PHD and PHD2 will accept same dither size levels from various imaging programs. Each of 5 Levels of dither aggressiveness corresponds to a number of pixels as shown in the table below. Additionally, this number of pixels will get multiplied by the number specified in dither scale in PHD's brain settings.
Dither Level | Pixels
--------------- ---------
1 - +/- 0.5 x Dither Scale
2 - +/- 1.0 x Dither Scale
3 - +/- 2.0 x Dither Scale
4 - +/- 3.0 x Dither Scale
5 - +/- 5.0 x Dither Scale
Since dithering happens on the Guider and not on the Imager, we need to calculate the difference between both of the imaging scales. For that we simply divide our Guider Scale by the Imager scale. In my example it would be 0.85 / 0.44 = 2. Simply means, that if star moves 3 pixels on the Guider it will move 3 x 2 = 6 pixels on the Imager.
Let's say I decided that I want to dither 30 pixels on the Imager, so I need that on the Guider it will move for 30 / 2 = 15 pixels. Now, according to the table above, in my imaging software I would select Aggressiveness Level 5, that would correspond to 5 pixels and in PHD's Dither Scale I would put 3, so the actual amount of pixels on the Guider would be 5 x 3 = 15 pixels.
Choosing an Appropriate Dither Amount
The 30-pixel imaging-camera dither used in the example is intended only to demonstrate the calculation process and should not be interpreted as a universal recommendation.
The optimal dither amount depends on several factors, including camera type, image scale, seeing conditions, and the nature of the noise present in the data. Historically, larger dithers were often recommended for DSLR and one-shot color cameras to help randomize spatial artifacts introduced by the debayering process and reduce the risk of walking noise. Modern monochrome CMOS cameras typically require much smaller dithers.
As a general guideline, imaging-camera dithers of 5–15 pixels are sufficient for many modern systems, while larger values may be beneficial when dealing with stronger fixed-pattern noise, walking noise, or color-camera interpolation artifacts. The primary goal is to ensure that noise patterns fall on different pixels between exposures so they can be effectively rejected during stacking.