Started looking at this.

I tried stripping out has_agents from this function.

https://github.com/FLAMEGPU/FLAMEGPU2/blob/c5007d284da9d8efbf0e937bb02034f571a7b0ef/src/flamegpu/visualiser/ModelVis.cpp#L37

But that leads to the visualiser being stuck at step 0 (with the splash screen open) when the model begins with no agents, regardless of pause state or beginPaused setting.

Debugging shows it's stuck waiting for vis to confirm it's done the initial buffer allocation.

            while (!visualiser->buffersReady()) {
                // Do nothing, just spin until ready
                std::this_thread::yield();
            }

Not too sure best way to handle this. Either strip out this check (or rather replace it with something that merely check's vis init flow is complete). Or give buffers a default allocation so they are ready assuming initial agent account is <= default.

I attempted this fix by:

• Init required size ~1024
• change updateBuffers() to only increase requiredSize, not outright overwrite it,
• change splash screen close logic to not check for agent allocation

This now pauses with splash screen closed at step 0, but agents are not visible in the paused frame if they are created before the sim.

I think it would be easier to change the beginPaused logic so it always pauses at the first step with agents.

Would this be acceptable?

(I think it would just require has_agents to be calculated differently, and the step count checks in the above ModelVis fn to be removed).