With this collimation, I've been consistently seeing this on-axis:
Not quite right on-axis. It's out of focus on the 9 o'clock/3 o'clock axis. And consistently a bit more corner aberration maybe than an APSC sensor has a right to have as well.
So I tested with putting an offset on the Cheshire center dot. Relative to the Vixen Rail at the bottom of the OTA, and looking into the Cheshire, I decided to butt the Cheshire dot up against the secondary donut at 6 o'clock, 9 o'clock, 12 o'clock, and 3 o'clock.
First I did 6 o'clock and checked the tri-bahtinov result, same not quite right tri-bahtinov image. Then 12 o'clock. Still no luck. Then 9 o'clock. Same problem. Then I tried offsetting the Cheshire center dot up against the secondary donut at 3 o'clock.
and I saw this:
Nice! It's a dimmer star than the earlier image, so not as bright. On-axis tri-bahtinov mask is right down the middle, or nearly enough so. It's quite a bit better than what I've been seeing.
So I tweaked the Hall of Mirrors since moving the secondary created a small but perceptible change in the HOM.
And just as it was getting into morning twilight, I got this image in Cassiopeia.
This, finally, is what I want (or very nearly so) out of my RC6 with an APSC sensor. Obviously, a full frame sensor would show field curvature, and to use the full frame, a flattener would be needed, but for a crop sensor, this is as good as I could hope for. No, it's not perfect and it's not going to be. But this is so much better than anything I've achieved previously. If you're seeing anything but round corner stars on APSC, it's not field curvature, it's mis-collimation. I'll have to duplicate it when it's not so close to sunrise and the image is less noisy (I pushed the black point to darken the sky). If any corner is worse than the others, it would be bottom right and it's not really all that bad. Still, you can tell the other three corners are round to "off-round", but the lower right corner stars are oval. The Cheshire dot offset was an arbitrary distance toward 3 o'clock on the secondary donut, and a perfectionist could continue to tweak it from there and perhaps beat the last bit of corner aberration out.
So, Hall of Mirrors is deadly accurate when pushed to the limit. The Cheshire centering is not (and now I'm wondering if that sloppy "find the optical center of the secondary" test I did might have been a better test than I gave it credit for?) However, if you're seeing more corner aberration than you should, or not quite perfect on-axis, just offset the Cheshire dot toward the secondary donut until you find the offset direction that gives a superior result to the others.
I've been staring at this so long, I may have myself convinced that I've gotten better results than I really have. Anybody care to put a fresh eye on this and let me know if I've upped my game, or am I laboring under a cloud of wishful thinking?
And if you've never considered a tri-bahtinov mask, I can't praise the item enough.
Edited by dg401, 08 August 2021 - 04:29 PM.