Page 1 of 1

Home-in is diverging.

Posted: Sat Sep 26, 2015 5:15 pm
by xiando
Hi,

While processing an image of the Bubble Nebula today, I noticed that when performing a Development task, the home-in function min/max is becoming divergent near the end point.

Normally, from the help entry description, I expect to see the home-in function converge on a single point.

In this case, it is homing in until it gets to (what appears to be) about ~5% or less to solution (jump to 50%, jump to 75%, jump to 85%) but then diverging rather than converging

ex: the ideal is trending towards 87.47% but the max/min values are 88.72% and 86.21% respectively), but then the max and min begin to diverge on each subsequent click of Home In

88.93, 85.99 > 88.97, 85.87 > etc

I can still find a "sweet spot" by doing the eyelid calculation, so no harm no foul, but it's a little confusing to see it bifurcating away from the centroid, so if it's of concern to you, fyi next time you're digging in that area of the code...

I've seen this behavior in several images so far, usually on second global stretch following a wipe action.

thanks,

Andrew

Re: Home-in is diverging.

Posted: Tue Sep 29, 2015 5:24 am
by admin
Thanks Andrew!

Yeah, I know it does that - it feeds back on what it last produced and 'sees' now. The idea is indeed that it converges on to a solution that works well, but the "trouble" is that once you get close to the 'right' value, there are multiple solutions that work equally well. Like you say, I may have a look at that behaviour and see if there is a better way of getting a 'single' answer. :)

Cheers,