Mikko Mononen

Coder and woodworker. Author of Recast&Detour and NanoVG. Previously at Epic Games, Unity, Tinkercad, Crytek, demoscene.

[Mikko Mononen @MikkoMononen]

Looks like bluesky is busy enough for daily doom scrolling. See you there!

[Mikko Mononen @MikkoMononen]

@rsms Bevy UI is the most comprehensive i've seen. https://t.co/Qgyi4Cf2qI

[Mikko Mononen @MikkoMononen]

@PetraNyqvist Jos on tehty paperille, niin skannataan (skannereita löytyy kaiken kokoisia). Tosi moni kuvittaja piirtää nykyään digitaalisesti, suoraan koneelle (sellanen erikoinen kosketusnäyttö joka toimii kynällä).

[Mikko Mononen @MikkoMononen]

@azmreece @whitebox_sys The issue is that oop turned the patterns into "religion". This article summarizes it quite well: https://t.co/j2LM71aQvw

[Mikko Mononen @MikkoMononen]

@anorangeduck When I was young and the games were made from scratch I got convinced that the only way to go is fixed update time + interpolate.

[Mikko Mononen @MikkoMononen]

@anorangeduck The more you dig into to it, the more complicated it gets :) There was blog post few years ago on the topic, which dives into the GPU details too: https://t.co/v1Eksj5zyD

[Mikko Mononen @MikkoMononen]

@amathlog Was pleasure working with you, Adrien!

[Mikko Mononen @MikkoMononen]

Last day at Epic. A whole bunch of cool projects over the last 5 years, and lots of really nice people to work with. Time to rest, maybe explore stupid ideas, and one day figure out what to do next.

[Mikko Mononen @MikkoMononen]

I did not know that ECS was a big thing in ActionScript world in 2010+. Here's a neat article on ECS and FSM from that era: https://t.co/pP6kMDV1Zc

[Mikko Mononen @MikkoMononen]

I know that there's a strong desire to have just one system (guilty of that myself too). Separating what and how just gets rid of a lot of shoehorning.

Relevant recent presentation:
https://t.co/GIyqh2vATE

@Bobby_Anguelov 's original talk:
https://t.co/Qcxu3ADLBs

[Mikko Mononen @MikkoMononen]

Unity Behavior Graph (behavior tree) supports Join nodes. I wonder why not just go for full on Activity Diagrams (https://t.co/Cz3uQS9Jja) for activities/abilities plus states or utility selectors higher level selection?

[Mikko Mononen @MikkoMononen]

@SebAaltonen Does the low end android have enough gpu bandwidth to do the anim eval on gpu (vat)? There are a lot of different variations of the method. You could do a lot with 2 bone weights + 2 anim blend, but that can end up being quite a few texture reads per vertex.

[Mikko Mononen @MikkoMononen]

@MikkoKuitunen3 Operators, modules, nodes... the particle system approach does not seem to have a consistent name for the "thing" that mutates data.

Rel DB is indeed one way to think about ECS.

[Mikko Mononen @MikkoMononen]

I have been researching authoring UI for ECS. My conclusion is that particle systems (Niagara, Houdini, etc) are the most successful ones.

Interestingly, they turn the setup upside down: the user defines behavior (systems), which then defines components.

[Mikko Mononen @MikkoMononen]

@rms80 I think the hacking will lead to jump-point-search territory (which includes visibility). I like the general idea, though, so maybe worth hacking so more.

In my proto I'm already using distance to previous node if possible, it helps some, but not quite enough.

[Mikko Mononen @MikkoMononen]

I wish I had better understanding of mathematical notation.

I have been looking how to apply the fast marching method to something else than grids, and I think I found the right paper, but it's so thick in math notation that I fail to see the algorithm.

[Mikko Mononen @MikkoMononen]

@rms80 I think I finally got the vector accumulation thing. "Practical anisotropic geodesy", Campen, Heistermann & Kobbelt, had nice explanation picture and also alternative (Short-Term Vector Dijkstra) that takes obstacles (more) into account.

[Mikko Mononen @MikkoMononen]

@Clarkbai223 Thank you!

[Mikko Mononen @MikkoMononen]

@rms80 The reason i was drawn to that paper was that they do not require triangles (as far as i got it). The geodesic FFM papers out there seem to all assume that.

[Mikko Mononen @MikkoMononen]

@rms80 I think i'm missing what the "simple multilinear interpolation" bit means (and some other obvious stuff around it). I have to check that reference. I'm kinda walking backwards from a 2D grid FFM code, i've been messing with.

[Mikko Mononen @MikkoMononen]

@rms80 Thanks!