I read his “Cognitive Mumbo Jumbo” section and couldn’t pull a thing out of it. Maybe it’s just me. No disrespect intended towards the man, but sometimes he seems determined to write cryptically.
@kofa you didn’t answer my question regarding the profile that you use in the base primaries tab to handle colors better, is it the working profile or the exporting profile ? because I didn’t find a clear answer in the previous discussions. I mean in sigmoid it is clear that Srgb works best for handling high values back into gamut.
Lol almost everytime I ask Troy something I feel instant regret; but hey, we’re here because of the mumbo jumbo mate.
Thanks for a good reply @kofa.
But it’s all quite complicated for me; I’m going to bow out, wait for it to be “productionised” then probably come back to it.
Good luck finishing it off.
There is a reason why he isn’t on this forum and the way he often talks down to people is completely unacceptable.
He may seem harsh but I learned a lot from him actually so let’s not get off of our main topic please.
I echo that sentiment and if we are lucky someone will produce another excellent video showing us how to use it ![]()
Eary_Chow said above that a larger space is often better:
Yes.
As for explanations, I find Troy’s writing very hard to follow. Sakari, Eary_Chow and Jed Smith have helped a lot. Jed has a very nice Wiki with visualisations.
Aurélien’s perceptually uniform space:
I don’t think so. Not because I actually understand the process deeply (I’m learning alongside you, that was the whole point of this project), but because those who know don’t use it.
No, the xy coordinates are different. Primaries always have RGB coordinates where one component is 1, the others 0. That’s why they are primaries: they define the axes of a coordinate system (they are the basis vectors of the space).The xy coordinates, which determine exactly what shade and how pure a shade of red, green and blue they represent, are different. They are the corners of the triangles drawn inside the ‘horseshoe’ xy chromaticity diagram.
I saw this response. So, to my understanding for now… we use rec2020 and push the attenuation further will give better results than using Srgb as a working space for primaries ?
Because he said that we can use wider primaries and “hard clamping” is fine, I guess the last part means the output color space of choise that comes after Agx or any view tranform ?
But I know that some tweaks and changes happened in the build after that discussion so correct me if I’m wrong or maybe @Eary_Chow can confirm that.
Thanks, I had taken a cursory look at that before. I think I shared the link to his github site at one point as a treasure trove of information and examples of tonemapping ( that may have been around the openDT implementation in ART though I can’t remember)
I agree, Troy is a bit hard to follow…I think of birds flying not hues so the descriptions are interesting…I did like the animation of the hues in Jed’s wiki…the one where he animates the 50% to 100% chroma range and then demonstrates how a small rotation in the primary alters that…its a nice visual as I often find unless your brain is wired for it (and I think mine is likely not ) it can be hard to visualize some of these diagrams and translate the 2 and 3D representations into something you tie back to changes you might make in an image…
Thank you for that explanation. That makes complete sense to me.
could an extra bypass switch be attached to the “primaries” to make it easier to make an a / b comparison?
We are experimenting, so sure
You’ll have to wait until the weekend, though.
i am excited ![]()
Maybe the curve/advanced block no longer needs to be a collapsible section, and the primaries also don’t need to be placed inside one, now that they are on their own page.
I hope it works. ![]()
https://tech.kovacs-telekes.org/dt-agx/Darktable-5.1.0%2B857~gbd6df65f4b-x86_64.AppImage
@MStraeten , @priort , @Dave22152 , please update your builds.
@Suki2019 : hope to get to your request, too (it’s not the weekend, yet).
So far, so very excellent. I am not qualified to comment on the primaries (Don’t really understand them and at present not enough free time to search and learn). It also appears that backwards compatibility with edits from recent prior versions remains which is equally excellent.
It was just a UI update, so params remained compatible. I’ll break them again when I implement @Suki2019’s request (plus some of the other stuff I plan to do). Sorry, it’s a prototype. Migrating parameters would be even more work.
Please do not apologise. Looking for white thumbnails in lightroom mode give me plenty of opportunities to re-experiment on images. Everything is working so very well. I can achieve with Agx alone what might take a lot of time and many modules. Re-visiting old edits and them comparing new Agx versions reinforces quite how bad some of my old edits actually were. I am very grateful for everything that you have achieved.


