Odd radius display on sunset time, 2022-10-16 (Jouhe, France)

Often, when I see the clouds coming from west, I try to do a high cloud sunset to see if there is something to see (like sun pillar and sunvex halos).

this day, we had mosly contrails making here and there some ice halos from my position, but nothing unusual to catch the eye (furtive sundogs, 22° halos, upper tangent and maybe Lowitz buddies but too short in time to be exploitable) (Time-Lapse of part of the day in here, South-South-East Field of View). To find other halos at sunset, the exercice can be tricky, as the global backgroud turn progressivly to yellow then orange so the b-r rendering is more complicated, with adaptation to make in the last 30-45 mins before the end of sunset.
As I don't have automatized this, I find myself resigned to make one setting for the whole sequence to save time in my processings.
And sometimes, something catch my eye, as it did this last 16th of october, last year.
The three next images show 3 differents times of the event, with slightly different processings:

2022-10-16-[17h14à18h58]-NNW-bw-v1-[4-2]-UsM-(49)-5s5-histo-UsM-h20-h28?-h35

2022-10-16-[17h14à18h58]-NNW-bw-v2-[4-2]-UsM-(47)-7s7-UsM-histo

2022-10-16-[17h14à18h58]-NNW-bw-v2-[4-2]-UsM(56)-6s6-UsM-h28?-histo


If needed, I can provide the bw sequence, so you can see (as I do with VirtuaDub) the evolution of the display and distinguish the 28° halo with dynamic (back and forth) play on the sequence.

A sequence you will be able to see there (Version1 & Version2), but with the compression due to GoogleVideo. Again, all raw images are still in my possession, as I guess I might not be done with this display yet.