Noise Ninja includes an automatic profile loader that selects the "best" noise profile when you open an image. It depends on profile annotations. When it is set up correctly, it works almost transparently. However, it is recommended that you read through this entire section to be aware of some of the requirements and idiosyncrasies of the auto loader.
The auto-profiler can be enabled or disabled via an option on the Profiles page of the Preferences dialog.
In a nutshell, here's how the automatic loader works:
During the matching process, the Maker, Model, ISO, Exposure Correction, and Megapixels attributes normally carry the most weight. For film scans, the Film and Multisampling fields also carry high weight. Sharpness, saturation, and contrast have medium influence. The remaining fields normally have low influence, though it depends on the specific profile set and the magnitude of differences between values. For more information, see the discussion about "How many profiles do I need?".
Matching will fail if the Maker, Model, or ISO are undefined either in the image attributes (after default substitution) or in the profile annotations. Other attributes or annotations are optional, though missing values may cause a profile to earn a lower ranking during matching.
Once it is set up correctly, the automatic profiler can be very effective and convenient. To ensure that it works as expected, you should adhere to the following guidelines:
In general, try not to make the matching process too difficult for the auto-loader. It has to optimize within a 15-dimensional space, and that is not always an easy thing to do. The auto-loader tries to make intelligent choices when there is not an obvious match, but this can lead to some non-intuitive results. For instance, if you only have profiles for 6 megapixel images, and you use the auto loader for a 3 megapixel ISO 1600 image, the auto loader might pick an ISO 800 profile. This is because the downsampling process used to create a 3-megapixel image from a 6-megapixel sensor also reduces noise levels. There are a number of relationships like this that the auto-loader attempts to consider. However, these relationships are most reliable when they do not involve large differences in parameter values.
So, to increase the reliability of the matching process, and to make it less likely that you will be confused by the automatic selections, try to use or create profile sets that are already well matched to your shooting preferences.
It is recommended that you work with full-size images instead of crops. This will avoid potential ambiguity in the profile matching process.
By default, Noise Ninja assumes that the dimensions of the image denote the full resolution, not a crop. It attempts to match this value against the Megapixels annotations of existing profiles. For instance, if you have profiles for both 3 megapixel images and 6 megapixel images from the same camera, then Noise Ninja will match a 3 megapixel image against the 3 megapixel profiles.
Downsampling reduces noise levels, but cropping does not change noise characteristics. So, if you try to filter a 3 megapixel crop from a 6 megapixel original image, Noise Ninja will match the wrong set of profiles.
If you work with reduced-resolution images, it is best to create profiles specifically for that resolution. Otherwise, Noise Ninja will select a higher-resolution profile, but it will try to select a lower ISO to compensate for the reduced noise in the downsampled image. For instance, if you have profiles for 6 megapixel images and then you open a 3 megapixel ISO 400 image, Noise Ninja might select a 6 megapixel, ISO 200 profile. This is probably a better choice than the 6-megapixel ISO 400 profile, but it might not be as good as a profile created specifically for 3-megapixel ISO 400 images.
Distinguishing RAW from JPEG images: Each image has a Quality field that is set based on the file extension for the image. If the file extension is ".jpg" or ".jpeg", the Quality field is set to "JPEG". Otherwise it is set to "Tiff/RAW". The automatic profile loader interprets an image quality of "JPEG" to be the same as the "Fine JPEG" when it matches against the Quality annotation in a noise profile. So, if you shoot both RAW and JPEG images, you can create one set of profiles with a Quality annotation of "Tiff/RAW" and another set of profiles with a quality annotation of "Fine JPEG", and the automatic profile loader will use the Quality field of the image to select the appropriate profile.
Exposure Correction: The Exposure Correction annotation is currently specific to Adobe Capture Raw. It is extracted from XMP data that Capture Raw embeds in converted images. It corresponds to the amount of post-capture exposure adjustment. Since the adjustment is occuring after exposure, it can change the relationship between noise levels and tones. By default the auto loader treats Exposure Correction like an ISO boost or cut, and it will try to select a higher or lower ISO profile to compensate. If you feel that this is not realistic for your particular camera, you can create camera profiles that cover different combinations of ISO and Exposure Correction, and the auto loader will try to pick the combination with the closest match to the corresponding image attributes. (Note that in-camera "exposure compensation" changes the exposure, it does not change the relationship between noise levels and tones. So, it isn't necessary to profile for this separately, and there is not a separate profile annotation for it.)
Regional variations in EXIF data: Realize that the Model field for some cameras varies in different regions of the world. For instance, if you are using a camera bought in Asia with profiles created using the American version of the same camera, you might need to change the Model annotations in the profiles and re-save them (with the automatic directory naming function enabled).
Sony camera owners: Several Sony cameras use the same name -- "Cybershot" -- for the EXIF Model field. If you use one of these cameras, then you will need to go to the NoiseProfiles/Sony/Cybershot/ directory and move the profiles for other models to another directory.
Automatic profile naming: In the Preferences dialog there is an option to automatically name profiles, using a mnemonic scheme derived from the annotation values. This feature is not necessary to use the auto loader. The auto loader ignores the actual profile names. Instead, it looks at the annotations that are stored inside the profile. The automatic naming scheme may make it easier to distinguish different sets of profiles for the same camera. However, if you prefer another naming scheme, feel free to use it.
Proprietary or missing ISO data: Some cameras store ISO and other information in a proprietary, undocumented EXIF field called the MakerNote. Noise Ninja can read some MakerNote fields, but not for all cameras. Usually, it can find enough information to make a match. However, two things in particular can interfere. First, ISO information is sometimes removed by other software applications when it is stored in the MakerNote field. Second, the "Auto ISO" mode available on some cameras might not store the actual ISO used for an image in a place where Noise Ninja can read it.