Need some feedback

[nikoli]

I'm making a 'warehouse' room and need some feedback on what most peoples rigs can run without issues. Granted it still needs a lot of work to be usable, but I'd like to make it semi-realistic. 

I plan on making it a normal room with various hook objects placed in it, initially disabled, so it loads faster. Currently the room is about 20MB empty, mostly from the racks and barrels which are probably more high-poly than the need to be. A room of this size for my rig takes about 30 seconds to load but I have over well 100GB of various objects I'd like to include to give it a 'lived in' look. Anyone know what a 'potato' rig can handle?

The screenshot is just the Blender rendering, I haven't even gotten to the point of converting it for the game yet.

[aardy]

37 minutes ago, nikoli said:

I'm making a 'warehouse' room and need some feedback on what most peoples rigs can run without issues. Granted it still needs a lot of work to be usable, but I'd like to make it semi-realistic. 

I plan on making it a normal room with various hook objects placed in it, initially disabled, so it loads faster. Currently the room is about 20MB empty, mostly from the racks and barrels which are probably more high-poly than the need to be. A room of this size for my rig takes about 30 seconds to load but I have over well 100MB of various objects I'd like to include to give it a 'lived in' look. Anyone know what a 'potato' rig can handle?

The screenshot is just the Blender rendering, I haven't even gotten to the point of converting it for the game yet.

I don't use Hook but if it's any help I'd be happy to test the basic room for you on my potato.

Might be able to give you some idea of any initial content over-loading issues.

 

[Oz70NYC]

I'm willing to give you a hand as a tester also. 

[hfg2]

Depending on how those shelves are created, if all are the same you could try to make instances of them, maybe it will speed up loading. Same thing for the wooden pallets. And for the lights and for the air recirculation tube.

Just wondering how you ended up with such a big number of vertices (I assume this is the case), looking at your scene looks something pretty basic.

If some meshes are not-lowpoly friendly, then you could try to decimate some of the faces, or replace them with something more usable.

 

[x17]

Many assets meant for static sceners and posers (like DAZ) have significant number of heavily unoptimized meshes ... some can reach insane levels of complexity thats basically "overkill" ( for example, I have seen props with 100s of thousands of vertices, as well as one blanket over half million).

Such assets should always be decimated - especially if theyre completely static and intention is not to utilize armature, weighting or morphs with them. So, if theyre room objects in our case, your options are definitely to decimate them (watch out for UVmap destruction in that case, and experiment with multiple types of decimation).

Especially in this case, the objects are part of the warehouse scenery... there really isnt vegetation or some complex oval objects and similar. All is fairly "blocky" and you can use a lot of that to your advantage to optimize the scene. Depending on the size and number of objects within the room, I would say optimal ranges are in 600-700k vertices max total, but could reach over 1 million.

IMO, many room scenes are such that they can be comfortably decimated and for difference between hi-poly and lower-poly scene to be barely recognizable. Many urban, SF and similar scenes and rooms... the only time when we really need to eye for upper limit regarding vertices is when we have extensive areas of vegetation in the scene.

[HDiddy]

4 hours ago, hfg2 said:

Depending on how those shelves are created, if all are the same you could try to make instances of them, maybe it will speed up loading. Same thing for the wooden pallets. And for the lights and for the air recirculation tube.

Just wondering how you ended up with such a big number of vertices (I assume this is the case), looking at your scene looks something pretty basic.

If some meshes are not-lowpoly friendly, then you could try to decimate some of the faces, or replace them with something more usable.

 

Instances?

Can you explain this concept more.  Are you basically saying making copies of things like chairs books and other items from one single mesh?  Does this help with limiting the poly count?  I remember I saw something like this in the RKRC but didn't get too deep into it.

[hfg2]

Depends where.

Daz has the concept of instances.

Blender can link same data for objects.

As for TK17 this is still an area where I'm a noob, but apparently, this game works with scenes, and in a scene you can have 0, 1 or more mesh objects.

For the sake of discussion, let's stick with 1 mesh object in a scene. To keep things simple let's say we have a toy. In fact you can make a toy using CollaTkane, have everything done as toy addon, but delete every file from there and keep only the scene file.

 

Now back to our room, more exactly its Ac... file.

Add something like

AppImportScene . {

        .NodeName "TestToy1";
        .ParentPath "/RoomRoot";
        .SceneFile "Shared/Item/TestToy1";
    };

And your mesh Toy will be added to the room.

Raistin Kane had some tutorials explaining this concept as well.

Now if you add a line, not sure how efficient and how helpful is this, in Daz and Blender it can help with performance, but in TK17 can't say yet. There are some usages why I prefer this approach, but for most people just making a room with all objects inside is way easier.

 

A ton of people goes the wrong way and over-abuse of toys, like making cars or wardrobe or beds as toys and loading them as a toy. The method described above should correct this problem, a bit of lua scripting can do miracles, certain thing is if you have a car to load it in a room is better to load it in other ways than as a toy.