I wanted to create a simple paint brush in Krita, but I couldn't.

Here's the problem:

1. press D to select black color
2. press F5
3. select Pixel engine
4. press "Default preset" button
5. increase Brush Tip's diameter to 100
6. change Opacity to 0.05
7. change Painting Mode to Build up

Now, using a mouse try to paint pure black color on canvas. You can't. Darkest color you can paint is rgb(10, 10, 10).

I'm using Krita 3.3.3.

Um...

"6. change Opacity to 0.05"

How do you expect that to paint pure, 100% opaque black? That's about as opaque as the Oort Cloud. This stuff really is just a matter of maths. It will never build up to a completely non-transparent black.

Well, I have such brush in GIMP (Airbrush Tool), which builds up to ANY color I want after say 10 strokes.

I can't help that: a) Gimp is a different application and b) an airbrush isn't an ordinary pixel brush, but something that uses a timer to deposit more color even when kept in place.

b) Yes, it's called Airbrush Tool, but it can be used as a regular brush by setting "rate" to 0. Build-up effect can be achieved by setting "flow" to some low value. I think that tool is only available in GIMP 2.9.6 (development release).

captaingox wrote:b) Yes, it's called Airbrush Tool, but it can be used as a regular brush by setting "rate" to 0. Build-up effect can be achieved by setting "flow" to some low value. I think that tool is only available in GIMP 2.9.6 (development release).

No, it's not, it's been in gimp since at the least 2.4, but the real thing here is again that Krita is not Gimp as much as Gimp is not Krita. So different formulas are used for the maths.

The only thing I can think of, Krita's default brushes have flow and opacity set to a pressure curve, so perhaps that is what is annoying you.

Yes... And while things like pencil or eraser or airbrush are tools in applications like gimp or photoshop, those things are brush engine options in Krita.

Thank you for your time I'll keep experimenting with Krita and other programs.

I solved the problem by using 32-bits float point precission for channel depth instead of 8-bits integer which is by default.

Cheers!