The Affinity Photo procedural filter has a huge selection of presets but if you want to create your own basic equation from a blank sheet. To start go to the left of the middle and click the + to add an equation. It will be added as a red channel by default and you will see the right side R set to ON. You can repeat this clicking the + and that will create an entry for the green and the blue channels. You can set all the channels ON as well by clicking the R G B buttons. You can enter a value between 0 and 1 in the field, if it is greater it makes no difference, if less it makes no difference. However, you can enter more than just 0.2, 0.3 etc in the field, you can enter functions such as tan and sin and cosine etc as well as variables such as rx (relative x / horizontal) and ry (relative y / vertical) which is more useful than just fixed x and y as you can then move the layer. You can also enter R (the red pixel value 0 to 1) and G and B. You can enter the same settings in each of the fields for the channels or different settings. For a very basic one, simply enter rx + ry into the red field and the same into the other channels and you will get an interesting diagonal gradient. You can also add custom inputs to the settings such as a * rx + b*ry and that does give you more options. You need to click one of the variable buttons / custom inputs along the bottom such as R or Z. I generally go for R which means you can enter real values such as 0.244 etc. You can set up different custom inputs for the entire set of channels such as you will have to click the custom input button 6 times to create a, b, c, d, e, f etc (you can change them to anything, they don't need to remain as a). You can then add the multipliers to all the rx and ry entries and set the a to 0.234 and b to 0.354 etc to create an interesting combination of gradients. You can also now save the current equations and values to a preset via the top right menu and create new preset etc. That is useful as you don't want to have to create it again and again every time you come to this filter.
rx+ry
rx*a + ry*b
sqrt(rx*a + ry*b)
oscsin(rx*a)
rx-ry
rx-ry*ry