]]>
]]>
Statistics: Posted by TattieBoJangle — 06 Dec 2015, 23:31
]]>
Statistics: Posted by TattieBoJangle — 06 Dec 2015, 21:49
]]>
Statistics: Posted by SolarPortal — 06 Dec 2015, 21:45
]]>
Statistics: Posted by TattieBoJangle — 06 Dec 2015, 19:38
]]>
Anyway when you scale the point light the widget is also scaled to very small or zero size so therefore not visible. Useful if they are too big/small in the scene.
Statistics: Posted by StarFire — 04 Dec 2015, 21:32
]]>
]]>
]]>
]]>
]]>
]]>
]]>
Culling would help here..
Statistics: Posted by epsilonion — 04 Dec 2015, 13:12
]]>
Each material has 2-3 passes and one of those passes is a light pass. This means if you have 1000 meshes with a light pass, the lighting has too be calculated for each material 1000x over in each frame. When no light is in the scene, this process is stopped and thus your framerate skyrockets.
In the case of deferred rendering, lighting is calculated as a post FX which means it only has too be calculated for each pixel by your resolution. so 1920 x 1080. This is why deferred is slower frames to begin with, but doesn't drop much as you add more lights as the same calculations happen regardless.
In your scenes case, if you used 2 directional lights, then the same processing occurs again 1000x more. The scene will speed up again when all the other meshes you cannot see are culled off reducing the number of meshes the directional light has to iterate over. I am guessing if you duplicated your entire building over again that you would have many many meshes now lol
Spot lights and Point lights have an attenuated range and when the objects are out of the range, they stop processing the other meshes as it is only the materials receiving light from that particular light that renders with that light. Whereas a directional light is an unlimited distance light meaning everything is affected.
Older ways of handling these sorts of levels, would be artists placing the level into a modelling application and reduce the amount of materials per mesh/per level. Inside the modelling application, they would also merge meshes together into larger chunks as in one mesh too reduce the number of batch counts and draw calls of the level. Too this day, many of the techniques are still in use for full AAA games, except now they have technology like simplygon which can do the lodding and merging for them.
We understand you don't want to have too do this and our plans with skyline is too make this possible for you without the extra effort, but these are complete tools that can take full development lifecycles to make lmao
Anyway, just thought i would mention that simplygon has a free online tool which uses the cloud and allows you to handle a certain amount of mesh data a month: . Check it out as it may help you.
All the plans for performance and lighting upgrades are already being discussed about and are in motion of being upgraded or integrated into the engine. We are already checking out an updated renderer which is why we are not wanting to focus on the current shader too much just in case we change, Lightmapping, batching and occlusion culling is also in motion, so we hear your concerns and feature requests and we are doing our best to sort things out. But as you know, these things take time but we will eventually deliver.
Thanks for your input m8
Statistics: Posted by SolarPortal — 04 Dec 2015, 12:21
]]>