Supersampling vs Virtual Resolution

[helmut.bressler]

Hello,

lets assume I have an output device whith a native resolution of 960x960.

What would be the difference, if I am rendering an image in 960x960 with supersampling 2 compared to setting the virtual resolution to 1920x1920 with supersampling 1 and relying on downsampling in the output device.

Which resolution would SSAO and other screen space effects use in each of those settings?

Im asking because we have a highly detailed scene, which will require some sort of supersampling and we want to decide which approach is better.

Many thanks

Helmut

 

[silent]

Hi Helmut,

In you specific case you will have absolutely identical results. In virtual resolution (unlike Supersampling) you can set independently width and height. Basically that's the only difference at this moment.

Post effects are rely on the actual render buffer resolution (in your case it would be 1920x1920). So if post-effects will be set to use half resolution they would be rendered in 960x960.

You can check render buffers used and their actual resolutions by typing render_show_textures 1 (or 2) in console.

Thanks!

[helmut.bressler]

Hello Silent,

thank you thats good to know.

I didn't think about show_render_textures . Very usefull. ( Usually I only look at if something is flickering or missing in a reflection...)

Btw, I am a bit surprised that I find 4 deferred lights textures, although I have only a single light source in the scene. The documentation says that there should be 2 textures, one for diffuse and one for specular lighting. So I wonder which are the other 2 deferred lights textures (which seem to be black in my scene).

Thanks, Helmut

[silent]

It's hard to tell without seeing the actual scene. You can disable lights and nodes and see what affects this behavior. Maybe it's some kind of voxel / env probe that left enabled.

I think also that not all the textures are being described in the documentation, so you may see some undocumented stuff here.