The right way to make profiles and use them?

[salarian]

Hi,

I have some questions about the better way to make and use profiles. I hope I can be clear in what exactly I need:

I've noticed that the center target in the calibration chart is actually a gray, with 127,127,127 RGB values. However, sometimes when you take a photo of the target and read the values in the photoshop, you can see that because of not having a perfect exposure (or the print quality or whatever) the centeral box is not the closest one to the 127,127,127 RGB but maybe another patch is a better choise. Now, wouldn't it be better to use the patch with the closest RGB value to 127,127,127 as the primary selection for profiling? Also, I see sometims that the darker patches show more visible noise. Is it better to manually fine tune the profile using those parts or right away after selecting the primary section, hit the auto optimize?

In practice, when you're going to use an already made profile, is it a better idea to use directly the matching profile on the image or to load the profile AND auto optimize (or even select some small parts and manually optimize) it for the each image for better results?

With RAW images, Neat Image at best is the SECOND program ever touching the data (sure, the first one is the RAW converter), so how important are the effects of WB adjustment on the overall output? Is it a good idea to load a ready made profile (with no specific WB adjustments) and just optimize it (auto or manual) and apply it to the image? And what about 8 and 16 bits outputs? Should one use a separate profile for 16 bits and 8 bits files?

And at the end, how really important is to have a top notch profile?!! It seems to be a strange question but actually I have once tested to see what happens if I use a WRONG profile (in my case a Sony DSC-75 profile for a Canon 10D image) and the results were pretty good!!! Or was it just a pure chance?

Best Regards
Arash

[NITeam]

I've noticed that the center target in the calibration chart is actually a gray, with 127,127,127 RGB values. However, sometimes when you take a photo of the target and read the values in the photoshop, you can see that because of not having a perfect exposure (or the print quality or whatever) the centeral box is not the closest one to the 127,127,127 RGB but maybe another patch is a better choise. Now, wouldn't it be better to use the patch with the closest RGB value to 127,127,127 as the primary selection for profiling?

No, it is not that important to have 127,127,127 in the primary selection. In fact nothing is important except what the user guide suggests.
Neat Image is designed to work in real world conditions where 127,127,127 is practically impossible to achieve. 127,127,127 is not needed either. You could start with any patch of the table.

Also, I see sometims that the darker patches show more visible noise. Is it better to manually fine tune the profile using those parts or right away after selecting the primary section, hit the auto optimize?

Auto fine-tuning in principle may be less accurate than manual fine-tuning (selecting areas, using manual fine-tuning analyzer). With the calibration target both auto and manual fine-tuning are practically the same accurate.

In practice, when you're going to use an already made profile, is it a better idea to use directly the matching profile on the image or to load the profile AND auto optimize (or even select some small parts and manually optimize) it for the each image for better results?

The second variant would produce more accurate result. Please see Profiling methods for some more details.

With RAW images, Neat Image at best is the SECOND program ever touching the data (sure, the first one is the RAW converter), so how important are the effects of WB adjustment on the overall output?

WB in general is not very important for noise reduction with Neat Image.

Is it a good idea to load a ready made profile (with no specific WB adjustments) and just optimize it (auto or manual) and apply it to the image?

Right.

And what about 8 and 16 bits outputs? Should one use a separate profile for 16 bits and 8 bits files?

No, the same profiles should be used.

And at the end, how really important is to have a top notch profile?!!

Just one thing - with an accurate profile, you usually do not need to adjust the filter settings, defaults are based on profile and produce good result in this case. With a less accurate profile, you may need to play with filter settings to compensate for that inaccuracy.

It seems to be a strange question but actually I have once tested to see what happens if I use a WRONG profile (in my case a Sony DSC-75 profile for a Canon 10D image) and the results were pretty good!!! Or was it just a pure chance?

This was luck. Generally it is very important to use accurate profile. Inaccurate profile may lead to incomplete filtration (artifacts, residual noise and compression artifacts) or to loss of fine details. Profile should be accurate to tell the filter exact noise levels to make the filter applying reduction to ALL noise elements and only noise elements (not image details).

Hope this helps.
Vlad