As some of you in the IRC notice, I've been having a problem understanding the visual value of the curves in the brush settings dialog. I notice that there is an absence of explanation of what the curves represent in the manual, I've seen only one video tutorial barely touching this topic, Revoy's explanation in his tutorial for calibrating the wacom stylus is the most complete piece of information I've found, yet still it was aimed to the stylus pressure in general. The curve may be something familiar for people with technical background and those who have had use them on other programs. For some of us is kind of hard to grasp what the dots and lines represent in values and numbers.

My idea is to try to understand as much as possible to vert that info into the manual, since I have no previous knowledge of this curve, I think my approach can be of help to anyone coming a new to digital painting.

On the brush settings, each function has its own parameters that can be activated in order to change the brush engine behavior and thus the outcome of the brush. Is rather simple to understand that activating "rotation" will rotate the brush tip, and that we can choose what action will trigger the rotation itself; pressure, tilt, angle, etc. That is the easiest part to understand. Now for the graph, that its a bit different. from what I manage to observe is that the curve (although it can be a straight line) represents two values at the same time, pressure sensitivity and the amount of effect delivered by the brush.

The concepts I can't grab are the dots and the line itself. are the dots coordinates with its own value or simply a visual modifier for the curve? Right now I am guessing is the second? since these two arrangements deliver the same result:


thus in the second graph, I'm thinking the third dot on the left-top corner simply push the up to the 100% range, but then what's the value of the pressure? low or high? if the dot has value, then would be low for the left dots (first and second picture) and high for the right dot?

Another thinking is that may be those two arrangements are invalid (visually) and the line/curve is just at 100% of the effect with the highest sensitive available regardless of been a line on top or a vertex, since the first graph (line on top) can't tell us weather position is on; low or high? this meaning that low = less required pen pressure and high = the opposite? Should those arrangements be used at all?

Experimenting a little bit I can see that in the following case, the curve/line represents the two values of the stroke, I think this is the easiest (if I am correct) that when the pressure is low (barely touching the tablet) the amount of effect is near to 0%, while the opposite is true. Putting the dots in the opposite corners will render an opposite stroke = a low pressure will bring the amount of effect to 100%, the strokes will resemble "hourglass".

If the mentioned above is correct, then the dots do have a value, in this example left dot is equal to "low = 0%" and right dot "high = 100%". But what the diagonal line means?

Now the following graphs represent what I think is the next step up.

In these I can see that the concept still applies, a dot in the middle will put the curves up or down increasing the values close the nearest dot, a curve up will increase the effect with less pressure required? I seems to be the case.

This following examples is when it gets difficult to visualize.

Are they equivalent to the previous curves? I appears that the space from the corner to the nearest dot is a subtracted value, meaning that the first graph will never have a 0% on the lower pressure? If the straight lines in these examples are converted to curves without moving the cornered dots, will that represent a mixture of the previous examples (more, less effects) with these? (minus pressure or effect?

To me the main problem here is to visually translate the curve into a desired stroke, thus if I want a stroke with rounded ends, what kind of curve I am looking for? If I want it to remain thin till I release the pen pressure, what variation should I try? answer to this would be in my humble opinion of much value on the manual. I think It will facilitate the creation of new presets and the manipulation of the strokes while painting.

The explanations don't seems to come easy though, especially when the curves become complicated.

Even when starting to understand the basics of the previous examples something like the above becomes really difficult to translate into what it represents on the actual stroke, do the dots still contain a proper value? is it mean to be an addition of curves (each curve represented between each dot)?

I totally appreciate any guidance and hope I will eventually understand the whole concept in order to share it with others. please feel free to comment, correct and add to the topic. I am not sure if I am making a big deal out of something simple, but I've been wanting to learn this since I starting using Krita, I don't think I should create presets (something I want to do) without a good background on this. Sorry for the long post. Thanks in advance for the help. : )

The curves represent a generated lookup table for which pressure which result value should be used. The dots are modifiers for the formula that generates the table/curve.

According to kis_curve_widget.h:

KisCurveWidget is a widget that shows a single curve that can be edited
by the user. The user can grab the curve and move it; this creates
a new control point. Control points can be deleted by selecting a point
and pressing the delete key.

(From: http://techbase.kde.org/Projects/Widget ... urveWidget)
KisCurveWidget allows editing of spline based y=f(x) curves. Handy for cases
where you want the user to control such things as tablet pressure
response, color transformations, acceleration by time, aeroplane lift
by angle of attack.

Now, the page it links to doesn't exist anymore. However, the class itself points to a cubic curve class, meaning this is the function used.
y=f(x) means that we manipulate x with function f then we get result Y. So a Krita example would be: pressure, as a value, being input into a cubic curve function becomes the size of the brush.

Now, I cannot tell you which of these is being used in Krita because power functions tend to be written very verbosely in programming, and curve-formulas tend to be very long anyway, so our blackbox is pretty lengthy: https://phabricator.kde.org/diffusion/K ... _curve.cpp

However, that isn't very important to the artist, as the artist can manipulate the formula by manipulating the curve, and can tell what the curve does by imagining a line from the bottom going up until it hits the curve, and which point the length of that line shows the result of that value.

So your examples become something like the following:






So the curve widget is like a black ox you put a value into, and while the maths itself is near incomprehensible to read, the curve is far easier to read. The result out of the black box is then used with the current value it's changing. So value pressure, thrown into our black box, results in a modifer for the size. So for the last one, a pressure of 0.25(with 1.0 being the max), results in 75%, and our brush size is, say, 30px. then 0.75*30=22.5 px.

The nodes are only there to manipulate the blackbox.

Colorspace trcs work quite similarly(though the convention of how the curve affects the values is a little different) https://docs.krita.org/Gamma_and_Linear

Finally, Quiralta, please do not upload 3000 px images onto the internet, I was deeply confused why Krita suddenly slowed down when I was editing these...

TheraHedwig wrote:T...So the curve widget is like a black ox you put a value into, and while the maths itself is near incomprehensible to read, the curve is far easier to read. The result out of the black box is then used with the current value it's changing. So value pressure, thrown into our black box, results in a modifer for the size. So for the last one, a pressure of 0.25(with 1.0 being the max), results in 75%, and our brush size is, say, 30px. then 0.75*30=22.5 px.

The nodes are only there to manipulate the blackbox.

Colorspace trcs work quite similarly(though the convention of how the curve affects the values is a little different) https://docs.krita.org/Gamma_and_Linear

I see now, is a constant variation not a static coordinates, interesting. I'm going to vert it into a little explanation then put it somewhere, with some illustrations. Hopefully will be of use for other hard headed folks like me

Finally, Quiralta, please do not upload 3000 px images onto the internet, I was deeply confused why Krita suddenly slowed down when I was editing these...

Aww, your poor little pc got a colic, nah, you know better, when lack of sleep, pixels are the least of the concerns, Thank you for the explanation and time as always