OK, another Newbie question. I notice that every mapper I've read tutorials for has a different take of this*... but is it better to use wide brushes for large panels, or to break those brushes down?

Would it better to have one brush 512x128, or two 256x128 brushes, for a wall, for example? Is it better, as a rule of thumb, to use smaller brushes, or better to merge small brushes into larger brushes when possible.

File this under "don't optomize to early"...

• Like the "build rooms out of brushes" and "build room out of a brush and hollow" holy war :D

(You know what would be neat (haven't seen something like this), but a Wiki page of "Words of Wisdom." Little things, like brush sizes, just little uncatagorized nuggets of truth about mapping...)

More brushes mean more work on the client which means more lag. Fewer is better.

If you're worries about optimizing, just do what I do. Hints and occluders.

I'm not sure if there is a newer number (and if there is I'd assume it is bigger) but hammer/source can handle 8192 brushes.

If you want to add in brushes for geometric detail, to allow more texture variation or pretty much anything else - do it.

Yeah, the less brushes the better. If you have a single, uninterrupted wall, there is zero reason to cut it up into smaller brushes.

That makes complete sense, but when I take apart Valve's Portal levels, I notice that every wall is broken down into little sections.

So the Valve mappers seem to have another priority in mind. But I have no idea how more wall pieces could possibly be a good thing.

The walls are made with many brushes so that they can apply lots of texture variation.

If you apply the same texture across a large surface, it shows.

Also, remember that if you are decompiling a map, that vmex does lots of retarded stuff. A decomipled vmf is not the same thing as the original vmf.

I was also wondering if the big brush/small brush different might be the epic battle between polygon drawing rate and texture-filling rate. I.E., you can clip out more triangles with smaller brushes, but wider shots have more triangles and textures?

So, basically, the answer is, unless you have a REALLY good reason for multiple brushes, it's better to use a big brush. Then, at the end, if you notice that the big brush is causing troubles, you can split it later, or use other hammer tricks to make the level work.

(Although having done 3D before, I'm struck by how fast it is to build a simple level in Hammer, and just how few triangles actually come out of the level as built...)

The larger and more complex your level gets, the more you will have to jump through optimumization hoops. It is a process that a rare few have mastered. There may be some instances where clipping brushes to create more is cheaper to render than the engine doing a texture fill, but don't expect to run into it any time soon.

This is just presumption on my part, but wouldn't the BSP generator split faces anyway?

msleeper wrote:
The larger and more complex your level gets, the more you will have to jump through optimumization hoops. It is a process that a rare few have mastered. There may be some instances where clipping brushes to create more is cheaper to render than the engine doing a texture fill, but don't expect to run into it any time soon.

Dude. Just use hints to cut off smart vis leaves. Using this wisely will result in an optimized rendering process.

Use occluders. These hide any entities behind them, and are a great way to disable any entities in past chambers.

Using these two correctly will result in better fps and r_speeds.

Overuseage of hint brushes can cause more harm than good, so I'm not going to recommend it to someone new.

Occluders stop the rendering of models, and only models that are entirely blocked up by the occluder. If you are making a prop-heavy map, these can be a lifesaver.

Hint brushes are nice, but areaportals reign fucking supreme.