weird behaviour of display and histogram

[Gaaned92]

W10 and ART_master_0.2-3-gce8e3ce4f_W64_SSE4_191004

using photo from https://discuss.pixls.us/t/play-raw-daylight-haze/10491

and _DSC0653.NEF.arp snapshot3

1- when changing zoom,, displayed image change.

2- selecting and deselecting tone equalizer (all sliders set to zero) , histogram change.

do you observe same odd behaviour?

‌

‌

‌

[agriggio]

Is this still valid? I cannot reproduce with the current master.

[Gaaned92]

With 0.2-13 I get the same odd behaviour. It is a rnadom behaviour. you have to randomly change zoom and select/deselect in order to observe.

‌

I have an other observation regarding interaction of log encoding and exposure compensation. do you want I open an other ticket?

[agriggio]

sorry for the very late reply. I just tried again to reproduce, but no success…

Regarding the other observation, please open a separate ticket, thanks!

[Gaaned92]

a screen capture is better than a lengthy explanation

[Gaaned92]

For the log encoding, I lack some understanding of the function. So I am really not sure if it is a bad interaction or if it is normal.

[agriggio]

Just tried again, still unable to reproduce

I’ll try on windows…

[agriggio]

Interesting… I can reproduce the histogram changes on windows (still no luck with the brightness changes though). Using the attached patch, the problem seems to go away, but I still don’t understand why… it would be great if you could try on your machine and let me know what happens

[Gaaned92]

I fear it is not solved. Evevn if all sliders are 0, clicking on the tone equalizer modifies all the histograms. They seem coherent with the small thumbnail on the left.

[Gaaned92]

Happy new year

[agriggio]

Indeed, it doesn’t solve the problem. In fact, the bug seems to be totally unrelated… In the meantime, I’ve been able to reproduce also the brightness issue, but both only on windows. At the moment I have no clue what the culprit might be, unfortunately…

[Gaaned92]

No glitch if I run art.exe under gdb.

So perhaps a critical race

Is there a way to use only one processor?

[agriggio]

definitely a concurrency issue. But I have to at least isolate the module in which it happens first (I’m still not sure it’s the tone equaliser)

[Gaaned92]

It doesn’t happen on all photos. I have to verify.

color correction? if I disable it there is no more glitch. Is it due to area mask?.

the feathering doesn’t seem very robust. I let show mask active. I go back to tone equalizer and I see that the mask is modified.

[agriggio]

area masks: properly initialize (and not overwrite) the guide buffer even when no parametric mask is active

Fixes #7

→ <<cset 17fe5608ee23>>

[agriggio]

I found the bug – it was due to an uninitialized buffer being used in some cases when generating area masks. It should be fixed now