Remove Noise - Raw & non-Raw

The Remove Noise filter will detect and remove noise caused by:

• poor lighting
• high ISO
• long exposure time
• inconsistency in the sensor used for image capture

Models

Topaz Photo AI comes with five Remove Noise models, each trained for a different situation.

The Raw models are enabled when Topaz Photo AI detects a Raw file. They process Raw Bayer sensor data, which contains the more information and detail than pixels, to remove noise. Raw images which do not have Bayer sensor data will not use the Raw models.

For non-Raw images such as TIF, JPG, PNG, etc, the normal Remove Noise models are used. The non-Raw models process pixel data to remove noise.

The same image processed as Raw vs TIF is below. Original image (left), Raw result, TIF result (right).

Raw Normal

Enabled when an image with Raw data is detected. Processes the sensor color filter array data to detect and remove noise. Best for low-medium noise images.

Raw Strong

Enabled when an image with Raw data is detected. Processes the sensor color filter array data to detect and remove noise. Best for medium-high noise images.

Raw Normal v2

A beta model that will eventually replace Raw Normal. It retains more detail for a high quality denoise that does not smooth out the image.

Raw Strong v2

A beta model that will eventually replace Raw Strong. This model will retain more detail than the Raw Strong model while removing extreme noise at very high ISOs.

Normal

Enabled for non-Raw images. Processes pixel data to detect and remove noise. Best for low-medium noise images.

Strong

Enabled for non-Raw images. Processes pixel data to detect and remove noise. Best for medium-high noise images with better detail retention and controls.

Extreme

Enabled for non-Raw images. Processes pixel data to detect and remove noise. Best for medium-high noise images.

Controls

Enable/disable Remove Noise by clicking the toggle.

When turned on, Remove Noise displays a Strength, Detail, Recover Original Detail slider. Control the results with these options.

• Strength: controls the amount of noise removed from the image. Increasing the Strength slider can remove fine details.
• Minor Deblur: controls a minor sharpening on the entire image. Use this slider to recover details that are lost during noise reduction.
• Recover Original Detail: Return texture and fine details lost during noise reduction. Start at low values and slowly increase.

Exceptions For Raw Models

The Raw models require Bayer sensor data. If the Raw image does not have Bayer sensor data, the Raw models cannot be used.

Fujifilm cameras with X-Trans sensors are not compatible with the Raw models as they do not create Bayer sensor data. The Remove Noise - Linear Raw model will be used instead.

Raw files which are processed and are no longer in the Raw Bayer sensor format are not compatible with the Raw models. The Remove Noise - Linear Raw model will be used instead.

Alternatives For Noise Reduction

There are two alternatives if Remove Noise removes too much detail and low noise reduction is needed:

• Enable the Sharpen filter and increase the Clean slider
• Enable Enhance Resolution and increase the Suppress Noise slider

The Sharpen - Minor Denoise slider performs a mild noise reduction on the subject in the image.

The Enhance Resolution - Minor Denoise slider performs a minor noise reduction to the entire image.