spectral S/D" tab and enable the "Spectral Sharpening" option. Thus, you add another spectral sharpening filter, which you can also adjust to your taste, without yet paying attention to the increased noise.

Since you have only turned on the sharpening filters, your noise level may rise noticeably. Therefore, to suppress noise, you need to turn on the appropriate filters.

1. For photos at relatively low ISO, the filter on the "Space S/D" tab, which is called "Deep Diff." is very useful. This filter is very effective for getting a nice background. If the initial background noise is weak, then in most cases the default values work very well, but if you want to increase the effect of noise suppression, then you need to start setting this filter by increasing the value of the slider D2 and only then raise the value of D1.
2. Another universal and very important filter that I used in my examples together with "Deep.Diff." is spectral "Amplitude denoising". Access to this filter is available both in the main panel and in the quick options panel ("Spectral A/C-D" tab). If you work with images at low ISO, then the value of "Threshold" should be raised approximately to the average value or even more (the higher this value, the higher the priority we give detail), and then you need to adjust the value "Strength".

The described filters are usually enough to get a pleasant background and a high degree of detail. It should be remembered that pre-installed filters are only a kind of basis that can be modified by adding the others. In case of rather high ISO “Dual scan” can be useful to get more pleasant background. Also in more complicated cases the spectral “Profiled denoising” can be used.

An even greater degree of detail on low-ISO files can be obtained by downloading raw data without LibRaw preprocessing. To do this, you must enable "Enable Wave Demosaicing" on the "RAW" tab before loading the data. The downloaded image will look black and white if the "Smart Box" is turned off. In this case, for the correct color reproduction on the same tab, it is important to choose the right type of Bayer matrix from 4 possible.

Thank you, I am using iWE-Wave with low ISO images with good results. The issue is high ISO 1600+ images, whatever I try I cannot get a good result. Some guide on how to zone in on a the correct parameters would be welcome.

Could you PM me your email address please?

Mike.

My E-mail is the manual (reference in OP). Also, in the manual one can get more info on the processing.