Do you need some VFX ๐พ for some juice ๐?
Preparing the examples for my talk at the #GodotCon2026 ๐
#godot #godotengine #indiedev #indiegamedev #gamedev #vfx #effects #actions
Posts by Beau Seymour ~ Sand Stack Dev
And I get there's a dropdown filter, I'm more thinking something more obvious to help popular as well as new
Yeah agreed, it'd be nice to have a few different feeds too, like new/emerging, recently updated, high like/download ratios (popular)
I mean.....why. What does this PROVE Google. What does it actually PROVE. Mans grandmas aren't going to provide valuable technical feedback. It's hoop jumping manufactured friction that achieves nothing for actual app quality while merking indie/solo devs time and sanity.
Urgh I am sadly reminded the Google Play Console exists for android development and suddenly finishing Sand Stack falls in my priority list again.
It's just in Google's DNA to make stuff convoluted isn't it -.-
Where's the PyTorch of the app store world.
That's How You Know It Means Business! lol
New Godot finally getting likes shown on asset store. Big!
Ah the classic 'its ok that she looks 8 but with makeup, because she's an [android, vampire, immortal, werewolf, has an aging disease and is actually 80] argument.
When was the last time I saw this, Star Ocean?
It's never not creepy -.-
Le Sigh. If you think Jesus had 'enemies' you fundamentally misunderstood the new testament.
And that's coming from an atheist.
Not sure yet why not ALL the game runs in a web browser yet, but if your weird and want to help design levels for me for no bank lemme know and I'll send a link sometime ^^.
Also still need more obvious connection status. Pillars lighting up isn't enough for complex levels I've found.
What's that, an actual video for my #GodotEngine puzzle game SandStack?!?
1. Finally made WASD/3 finger drag movement work right rendering the on screen joystick moot
2. Made right click delete work so I don't need to toggle the delete button when designing on PC
3. The editor runs in a web browser!
tells me a lot about your readers :S
As I've just fixed the bug where finishing the last level of SandStack caused it to crash... is it technically now a complete game?.... scary times!
Ghost in a mirror
Idk why, but found myself watching some of my own, now old, #GodotEngine videos. And you know what? I'm a little proud of this one, so thought I'd reshare!
I'm still happy with this scene and how it came out, and who knows, a beginner may find it interesting still!
www.youtube.com/watch?v=T1Er...
For a split second I thought you'd leaked some internal Godot 5 build ๐
Which then makes me wonder, reflected visibility must be a nightmare for LOD systems. Does Lumen just solve the need to frustum cull ever?
At a glance my guess is they're culling buildings too aggressively behind/to the side of the player which then causes its shadow to switch off, only for it to be unculled moments later requiring the GI update. Temporal delay then makes it even uglier.
He's back! Was a sad day when he 'retired' but kinda expected it wouldn't last ^^
Which is what retracing last legs of journey probably is. This is presumably rereading previously cached edge data, then potentially triggering a new routing request which could take seconds to process if it can't find a solution in a given amount of time.
When I say dynamic I'm more referring to responding to traffic live. 'The route I'm on is more congested than normal so I'll detour' is computationally nightmarish. There would still be scenarios where a vehicle needs to compute a new route, but you'd need to delay and queue it to avoid lag
I also have no kind of traffic light implementation yet, but at a basic level its just a phantom car holding a junction turn for a specific amount of time rotating. Actuated would be more interesting to assess inbound load per edge, but I'm aiming for 10k simultaneous so the lighter the better
I am not yet accounting for multilane roads, but its probably a pick_random() scenario when joining the road, then lookahead to reroute to another lane if the next edge is full, without leaving the 3 lane road. This feels like it would be separate from the overall routing, and kinda messy.
This should mean basically all heavy computation is pre-calculated across several frames and can be updated incrementally.
Obviously MultiMeshes for draw call batching and as long as I am storing stationery times per edge over a given avg time, I can visualise congestion too.
Whether this = 10k....?
... but existing vehicles would always keep trying their original route until the network topology is changed which:
5. Network Topology changes prompt all vehicles affected to re-request OD pair computation in a queue.
4. No Dynamic routing. I'd be suprised if Cities is doing anything dynamic, I think instead they spawn the vehicle, add its OD pairs to a queue to be calculated one time and cached so each vehicle pre-calculates it's route once on network change. New cars would calc ratruns if congestion this way...
3. Currently I'm casually just recalculating the whole network whenever I move something. At scale this won't fly, so need to better handle changes. This should be trivial though, as control points are hashed to their grid positions, so I literally have the func to do it already, just logic change.
2. Traffic can't be collision based. 10k raycast at a minimum would become nightmarish fast and doesn't even account for junctions. So cars need to register which edge they hold and previous cars need to look ahead to say nah, can't move forward yet. Junctions are messier, but in principle the same?
1. Vehicles can't be calculating orientations on the fly, so Pre-calculate per edge direction vectors on network change. Store in edges dict for fast lookups.