Lum sub solves, filter subs fail

Sitting at my desk experimenting with solving previously captured frames. I see a curious behavior when solving luminance frames versus filtered frames. With both PlateSolve2 and ANSVR, I get reliable solves with the luminance frames but mostly get fails on the red, green, blue frames – these are frames of the same target and they were all shot after centering so the center point hint is exactly the same in all.

In my subs of the Cocoon Nebula, the luminance focus and centering frames all solved quickly and accurately. The blue filter frame also solves but the red and green fail. All are good quality images. On another target, the luminance and Ha frames solved but the RGB frames all failed.

All these images solve with Since I always frame / focus / center using the luminance filter, these failures aren’t causing a problem. I am just curious if any other SGP users see the same results?