Ron Baird wrote:
...
What you should consider is that low light or higher ISO settings in digital
camera can yield an increase in 'noise' the digital equivalent of grain. You
may want to get a good filter for it, or in your case improve on some shots
with flash. The P20 can talk to your camera and would be an advantage.
...

I'm not an expert on this, but I hypothesize that the explanation
for Ron's comment is as follows:

A digital sensor accumulates charge (i.e., electrons) in each
pixel position as a response to light falling on the sensor in that
position.

The sensor also accumulates charge in each pixel simply as
a result of the random motion of electrons (due to heat,
radiation, and perhaps other factors other than stimulation
by light.)

We see digital "noise" when the ratio of random charge becomes
a significant percentage of the total charge in a pixel, causing
the value of that pixel to have a noticeable random component
to it.

The sensor accumulates random charge as a function of time.
The sensor is discharged just before the image is made, and
read after the image is made.

In bright light, the time between pre-image discharge and
post-image reading is short, allowing little random charge
to accumulate.

In dim light the time is long, allowing more random charge
to accumulate. Hence the extra noise in low light situations.

Dark areas of an image are most subject to this because
the random charge in all areas of the image is the same, but the
dark areas have a higher percentage of random to light
stimulated charge accumulation.

The "filtering" that Ron talks about is the digital process of
averaging adjacent pixels with similar values in order to
eliminate differences that are due to random factors. But
no filter, no matter how smart, can truly distinguish
differences due to random charge as distinct from differences
due to different amounts of light. It uses heuristic algorithms
to make intelligent guesses, which are often but not always
right. Hence there is always some loss of detail when
aggressive noise filtering is performed.

Alan