Radar LOS

If it can't be easily blocked... won't pass straight through the things you want to detect?

So the planets are not just spheres with a map put on top? Then I have not idea what will happen.

Unobtainium blocks it and robots need to be made out of it for robot reasons. Stealth then becomes similar to today's passive/active stealth systems.

QED robots solve all robot problems.

I just wanted to point out that the LOS for radar thing is a pretty simple one performance wise. You don't have to raytrace everywhere from the radar, only to enemy units. A single raycast to all the enemy units is considerably cheaper performance-wise.

But impossible with over the horizon radar as it requires refraction and that means much more work then a single collision check. You should have understood by now, "real" LoS is insufficient when talking about spherical or complex shaped maps as visual is much more restricted than radar should be and it is not possible either to create an satisfying effect with acceptable computational overhead either.

Just accept that LoS is a broken concept with such complex maps as PA is going to offer.

wtf are you on about?

I heard that strategic zoom was supposed to be impossible to implement due to how taxing it was on resources, once long ago. And yet we have Supcom.

TA was impossible too.

In reality the tradeoffs you have to make for something like Strat Zoom are pretty crazy. The tech we build on keeps getting better and better though. For example I'll be able to move way more of the rendering work to the GPU these days than I could before.

Will you be able to dynamically spread the rendering work to cpu or gpu depending on the configuration the game is running on ? I mean let's say i'm lucky enough to have a 8 cores machine but my graphic card is not a high end one, will you be able to spread much more rendering stuff to the cpus instead of the graphic card ? Does this means there are minimum requirements for the graphic card in order to make the game run as smooth as possible (and especially regarding strat zoom ?). Another question regarding resources consumption : Any chance to link a second PC (using a switch) to the main one running planetary annihilation, and having the additional PC running planetary annihilation core engine in order to help the first PC ? I mean there are softwares like raytracing that can spread stuff accross multiple nodes. Is that something that would be possible in order to be able to scale much more than on a single PC ?

You can't shift work from the GPU to the CPU, thats not possible. GPU just became much more flexible and capable of handling many tasks independently from the CPU which wasn't possible 5 years ago (basic animations, complex shaders). If you GPU is to weak, then you won't be able to achieve the same level of detail as someone with an high end GPU, no matter how much CPU power you got spare.

Having 2 PCs running is possible, it's just a LAN-game where you have the game server running on the one PC (that one doesn't require a GPU) and the actual game client runs on your main PC, relieving the stress on the CPU but still requiring a strong GPU. Of course that will make only sense if your dedicated server is much stronger in terms of CPU power then your desktop PC, you won't gain much otherwise.


Now again a question at Uber: Would it be possible to adept the level of detail based on the current framerate rather then on a predefined setting or visual ranges? I'm talking about model resolution, number of particles in particle systems as well as the accuracy of client side physic simulations.

I'm currently thinking about late game when you have armies of an insane size or battles with them. People should be less aware about the lack of high resolution models or particle simulations when watching a full army than they would have been when just micromanaging 5-10 units, even when playing on the same zoom level.
This could also solve the problem of the game starting to lag at a certain point since most people will not tune their settings to the demands of a worst case scenario but rather to medium load at most.

Only settings which make sense to be fixed are AA, AF and texture resolution. Latter one because you don't gain much performance by reducing texture resolution since even most modern mobile GPUs have at least 1GB of memory, you only loose details, much faster and much more obvious than when you are just using low resolution models.

Actually, i was just asking myself in which degree of magnitude neutrino engine can scale :
If the engine perfectly scale, then we could imagine a 4 node cluster, one node being the master, spreading the workload across the 3 slaves. I guess it's unrealistic, but if Neutrino says the more horsepower you have, the bigger game you will be able to play, then at some point the strongest single PC might not be enough.

Just making an engine scale with a multiprocessor system (shared memory) and making it scale with a master / slave architecture are two completely different tasks. Synchronization overhead can easily kill the performance on the latter one.

So, no, you just won't be able to spread the server over multiple systems, at least not for a single skirmish.
What you could do, is using multiple clients on the same server together with KVM switch to get even more view ports in parallel with full (pure visual!!!) debris simulation, smoke and whatever other breakables and particle systems you could waste your GPU's computing power on.