rd-rend2 (old thread)

[Boothand]

Wonky glowing duck statues.

 

[Archangel35757]

@@Boothand - Are those T2 duck terminators?

[Boothand]

@@Boothand - Are those T2 duck terminators?

 

Cubemap duck terminators. It doesn't exactly map very convincingly.

[DT.]

So @@Xycaleth, what would be the estimation on when rend2 would be complete enough for a port to SP? I realise there would be 2 rend2's to keep updated then.

[mrwonko]

So @@Xycaleth, what would be the estimation on when rend2 would be complete enough for a port to SP? I realise there would be 2 rend2's to keep updated then.

I'll look into having universal renderers (i.e. working for both SP and MP) soon; should have results within the month (although they may just be "not possible", if we're unlucky). Assuming it works out, it will probably be fairly easy for Xyc to update rend2 so it works in SP.

[DT.]

That'd be great.

[Xycaleth]

So @@Xycaleth, what would be the estimation on when rend2 would be complete enough for a port to SP? I realise there would be 2 rend2's to keep updated then.

It's not really a question of when, it's more the fact that I don't particularly don't want to do twice as much work :/ If mrwonko can unify the SP and MP renderers then that'd be great and I'd happily get rend2 working so SP.

In other news, I've seen issues with using misc_models and normals. It's like they're aren't rotated properly by q3map2..

 

Anyhow, I've been working on deferred shading in rend2. Progress is going at a steady pace. Here's mp/duel1 combined with rainbow road from Mario:

 

 

[DT.]

I don't remember seeing anything odd when I was testing out misc model, I'll check again.

 

So what has deferred shading done to the pic you posted?

[Xycaleth]

I don't remember seeing anything odd when I was testing out misc model, I'll check again.

 

So what has deferred shading done to the pic you posted?

It's broken the lighting on misc_models and added the rainbow effect around the map When it's done, it'll allow me to optimize the rendering further.

[Xycaleth]

This is just some debugging visualisations but I think it looks pretty cool When a map gets compiled, a lightmap is produced but also a light grid. This is basically how much light there is at every 3d point in the world (evenly spaced out).

 

I've made a visualisation of the ambient light at each of these points:

 

The same effect can be used for fog and volumetric effects so I expect I'll be putting this same technique to good use somewhere else

[Didz]

Are those light values in linear-space but being outputted as gamma 2.2 RGB values? It looks a bit darker than I expected

[Xycaleth]

There's no gamma correction in JKA

[Didz]

So for a room with light value 0.5 everywhere, JKA just multiplies the raw RGB value from the texture with 0.5 and puts that on the screen? I guess it's only noticeable when you know the problem exists!

[Guest KENNITHH]

just a quick question, would rend2 eventually be capable of doing this?

 

 

 

or too early to tell?

[ensiform]

In what way?  The BSP is not changing simply because of the renderer being used.

[Archangel35757]

I think he may be hinting at another data structure for large outdoor scenes...

 

http://stackoverflow.com/questions/99796/when-to-use-binary-space-partitioning-quadtree-octree

[eezstreet]

just a quick question, would rend2 eventually be capable of doing this?

 

&&0

 

 

or too early to tell?

yes, rend2 will be able to render polygons

[Archangel35757]

Again... I think he's really asking for an Octree Spatial Partitioned (OSP) data structure for large outdoor maps.

 

http://www.gamedev.net/topic/335852-octtree-vs-bsp/

[Guest KENNITHH]

no I mean : the complexity of maps... the maps we know of can't handle this much models, definitly not in the open as FPS will be really low.

My question, which I shouldve asked better is if rend2 can actually handle all these models in the open better than the original renderer, same for a bigger terrain

[DT.]

I guess you might get an fps improvement due to models being done on the GPU rather than CPU, but pretty sure it still comes down to the map format. Plus, you can only go so big with current format.

[Spaghetti]

Here's something that's been bugging me so I thought bringing it up might be a good idea. Has any consideration been made to making the transition between the legacy renderer and rend2 easier? Updating shaders is my concern.

 

For those who do not know, I am part of the Movie Battles II development team. Internally the topic of rend2 has come up many times. The inevitable conclusion though is that we have too much legacy material to convert manually. Between our maps, models, and UI, our shader count numbers over 600 discrete files (or 3 MBs of text). We will never have the resources to manually rewrite everything to the bare minimum point of looking the same as before.

 

The only solutions I can think of would be to either have a hybrid render mode which could read and display the old shaders while still allowing the use of newer capabilities and optimization, or an external tool which could automatically parse old shader files and convert them into equivalent versions in the new shader format. I have no idea how technically feasible either would be though.

[DT.]

Rend2 has a feature where you can have an .mtr file (exact same as .shader, just different extension) that it will load first. For example:

 

Vanilla renderer = kyle.shader

Rend2 renderer = kyle.mtr

[ensiform]

Enough with the octree shit archangel...  We get it, you want something else besides just BSP or along with.  Stop pushing that other people are making slight hints about it too.

[Archangel35757]

Enough with the octree shit archangel...  We get it, you want something else besides just BSP or along with.  Stop pushing that other people are making slight hints about it too.

Oh...

Consider

This

Request

Efficacious

Ensiform

[DT.]

Well there's my lol for the morning.

 

@@Xycaleth How's progress? I hear a certain large project for JKA might make good use of rend2 when it's stable and finished.