Bug: area selection changes with Smart Histogram measurement

[neverfox]

It's not the case, right, that perhaps sensor analyses didn't survive the upgrade?

[neverfox]

Unfortunately, I can't revert to 3.3.7232.0 beta (the version prior to this change) because it has expired.

[admin]

Hi,

I have made changes for the next update - you will be able to choose which colour channel to measure (R/G/B or darkest for colour cameras, always Lum for mono) and tell the algorithm to stay clear of the frame corners/edges.

cheers, and apologies for the problem,

Robin

[Menno555]

Thanks Robin!

And 100% no need for apologies! Your doing a more then great job with all this and like you said: technically it made sense, so it was a logical step

Menno

[admin]

Hi folks,

the new update (now available) lets you choose the channel and whether to avoid the edges/corners (for info, this means anywhere further from the center than 50% of the distance to the corner).

Let me know how you get on.

Robin

[Menno555]

Thanks Robin!
Gonna test it asap

Menno

[Menno555]

Hi Robin

Just did test but sadly no improvement. I did choose the Green channel and the option to stay clear of the border, but still solutions of minimal 9 minutes are given.

Menno

[admin]

Hi,

do you get different results if you use different channels? Just wanting to check at this point that the changes I added are having an effect.

cheers,

Robin

[admin]

I just tested side-by-side 3.2 and 4.0, selecting the same area in 3.2 that 4.0 had picked automatically. Admitedly this is the camera pointing into the dark area behind my monitor with a very narrow aperture, but it's all just light after all, so should give the same results pointing at the sky...

As you can see below, the two versions gave very similar results in terms of both the measured brightness and the suggested exposure.

A bit of a puzzle !

cheers,

Robin

sc3.2.JPG (243.86 KiB) Viewed 1295 times

sc4.0.JPG (260.05 KiB) Viewed 1295 times

[Menno555]

Robin

My mistake, sorry I did the measurement yesterday with 1% RNL. Now with 5% it gives a nice solution of 232 seconds.
So, false alarm.

Menno