Filtration-related questions

Q Why do I receive some crystal-like artifacts in the filtered image?

A The crystal-like artifacts (that are in fact introduced by JPEG compression) look like thin lines in the filtered image. They can be eliminated by increasing the value of the high frequency setting of Noise Filter.

Note: presence of many residual artifacts is usually a consequence of using a poorly built device noise profile or a profile built for another device and/or device mode.

Q Filtered image looks 'plastic'. Why?

A The reason is that too much filtration was applied. Let Neat Image keep some noise to have natural-looking results. Adjust the noise reduction amounts; for example, reduce the noise reduction amount in the luminance (Y) channel to 50-70%.

Also, make sure the device noise profile does match the image processed. Using an incorrectly chosen or poorly built profile can both produce plastic-looking results and leave residual artifacts (see the previous question).

Q What is frequency?

A The term (spatial) frequency is used in Neat Image to denote image elements (both important details and noise) of specific size.

High frequency corresponds to image elements of smallest size. Medium frequency corresponds to elements of medium size. Low frequency corresponds to image elements of (relatively) large size.

For reference, see the noise samples of different frequencies in the Noise Filter Settings tab of the Filtration Job Editor.

Q The filtration process is slow, is this normal?

A This is normal, because the filtration algorithm is quite complex. We are working on further optimization to provide better performance.

Q Is the input image automatically changed according to the profile that is being opened?

A The input image is NEVER changed. Neat Image always creates a new output image, which contains the filtration results. The output image can then be saved to any file.

Q How to filter only the color noise (not the brightness noise)?

A When the YCrCb space is used, set the value of the luminance (Y) channel noise reduction amount to 0%. This will disable filtration in the luminance (brightness) channel.

Q What is YCrCb?

A YCrCb is the name of a family of color spaces widely used in digital imaging, television, image compression (e.g., the JPEG compression transform RGB images into this space to efficiently compress image data), etc. In 'YCrCb', 'Y' corresponds to the luminance channel, 'Cr' - to the Cr chrominance channel covering the red to blue-green color range, 'Cb' - to the Cb chrominance channel covering the blue to yellow color range. Because this space enables easy separation of the luminance and chrominance information, it is very suitable to conduct noise reduction.

Q Is processing via Neat Image best done before or after any other processing (i.e. tonal/color correction)?

A Such operations as tonal/color correction are quite conservative from the standpoint of noise, i.e., they do not significantly change the noise characteristics of the image. Therefore, filtering before or after makes little difference - as long as the noise profile is built and applied at the same stage of image processing. For example, don't use a device noise profile built with an unprocessed (with the color correction not yet applied) image to filter a processed image.

Some digital cameras apply some color correction internally. Other cameras allow access to unprocessed RAW data. Neat Image is a generic filter, which can be applied in both cases. The only requirement is to use profile that matches the device mode of the input image.

On the other hand, image sharpening applied to a noisy image makes it much noisier. It is best to apply Neat Image filtration before sharpening. However, the sharpening and noise filters of Neat Image can be used together because the sharpening is applied AFTER noise filtration.