A Question About Repeating Models

[Rob_T]

From a rendering perspective, is it better to use many instances of the same model or is it better to individually model each instance with it's own name and file?

 

For example, in my current project I have lots of buildings. Buildings have lots of windows. Would I be better off modeling the buildings with the windows built into the building as part of the model. Or would I be better off making several types of windows and then importing the building and window models into my choreography and then inserting the windows into the buildings by creating many instances of the same windows until the buildings are complete?

 

I ask because this is occurring a lot on my set. I have a bunch of balconies on my buildings as well as trees.

 

[robcat2075]

I'd suggest doing a small test each way and see if there is a difference

[Rodney]

The experts will have to weigh in on how using instancing versus additional models effect render time but my opinion...

 

You should see a significant enough increase when using references/shortcuts to the same model versus separate models to make it worth your while.

Don't forget also that once you've got all these instanced models into the Chor you can Save them out as a new model incorporating all of the referenced resources.

 

Where I would expect you to see the biggest savings in workflow is when editing your models which is likely to happen at the most inopportune moment. If you've used instancing... when you change one model... all of the instances will automagically change too.

 

(Tip: If you need to change one of the instances for the sake of getting a perfect image you can adjust them separately by adjusting the geometry in Muscle Mode (F7). If you want to adjust them all at the same time while in the Chor go into Modeling Mode (F5).

 

I'd suggest trying not to think here in either/or terms.

You can get the most bang for the buck by using both.

[HomeSlice]

It depends on your visual style and how close the camera gets to the windows. In most cases, you can just slap bump and color maps on the buildings for windows. If the camera gets close to any of the buildings, you can model a few detailed "foreground buildings" with the window geometry as part of the building model. Even in the "foreground buildings", you only have to add window geometry to parts of the building the camera gets pretty close to.

[Rob_T]

It depends on your visual style and how close the camera gets to the windows. In most cases, you can just slap bump and color maps on the buildings for windows. If the camera gets close to any of the buildings, you can model a few detailed "foreground buildings" with the window geometry as part of the building model. Even in the "foreground buildings", you only have to add window geometry to parts of the building the camera gets pretty close to.

 

I love how you say stuff like that like it would be so simple. I hope that stuff gets so easy for me that I can be blase about it myself someday... sooner the better.

 

But just slapping maps, of any kind, I don't even know what that means. I've never mapped anything, bumped, slapped or what have you, to my knowledge. Was that even part of Tao?

 

I do understand what you mean about geometry though. I'm not making windows on any of my buildings right now. My intention is to either decal them with a window picture if they are far away or insert a window model into the building in strategic places if the camera gets close.

 

My question was, as far as rendering goes, if it is better for me to repeat those window models in the choreography as many instances of the same model, or if I should insert them into the building model in the modeling window and then import the building into the choreography (with the windows already in place). If one was better for rendering over the other.

 

My thought process was that perhaps the program may have an easier time rendering the same window model over and over again rather than having to consider each window as part of a larger model and therefore distinct on its own from every other model.

 

I think Robcat is probably right I would have to test to find out. Unfortuntely I haven't a lot of time for testing right now. I'm going to have to make a choice and stick with it. For better or worse.

 

Rodney, your info about altering the models in muscle mode has already been a huge help. Thanks a ton for that. Sometimes when I load the project the geomentry on the buildings I altered already is wacky and all over the place but as soon as I select them and click into modeling mode they snap back. Wierd but I can live with it.

 

Thanks for the help.

[petokosun]

considering your scene, I think a good question should be how would you configure your camera so that it will not render the whole scene each time you render a camera view, that is rendering only the area in the camera view. Any help on this

[Rob_T]

considering your scene, I think a good question should be how would you configure your camera so that it will not render the whole scene each time you render a camera view, that is rendering only the area in the camera view. Any help on this

 

Actually, that's a great point. It never occurred to me that the program would render things that cannot be seen by the camera. I wonder how that works. <_>

[Rodney]

But just slapping maps, of any kind, I don't even know what that means. I've never mapped anything, bumped, slapped or what have you, to my knowledge. Was that even part of Tao?

 

Hehe. He's talking about Decals here.

Instead of having dozens upons dozens of giraffes in your scene... try decaled buildings instead.

 

Seriously though, think about it.

If you had to model in every spot on a giraffe's body you might want to map, slap and bump them on too.

[Rodney]

Actually, that's a great point. It never occurred to me that the program would render things that cannot be seen by the camera. I wonder how that works.

 

Where possible you should consider rendering your larger scene and using that as a background.

Then you can focus on the detail of the 'hero models', the ones that will need to have greater detail because they are up close to the camera.

 

A:M does a pretty good job of determining what in a scene to render but there are things that can complicate the process.

For instance, if you have a lot of reflectivity on your objects.

When in doubt its a good idea to hide or inactivate unnecessary elements in a shot.

 

Interestingly, this brings us back to the subject of instancing.

You may find it easier to build your set in separate sections (saved as different models), combine what is necessary in a sequence into one Choreography and then as you need them turn them On/Off.

 

In the TWO and SO movies pretty much everything needed in a sequence or shot is saved in separate Choreographies.

You can have as many as you want.