No, and I find the hostility off putting.
Sorry, I did not mean to be hostile. I just find this discussion pretty frustrating.
Still, my impression is that many people who are criticizing here have not tried it. When using it, the benefit should be immediately apparent.
3 Likes
They donāt bother me either, and I think that argument of ātoo many modulesā is a poor one. Users can happily ignore them and hide any module they want.
And as mentioned previously, we have multiple options for tone mapping, colour grading, sharpening, denoising, etc., so I wouldnāt see a problem at all with having a Tone Equalizer alternative. Itās not like we have a glut of them.
Am I wrong but to me you have just introduced some new flexibility into the mask exposure and contrastā¦the presets are shifted left, shifted right, centered and custom which allows pretty much what you could do by tweaking the old separate controlsā¦ yours seemed quite responsiveā¦ Then you collapse the other mask controls on to the main tab and well we debate on about the simple set of slidersā¦
AP has merged the two tabs in Ansel and dropped the need for that second tab and reference bar graphā¦ and he just left the simple tab as the second tabā¦
Other than that nothing really changes in the output correct or not correct???..so its just a UI update at this point and with a couple of enhancements to set the mask exposure and compensation??
I thought it was pretty good. I thought that there was a pretty good case to have the mask controls with the graph but maybe others are opposed and maybe things have gotten side tracked around the curve and the simple slider interface so that the main things offered here are not staying in the fore front of the discussionā¦
I fear I cannot offer to much constructive input on this thread. I use tone equalizer not really often. I somehow donāt like it that much, unfortunately without exactly knowing why. I know that sometimes the produced output has some flaws like halos. Anyway this counts for other modules as well and I still like them - like local contrast. So I think all in all it is the handling.
What I definitely like on this proposal is āauto alignā. This is something I always hated on tone equalizer, especially because masking is in a separate tab.
On the rest I fear Iām useless. Is there a easy to use linux version for testing?
1 Like
This.
1 Like
As Iāve understood from the videos of @s7habo he uses the shadows and highlights sliders under perceptual brilliance grading in CB RGB to obtain this in a quick way.
Wasnāt it stated in a thread a couple of months ago that the visible curve in the UI also reflect how the function behind behaves. If that is correct, shouldnāt the function then for the segment where the function changes from a positive to a negative value be expected to result in a change in the opposite direction of what the user have intended?
But they are tonal (pixel by pixel) adjustments, while the tone EQ is a spacial (area by area) adjustment.
1 Like
Iāve been playing with this for about 20 minutes, and it doesnāt really feel like the same module to me.
The auto fitting is nice in theory, but Iāve never really had an issue in the current module fitting the mask to what I want to adjust.
This curve still undulates when pushed, so there isnāt any improvement there for me.
When I push to nodes that are close together, I feel like I get artifacts where I donāt in the current module.
The sliders in a collapsible section, I donāt feel differently about this, I think its a bad move, it takes up more vertical space which is prime real estate. Not an improvement for me. It also leaves the masking tab really empty; its weird.
For me, this feels like a new module, not an improvement to an existing module. I am not against getting new/more modules, I am against loosing what I already have.
1 Like
Interesting I find that with it I rarely have halos and for me a very small tweak I use when it happens is to bring the feathering slider roughly even to the default position of the smoothing sliderā¦
I use the relight preset a lot and then just tweak the darker and lighter parts up and or down apart as needed and I find it really does a nice job on so many photosā¦
2 Likes
the module cannot know the intention of the user. It just follows the implementation - and so the intention of the developer.
The developer of the module expected a user who knows about the characteristics and limits of the module ā¦
1 Like
As is the case for any toolā¦ using a hammer to drive a screw doesnāt work all that well, nor does using a big hammer on a small nail (or vice versa)ā¦
Maybe the developer used a curve prone to oscillation because they didnāt know of a better solution. Thereās no reason not to change that behaviour, if a better solution is found.
But this topic is (was) not about the oscillations, right? There was another to discuss curve behaviour.
3 Likes
This is the point that I like most from your proposal.
Lately, Iām taking photos of events (voleyball, family and club hikings, football, MTB, ā¦). You can imagine, a huge number of photos each time.
When I work the photos and apply TE changes then I have to go one by one, doing double click in āmask exposure compensationā which it is a pain in the neck.
Actually, in the last session, I avoided its usage because of it.
Personally, regarding to curve behaviour and the sliders, I do not lose sleep over it. But it is because I only use the graph or the photo to modify the curve but no the sliders.
Thanks for your effort/work
3 Likes
iām pretty sure AP was exactly knowing why he used that curve ā¦
Reading the preamble in the source codeā¦AP acutally notes that oscillations might happen
The exposure correction is computed as a series of each octaveās
- gain weighted by the gaussian of the radial distance between the
- current pixel exposure and each octaveās center. This allows for a
- smooth and continuous infinite-order interpolation, preserving
- exposure gradients as best as possible. The radius of the kernel is
- user-defined and can be tweaked to get a smoother interpolation
- (possibly generating oscillations), or a more monotonous one
- (possibly less smooth). The actual factors of the gaussian series
- are computed by solving the linear system taking the user-input
- parameters as target exposures compensations.
Happy to move this to the thread you mention and maybe it still doesnāt change anything related to peoples concerns about the the curve mathā¦but even reading the preamble to me the fact that the data comes from an estimate of luminance because of where the module is in the pipeline and then some of the other comments related to the math and transforms where I get a bit lost but as suchā¦I think the points on the curve are target adjustments but its not as simple as a one to one change in output based on a strict interpretation of the curve or the individual sliders in the simple tabā¦
I quite like this idea as an alternative solution (rather than updating Tone Equalizer). Or even something more like the current Color Zones module in appearance/usability? I prefer the idea of a horizontal spline rather than diagonal, but thatās not particularly important. But one thing I particularly like about RGB Curve and Color Zones is the ability to add/subtract nodes for smoother transitions or finer control as required.
I know the idea of a new module is not what this thread is about or what you have been working on, but I want to harness your enthusiasm and give you encouragement because not everyone is on board with changing the current Tone Equalizer. Iāve found that new modules are generally received more positively than changes to existing ones, especially when the module in question is a one-of-a-kind.
At the very least IMO this would be an improvementā¦ie the tab mergeā¦ this is what AP has done in his fork and it would cut down on back and forth
10 Likes
Yep, agreed. I still think there are some improvements to be made with the module. And I have to agree with others that itās annoying to have to revisit Tone Equalizer each time you make an exposure change, although that can be mitigated by using other modules later in the pixelpipe.