Smudge Tweaks testers needed

https://github.com/mypaint/libmypaint/pull/122

If you can compile mypaint please try this out, and the brush pack. Pretty nifty new settings that surprisingly don’t require a lot of CPU. I actually bundled a windows installer if you want to try this alongside some funky CIECAM color adjuster stuff:
Windows installer:
https://github.com/briend/mypaint/releases

Brush pack:
https://github.com/briend/Brushes/blob/master/README.md

That’s a great step towards realistic brushstrokes!

I’m currently trying out the smudge_tweaks branch a little and I really like the Smudge-Bucket setting in combination with Angular-Offset-Side. Will definitely use it in my next paintings.

One little “problem” I found is how the bucket order looks inverted on every back-and-forth stroke, because it basically follows the painting direction.

1 Like

Thanks for trying it!! You are absolutely right, the order of smudge buckets switches when using direction based offsets. I didn’t notice this for a long time because I was using the Offset Asc versions, which follow the Ascension of the stylus (which requires tilt data.) instead of the direction.

So, if you have a tilt aware stylus it is a simple fix to switch it. I’m planning to make another brush pack that is optimized for tilting stylus, whereas this one is meant to work well with just pressure alone.

Maybe there is some way to get it to work with directional offsets but I’m not sure

Oh and keep an eye on the brush pack, I’ve added a bunch recently using some other settings, some pretty cool :joy:

heyaa briend, do these brushes include the REAL MIXING ones ?

“real” as in the spectral mixing method? No, but it does use the same subtractive mixing method, which is based on weighted geometric mean. So, it kinda feels like real paint despite being just RGB data. Some day I want to revisit spectral rendering but it was too cpu and memory intensive for now

yes the one discussed in the Thread where you implement Scott Burns’ theory :D, I’m quite a noob when it comes to compiling and stuff like that, I downloaded your RGB.txt, and the latest alpha version of mypaint you uploaded 2 days ago for windows, but it doesn’t want to open, I double click and nothing happens, in the debug mode, it says libpaint module is missing. Any help with this would be greatly appreciated, and god, those videos on youtube show great potential, as I always was looking for a program that would mix like traditional media :3

People watch those videos? :slight_smile: Thanks! I think the current smudge tweaks is almost as good as spectral. Sure, pure blue+yellow doesn’t make a green, but it won’t make black either (like normal multiply mode does).

Crap, I really feel bad for flooding the build log with these builds that don’t work. I’ll have to stop pushing patches to that until I can figure out why that error message shows up. Sadly, the windows alpha builds are not setup to use anything other than the official libmypaint, so it wouldn’t have worked anyway. Well, you would have had my cool CIECAM based color adjusters but not the smudge tweaks :slight_smile:

I will try to create a windows bundle w/ the smudge tweaks and link it here.

I created a windows bundle and installer that should actually work. It has the smudge_tweaks as well as CIECAM color adjuster stuff:

https://github.com/briend/mypaint/releases

wow, thankyou :3, i’ll try it now and feedback :smiley:

Soo, I put the RGB.txt in the Standalone folder where Mypaint.exe is located, and imported your brushes, didn’t see any hiccups when selecting any of your brushes and also my RAM usage didn’t increase, and as expected the mixing of blue and yellow turned greyish black, in the Debug.exe there doesn’t seem to show any errors saying anything about RGB.txt even when I delete it from the folder. I don’t know if I did something wrong hehehe :smiley:

Sorry, I don’t think I was clear-- this patch does NOT use Scott’s spectral stuff (rgb.txt), however it is using a subtractive mode based on his same ideas (weighted geometric mean), and it can use linear RGB (the smudge pow setting) for the smudge mixing. So, it is quite a bit more natural than the previous methods. Let me know what you think! And sorry you downloaded that huge RGB.txt. The spectral code is still on github and I will revisit someday but don’t have time to dust it off right now. And really, the performance is pretty darn bad.

oooh, i see, so you mean that I won’t be able to see results like in this video https://www.youtube.com/watch?v=RUa8IHNtGb4&t=329s , and can expect only the results of linear RGB ?

Actually, no, it won’t be like linear RGB “normal” mixing in that video, it’ll be very close to the spectral “pigment”. In that video I hadn’t yet separated the concept of spectral wavelength mixing from the “pigment” weighted geometric mean (WGM) “subtractive” smudge-blend mode. When I separated them into their own settings I realized that using linear RGB (a smudge pow/gamma of ~2.4) in combination with the subtractive WGM blending gave a very similar effect to the spectral method.

hmm, that’s strange then, can’t seem to get results similar like the ones in the video

I see what you mean. Spectral simulation gave us that “blue+yellow==green” benefit. Pure Blue + Pure Yellow is sort of a contrived example though, I’ve found. And, actually, we can still get blue+yellow=green if we are willing to sacrifice some color saturation. Here’s an example:

Top-Left is pure yellow and pure blue. Notice it’s kinda gross like your example. However if at least ONE of the colors is somewhat desaturated we get a nicer blend. Lower Left is both colors slightly desaturated. Upper Right is both colors even more desaturated. And the the lower-right is a pure yellow and a fairly desaturated blue. These are all the same exact hue angles for yellow and blue so I haven’t done any tricks.

If you use your color picker you can see (and hopefully you can tell with your eyes) that there actually IS green in the three of these that have a desaturated color.

I think one reason I’ve been pretty happy without spectral is that I’ve been avoiding these fully saturated colors that bump into the gamut limits of sRGB. In fact that CIECAM code I included in the build above includes some features to specifically limit your “saturation” automatically without having to be so conscious of it:

image

wooo, now the things blend like they should :smiley: thankyou very much :3

1 Like

Woohoo! Hopefully you have a pressure-sensitive stylus!

hahaha, of course :3 and btw i like your brushes very painterly <3

Thanks! Let’s see some painterly paintings! :wink: