That. Dithering is useful if you can trade samples for accuracy. It is useful if your quantization scale is too coarse. As a result, one can inject noise to s

That. Dithering is useful if you can trade samples for accuracy. It is useful if your quantization scale is too coarse. As a result, one can inject noise to sense signal that is normally below a sensor threshold, for instance. In graphics, it is still very useful to add some dithering in some cases; one obvious example being to reduce banding on gradients: https://en.wikipedia.org/wiki/Colour_banding https://en.wikipedia.org/wiki/Colour_banding Regardless of the chosen quantization scale, there are corner cases where values measured will fall close one to another; dithering is useful in these cases, and is not limited to images, but any data: https://en.wikipedia.org/wiki/Audio_bit_depth#Dither https://en.wikipedia.org/wiki/Audio_bit_depth#Dither