R H
@flinflonimation.bsky.social
Programmer, in Winnipeg, made a couple video games with Unity. Now learning Godot. He/Him
Some games here: https://rhill.itch.io
Some games here: https://rhill.itch.io
Like I guess it’s technically running a shader to generate the sdf but the shader is running a 2D for loop to find the shortest distance which is what I’m calling brute force.
In theory a compute shader storing intermediate values could do it O(n) passes instead of the O(n^2) but I didn’t bother
In theory a compute shader storing intermediate values could do it O(n) passes instead of the O(n^2) but I didn’t bother
December 2, 2025 at 5:09 PM
Like I guess it’s technically running a shader to generate the sdf but the shader is running a 2D for loop to find the shortest distance which is what I’m calling brute force.
In theory a compute shader storing intermediate values could do it O(n) passes instead of the O(n^2) but I didn’t bother
In theory a compute shader storing intermediate values could do it O(n) passes instead of the O(n^2) but I didn’t bother
You can see on the right side me deciding I don’t have a pale enough blue brush marker and switching to gold instead.
November 29, 2025 at 4:50 AM
You can see on the right side me deciding I don’t have a pale enough blue brush marker and switching to gold instead.
It will happen if there are genuinely only two possible values.
But in this case, that requires confidence you can classify every article into AI or Human and never have to say Unknown or Secret Third Thing
But in this case, that requires confidence you can classify every article into AI or Human and never have to say Unknown or Secret Third Thing
November 28, 2025 at 4:57 PM
It will happen if there are genuinely only two possible values.
But in this case, that requires confidence you can classify every article into AI or Human and never have to say Unknown or Secret Third Thing
But in this case, that requires confidence you can classify every article into AI or Human and never have to say Unknown or Secret Third Thing
In both cases I find I have to turn off compression on the texture’s import settings. Compression that looks fine in an image looks terrible when it’s a vertex offset.
November 27, 2025 at 7:01 PM
In both cases I find I have to turn off compression on the texture’s import settings. Compression that looks fine in an image looks terrible when it’s a vertex offset.
I think that one I baked using this Blender plugin
github.com/Knetic/verte...
More recently I’ve tried out OpenVat which streamlines some of the process in comparison so I might have an easier time convincing collaborators to use:
openvat.org
github.com/Knetic/verte...
More recently I’ve tried out OpenVat which streamlines some of the process in comparison so I might have an easier time convincing collaborators to use:
openvat.org
November 27, 2025 at 6:54 PM
I think that one I baked using this Blender plugin
github.com/Knetic/verte...
More recently I’ve tried out OpenVat which streamlines some of the process in comparison so I might have an easier time convincing collaborators to use:
openvat.org
github.com/Knetic/verte...
More recently I’ve tried out OpenVat which streamlines some of the process in comparison so I might have an easier time convincing collaborators to use:
openvat.org
I guess most of my effects lately have been explosions which don’t repeat which is why I haven’t been relying on it more?
November 26, 2025 at 3:46 PM
I guess most of my effects lately have been explosions which don’t repeat which is why I haven’t been relying on it more?
Code for it up here:
The neat work-flow with it is how you can just type in a string like acgbef or aannzzaa and that becomes a flicker pattern. Makes it very quick to iterate different animations.
The neat work-flow with it is how you can just type in a string like acgbef or aannzzaa and that becomes a flicker pattern. Makes it very quick to iterate different animations.
display_light_pattern.gdshader
GitHub Gist: instantly share code, notes, and snippets.
gist.github.com
November 26, 2025 at 3:41 PM
Code for it up here:
The neat work-flow with it is how you can just type in a string like acgbef or aannzzaa and that becomes a flicker pattern. Makes it very quick to iterate different animations.
The neat work-flow with it is how you can just type in a string like acgbef or aannzzaa and that becomes a flicker pattern. Makes it very quick to iterate different animations.
You got me curious so I looked into it.
I think it’s supposed to be size of the near plane.
This works for me:
Size = near * tan(fov / 360.0 * PI) * 2.0
I think it’s supposed to be size of the near plane.
This works for me:
Size = near * tan(fov / 360.0 * PI) * 2.0
November 26, 2025 at 6:41 AM
You got me curious so I looked into it.
I think it’s supposed to be size of the near plane.
This works for me:
Size = near * tan(fov / 360.0 * PI) * 2.0
I think it’s supposed to be size of the near plane.
This works for me:
Size = near * tan(fov / 360.0 * PI) * 2.0
A month ago I was describing some of my stuff as procedurally generated and in retrospect I think someone definitely interpreted me that way.
November 25, 2025 at 6:00 PM
A month ago I was describing some of my stuff as procedurally generated and in retrospect I think someone definitely interpreted me that way.
It’s like the Facebook fake ai generated accounts. Weren’t we supposed to be on there to hear what our actual friends were up to?
But these companies are willing to make fake people if they’re more “engaging” than real people.
But these companies are willing to make fake people if they’re more “engaging” than real people.
November 24, 2025 at 6:04 PM
It’s like the Facebook fake ai generated accounts. Weren’t we supposed to be on there to hear what our actual friends were up to?
But these companies are willing to make fake people if they’re more “engaging” than real people.
But these companies are willing to make fake people if they’re more “engaging” than real people.
Now, it’s Musk so of course he’s vastly overestimating whether gen AI can accurately anticipate what pixels and sounds you want to see and hear. But it’s also to even treat that as a good goal is a rejection of the value of truth and reality.
November 24, 2025 at 5:58 PM
Now, it’s Musk so of course he’s vastly overestimating whether gen AI can accurately anticipate what pixels and sounds you want to see and hear. But it’s also to even treat that as a good goal is a rejection of the value of truth and reality.