I’d turn off XMP writing and specify a different database, cache, and config location for the AgX version. If you update your db, there is no way back to dt stable.
Also make a backup and tag it before you start.
I’d turn off XMP writing and specify a different database, cache, and config location for the AgX version. If you update your db, there is no way back to dt stable.
Also make a backup and tag it before you start.
That depends on how you interpret the scene.
You wanted the area in the shadow with the road to be brighter and at the same time keep the contrast in the clouds.
This is actually not possible if you want to keep the brightness ratio between these two elements.
In your interpretation, the area with the road in the shadow is just as bright as the white clouds directly illuminated by the sun’s rays. This makes the clouds look very dark, almost like a gray smoke.
In my perspective, this would be a much more realistic interpretation:
I would only take one more instance of the exposure module and use the mask to brighten only a small part of the road in the foreground slightly more to put the focus point on the road:
And at the end I would use the color look up table to darken the blue sky a little so that the clouds come out even better:
Because the surroundings have remained darker, these two elements have been emphasized.
What I wanted to say with regard to your question is, yes, the tone mapper does not really play a role here. This is about an interpretation of the scene that has nothing to do with what a tone mapper is supposed to deliver.
I don’t know. It’s probably just me. This has become a very complex module to use. I understand that for someone like @s7habo, it’s just another day in the park, yet I do appreciate the depth of understanding that he has when working with all of the sliders along with the interactions between them.
I’m just wondering if I could simply use AgX along with cb rgb and still achieve a nice result which is similar to what @s7habo is able to get using the AgX/Sigmoid module. A result that might be better than using filmic or sigmoid as I have done in the past.
Really, this might not be another DoS but for simple minded folk like me it’s quite overwhelming. Then again I am sure that I am in the minority so, carry on guys.
I think the module has largely remained the same since it’s inception. I think we can make the sigmoid section collapsible. If you only use the top part of the module then it’s basically the same module as it’s first version with all it’s quick and easy beautiful results. I do appreciate the sigmoid section though, and often play with the white relative exposure and other options.
I even enjoy the saturation effecting the contrast, somehow it feels more intuitive for me this way.
The top section is kind of the gateway drug that can convince newcomers and veterans to use the module, and as your curiosity grows you can experiment with the sigmoid sliders and develop an intuition of what they do and when to use them.
All in all I love the current configuration and it’s already the main way I develop my pictures. (I know it’s not stable yet but I’m fine with re-editing my pictures as new versions come out)
I think your opinions are just as valid as any, so don’t worry about sharing them. I’ve no idea if you are in the minority because there are really only a handful of us providing feedback in this thread.
Personally, I don’t think it’s another DorS module. That module is quite an outlier in how non-intuitive it is to most people. As some of you know, I have fairly strong opinions about that module, so I would definitely be speaking up about AgX if I thought it was getting anywhere near that!
I do understand the point of view that AgX is becoming quite complex, especially compared to earlier versions when it was just 4 sliders and still producing excellent results. There was a wonderful simplicity about it. However, as long as the more advanced controls remain optional and are hidden behind a collapsible section, I think we’re still in good territory.
I think @kofa’s original idea was to have fewer sliders, and I’m still not convinced we do need all the current ones. @s7habo has shown that they can be useful, but I still get a nagging feeling that we’re getting incremental gains from having several sliders that can do similar things. I’m still on the fence about the two relative exposure sliders in particular. I find myself rarely using them. Yes, they can all help, but we’ve already seen that sometimes the Tone Equalizer is still required, so maybe we don’t need to cram so much into AgX? I’m happy to be convinced otherwise of course.
I honestly don’t get it why it’s more difficult to use than the initial version. There you had:
So ignore all sliders other than those of the look, and it should work the same way. You don’t have to move all the sliders only because they are there.
After over a decade of talking to people about raw processing software, mostly about dark table, I convinced there is something inside most people that makes them have to fiddle ever knob. I don’t know why and I don’t understand, but it’s like something that they have to do and no amount of explanation will make them not do that.
Yes, this is an extrmey broad generalization. Its. Of leveled at any one person. It’s not meant to be an insult either, just an observation.
@Popanz I have a simple idea for the offset, I’ll try to add it tonight.
Exactly.
For those who are afraid of complexity. This is the famous Swiss Army knife:
In other words, if you want to cut a slice of bread, despite all the possible options, the knife edge alone is enough.
But you can also open a can and a bottle of beer and even use the toothpick after the meal.
For a nice picnic, this reduces the number of tools you would otherwise have to take with you.
So, it’s not some complex instrument to do a very specialized operation that you have to master first. It is a very simple knife with a few additional, useful features.
That’s how agx should be seen.
If additional settings can be hidden, it won’t be much different than the handling of the other two tone mappers we already have.
I’ll try to do that, but I know nothing about the UI. I’ll see what I can copy-paste this time.
I think many of the sliders in AgX have analogous controls in Sigmoid, particularly in the 4-ways tab, so you ought to be able to do your initial edits in the Look tab and hop over to CB RGB for refinement.
The sliders of agx either:
color balance rgb is unbounded, scene-referred.
Thanks for the clarification. What I had meant was that I could perform similar operations in CB RGB.
For me that’s a huge advantage with your AgX module. With Sigmoid and Filmic, I perform the tone mapping and refine in the four ways tab in CB RGB, where in AgX, it’s there beneath, so I don’t have to hop from one module to the next.
I bet people once said that about display-referred processing!
here Claudes attempt to adapt the sigmoid way using collapsible section to the agx code:
just had to manually add the sigmoid section definition to typedef struct dt_iop_agx_gui_data_t
but the generated gui_init
was perfect
Thanks, it’ll be in the next build.
No rush, it’s something that can be added later on if/when this is added to master. We’re still in the testing and early feedback stage, and I certainly don’t need it collapsible at the moment. I’ll probably always have it expanded anyway because I like tinkering
My gut feeling about this is that it’s more of a workflow and choice paralysis issue. Perhaps the fact that there are more sliders has made it less obvious which sliders should be moved first or in what order, especially as several of the sliders can all be used to set the black/white points and contrast?
In the earlier version, it was adjust Slope, adjust Offset, adjust Power, done (or you could do it in another order, but essentially just 3 adjustments). Now, you might wonder whether to use Slope, White Relative exposure or Target White to adjust the white point, and after that should I use Offset, Black Relative or Target Black to do black point. And then should contrast be further adjusted using Slope of Linear, Power, another module even… etc.
More sliders = more choice = greater chance of choice paralysis
… which is something people suffer from in varying degrees.
The module actually hasn’t become harder to use; there are just more options now. You can still just adjust Slope, Offset, Power and get good results, but the other sliders are there seducing people into thinking they should be used (which is more of a psychological problem as @paperdigits alluded to).
I can only speak for myself really, but I’m happy to have lots of sliders as long as it’s clear what they all do. And we’re still very much in the experimentation phase here, so I don’t think we need to get too hung up on this issue yet. We let @kofa iterate and experiment, and gradually we will probably reach some consensus on which sliders are the most useful and what the overall goal of the module is. It’s then just a matter of labelling the sliders clearly and providing helpful tooltips, which I’m happy to contribute to if I can.
At some point it would probably be good to take a step back and consider what this module can and should handle vs. offloading to other modules. I’m all for options and choice but with that comes more code complexity and the potential bugs and issues that come along with that. And of course, confusion and questions from users. Having a Swiss Army knife can be useful, but if we already have a can opener and toothpick then it might be unnecessary. There’s no right or wrong answers, but it can help justify features and begin the documentation process.
Of course, the ultimate solution to ease of use and decision paralysis would be to have one module that has a slider ranging from “Bad <—> Good”
Kudos to everyone’s work on this. It’s really cool to see the work from @kofa and all the discussion and feedback happening in real time. I’m about as ignorant to all the maths and reasoning on this stuff as you can get but it’s been fascinating watching all this unfold and trying to learn as I go.
@s7habo made a few photo edits above where he also used the Sigmoid section.
Not restricted to only using the AgX module, but without Sigmoid, I would like to see how he would achieve a similar result as when he used the Sigmoid module.
Would this be of interest to anyone else besides me?