Forwarded from Yuri
🧵 Thread • FxTwitter
Tim Soret (@timsoret)
This isn't getting enough attention.
Rive has solved vector feathering!
The blur isn't a costly post-process convolution, it's mathematically pure at the vector level = incredible performance even with ultra large kernels.
Incredible work from @csmartdalton…
Rive has solved vector feathering!
The blur isn't a costly post-process convolution, it's mathematically pure at the vector level = incredible performance even with ultra large kernels.
Incredible work from @csmartdalton…
Forwarded from Oleksandr Petulko
This media is not supported in your browser
VIEW IN TELEGRAM
Ладно іноді і вайбкодінг вистрілює😅
Forwarded from Yuri
навіть не знав про roundEven, https://fxtwitter.com/XorDev/status/1784058228667465777
🧵 Thread • FxTwitter
Xor (@XorDev)
fwidth only can detect the changes within it's 2x2 block, but not across blocks. So if there is a change between two blocks, fwidth misses it.
Here's where my trick comes in:
vec2 p=roundEven(FC.xy)-r/2.;o.xy=fwidth(ceil(p*r.y/dot(p,p)));
Here's where my trick comes in:
vec2 p=roundEven(FC.xy)-r/2.;o.xy=fwidth(ceil(p*r.y/dot(p,p)));