darktable 2.7 strange margin - bug?

yes, I did that, I re-edited from clean with version 2125

I did another edit with version 2128 and the stripe is still there

well and no: MX is based on Debian stable

apparently it is an issue with denoise profiled, I exported after each step of the edit

What’s the full version name?
Mine is this:
image

which, according to github commits page is one of the latest:

so it’s this you have to follow, you only have to choose the proper Debian version:

no, that is not the problem. apparently, dt 2.7 always produces this margin when I activate denoise profiled chroma and edit a raw from my olympus e-m10

I can’t reproduce this. Do you use a self created preset in profiled denoise?

Profiled denoise has received a massive overhaul, which thankfully makes all old “tricks” needless:

did you edit a raw form an Oly e-m10? I tried it with a raw form a Canon 6D mark II and the result was perfectly healthy

Sorry, I’m still not sure about that. I don’t know what release is that.
The version from your screenshot doesn’t show among the latest commits (see my previous screenshot, the one from the commits page. As I understand, those are master’s latest versions)
EDIT: my line of thought is: before concluding for a bug, we must be sure that we’re running the latest version. Talking about a bug from older versions doesn’t make sense because it could have been fixed on later versions.

you are making me furious!!! the verson is not the problem. I have tried it today with the latest version, too

1 Like

well, it is a bit strange. Apparently I cannot always reproduce this
edit: I can. It just does not work if I crop the margin away.

yes, on that one you posted in the first thread.

which system are you using?

@betazoid: you are also making me furious then :slight_smile:
"So, as you can see, darktable 2.7 produced a strange light margin on the right edge. "

And many people can’t reproduce. So that’s certainly a configuration issue, computer, graphic card, opencl or not, driver, full history of changes (modules used)… But as in many case people create report as if there was an universal issue!!! How many people will use dt if exporting was producing a “right margin” on all images when denoising was used?

So, we do our best to understand and fix issues, but we also do expect people reporting issue to give a proper context. We cannot guess or read in crystal balls. Thanks.

I do not want to make anybody furious but… I have the feeling that I am not taken seriously. Unfortunately I only have 1 computer so I cannot test on other systems.
I am not a computer scientist/programmer but I am doing my best to help.

Well…
Anyone will be able to judge:

Thank you for that. We all need to be patient. Bugs are hard to reproduce, and while its frustrating to have encountered one, the perseverance to accurately report and then resolve it will benefit us all.

thanks!

I think I have provided all info I can provide.

No you have not, please reread my comments, especially:


I could reproduce the issue again, this time following the “clean” steps suggested by @Pascal_Obry.
Details on my github comment.

Maybe this is nonsense, but may this have to do with spurious leftovers from earlier installs? Did one of those able to reproduce try with a fresh checkout of master, submodules updated, fresh, unused install location and empty directory for database and config?