How to resize a knob in Knobman?

[SampleScience]

Sorry if it's a stupid question, but I'd like to resize some knobs I've found on the Knobman website but I don't know how to do it properly. If I resize the knob with another program it makes them unusable in the plugins I program (you see 2-3 knobs instead of one that turn properly).

Any idea how I could do that?

[EvilDragon]

Best would be loading the .knob file of the knob you want to resize, then simply adjust the width and height on the Properties tab, then export. If you resize just the resulting knobstrip, in most cases you will not get a good resize because of resampling artifacts.

[RunBeerRun]

If you're seeing 2-3 knobs, maybe you need to define the size in the program used, that's where I've seen the double knob errors.

[SampleScience]

Best would be loading the .knob file of the knob you want to resize, then simply adjust the width and height on the Properties tab, then export. If you resize just the resulting knobstrip, in most cases you will not get a good resize because of resampling artifacts.

I don't see the properties tab. But I see in the preference tab a output image size, is it good for a good knob export?

[EvilDragon]

That's it.

[mystran]

Best would be loading the .knob file of the knob you want to resize, then simply adjust the width and height on the Properties tab, then export. If you resize just the resulting knobstrip, in most cases you will not get a good resize because of resampling artifacts.

In most cases, if you use an image editor with good quality resampling, then exporting at 2x or 4x from Knobman (and resampling afterwards) is likely to give you better quality (neither knobman's anti-aliasing or the built-in oversampling is that great in quality; they aren't poor, but the average image editor does better). Just remember to leave a bit of margin around your knobs (or whatever you are rendering) so the resampling filters don't bleed from one frame to the next.

[SampleScience]

Thank you for your help guys!

[antto]

mystran: yeah, the antialiasing in knobman/skinman is kinda poor.. the text antialiasing is not done in linear color space
and also a funny thing - in knobman, make thin circle... now rotate it - it's not quite a circle
i tend to rotate those circles at 45° because they look a bit better then

[mystran]

mystran: yeah, the antialiasing in knobman/skinman is kinda poor.. the text antialiasing is not done in linear color space

I think the whole "anti-aliasing in linear color space" stuff is a fallacy to begin with. You just end up with far too much weight for the brighter values, so everything that doesn't perfectly align to the pixel grid will look washed out. Essentially, it's one of those things that seems to make sense, until you actually turn your brain on and work out what's going.

edit: for text specifically though, I can see why some people might prefer messing with the gamma curves, as dark-on-bright glyphs will look a bit thinner, bright-on-dark a bit heavier, and pretty much all glyphs a bit less "dirty" at the cost of losing a bit of anti-aliasing.. but really to claim that it's somehow "correct" is just non-sense.

[antto]

well, no idea about how vector fonts are rasterized, but here's a comparison:


(this is from skinman)
at the top, i used colors 10,10,10 and 211,211,211, then i loaded this in an image editor and applied gamma (which made everything brighter) ... those two colors became 63,63,63 and 235,235,235 respectively
then i went to skinman and just changed the colors and that's the bottom image "direct"

surely the ellipse rendering is good, because it looks proper, and then looks ugly on the modified variant, but tbh the text rendering looks better on the modified image (IMO)

now, i changed the dark color to 1,1,1 (thus even darker)

after shaping this becomes 42,42,42

i think in this case the difference is a bit more visible

and, finally, here is "directly" how it looks with the colors 1,1,1 and 211,211,211

[mystran]

surely the ellipse rendering is good, because it looks proper, and then looks ugly on the modified variant, but tbh the text rendering looks better on the modified image (IMO)

Like I pointed out above, thinner blacks vs. heavier whites can look subjectively nicer, but you could get it to look a lot better by tweaking other aspects of the text rendering (and I do think the modified versions look considerably worse in each case, but then again doing gamma on LDR images is going to look poor no matter what). At sizes this small, you might want to do hinting at least vertically, consider sub-pixel rendering where possible (which usually allows you more horizontal resolution, so you don't need to accept poor kerning as would typically happen with full hinting.. though even in gray scale it might be better to prefer kerning accuracy), and possibly tweak the AA filter weights (eg give more weight to values near the pixel center; this can give you a bit similar results to what you get from gamma tweaks, except a lot more consistent).