Though, in Blender, I think look was provided because the module has no other parameters: you cannot move the pivot, adjust the exposure, the contrast, the target black and white, or the primaries. Look provided some final adjustments, and its controls overlap the other parameters, which are exposed in darktable as sliders. look offset slope power | Desmos
Overall brightness should be the same, but you might feel that it’s getting darker because the highlights are compressed when AgX is first enabled.
The Look section includes similar controls to what you can find in Color Balance RGB. I know that Color Balance RGB comes before AgX in the pixelpipe, but are there any real benefits to using the Look section rather than Color Balance RGB’s similar controls? Or is it more for convenience?
They are display-referred operations, with all the benefits and drawbacks that brings. I find they are useful for some small adjustments (mostly saturation), but I rarely use them. As I said, I think they were offered in Blender because the AgX transform itself was not parameterised (had all params baked into the LUT). As far as I know, this may change in Blender, too.
Not sure if this has already been mentioned, but would it make sense to use the same slider graphics as RGB Primaries for the rotation sliders in the AgX Primaries tab?
They give you an indication of which hue the rotation will shift towards, whereas it’s not so clear with the current AgX sliders.
Note the CDL adjustments, as I said before, was part of the very first proof of concept AgX as an “Outset”, which was later replaced in AgX SB2383 by the outset matrix. The AgX Looks in Blender 4.x and later, are additional contrast curves applied before AgX in a log encoded state. The Punchy in Blender is also a simple darkening adjustment without the CDL saturation adjustment, becasue that particular part was replaced by the outset.
Is that little DNR offset maybe a little bit of what he might see. If I recall its in there to just be sure that highlights stay in check?? Or maybe not but actually in many images when I tested it I thought that it looked better set to 0 at the start, slightly brighter (in a good way) than with the correction…but I may have this wrong…
As for the look it might just be a workflow thing but I often tweak at least one of the sliders. Maybe its just convenience or whatever but I like not going back and playing too much with the curve once I get things “about right” . I find small bumps in the slope add a really nice brightness effect and likewise the offset can just bump contrast and saturation is nice to nudge as well… I don’t often apply large changes but I do find them really nice. I suppose it could be broken down and someone would well you could just do this or that instead but it just seems like a logical way to make a small but important change…
Thanks for the feedback. I’m sorry, I was mistaken; the original code on which I based the first prototype was from Minimal AgX Implementation | IOLITE. I never rearranged the application of the look. I guess it’ll stay as it is, now.
I also don’t use it. Someone suggested it as a default, others liked it, so there it is. We can always create our own presets.
Note that, due to per-channel application of everything, contrast influences saturation, and purity boost will increase the difference between channels, pushing the highest-valued component up (not limited to 1), the lowest-valued down (not limited to 0). So a bit of caution may be useful.
I would like to make desaturation of highlights kick in earlier/stronger without desaturating the main subject/midtones. I tried to play with the primary color desaturation sliders but that seems to:
Affect the whole image, not just the bright/dark portions of it going out of gamut
Act on separate colors - while it is gives the user the most control, it is a bit of an overkill for what I need.
Or perhaps you have other suggestions for desaturating the sky and (literal) shadows while keeping the main subject saturated and contrasty?
Thanks. I tried them and in the end I ended up using Color Balance RGB, (it allowed me to set different chroma for highlights and shadows).
But I am still confused about the desaturation settings in AgX. AgX’s key feature is to preserve hue of bright pixels by desaturating them (and does it better than Filmic RGB or Sigmoid), so I thought these sliders are provided to control this behaviour. But instead they seem to desaturate the whole image, something that could be done with many other modules. Is that intended?
You are redifining or modifying the primaries to compensate for tonemapping shifts esp for input data with harsh lighting so they are not really tools to specifically tweak highlight saturation… sounds like you got CB to work …you can also always create basically a luminosity mask for color eq and it would likely work just fine and I think the old color zones would let you select by lightness and then adjust color parameters so there could be outher ways as well
modifying the primaries to compensate for tonemapping shifts esp for input data with harsh lighting
Yes, this is what I expected it to do and I assumed these sliders are for adjusting this behavior (how much to attenuate and rotate primaries of bright pixels to preserve their perceived hue). It would be useful to have a slider balancing hue accuracy and saturation of bright pixels.
I thought the existing sliders in the Primaries section are doing something like that but instead they seem to affect all pixels, regardless of their brightness. That’s useful, of course, but it is loosely related to the main function of the AgX module and is already covered by other modules specialized in it (RGB Primaries).
We can always create our own presets.
which come to think of it I am not sure how well that translates beyond English for others