4
\$\begingroup\$

Does software used to create 3d assets (for game development specifically) have an impact on the efficiency of the program?

In other words, is it possible to reduce the operating footprint of an application merely by utilizing alternative development software during production of 3d assets? If you use two different applications to create a 3-dimensional image of a box, can one of them result in better performance if aspects of the image are identical?

I am attempting to get the information I need without causing unnecessary debate over specific software choice.

\$\endgroup\$
8
\$\begingroup\$

Indirectly, yes.

A 3d model is a collection of polygons*. The less polygons it has, the faster it can be rendered by a 3d engine. That means a 3d modeling program for game development should assist the artist in creating good looking 3d models with the fewest number of polygons possible. On the other hand, a software which focuses on use-cases where image quality is the primary goal and rendering time is just of secondary concern (like CGI effects for a movie) can be much more wasteful with the polygon-count.

The difference might not be apparent with a test-case as simple as a box. I would dare to say that any 3d modeling tool which uses more than 12 polygons to create a box-primitive (2 triangles to form each of the six rectangles) would be quite strange.

But it might be more apparent when it comes to approximating round shapes with polygons. You can not create a perfectly round shape with a polygon-based 3d model. When you want something round in a game, you need to approximate the shape with a large number of polygons. How many you need to make it appear round-ish enough depends on how large the object is going to appear on the players screen in the finished game. This is something the 3d modeling software can not anticipate. The artist has to make this decision during the design phase.

An inexperienced 3d modeler who just clicks on the "make sphere" tool might not realize that it causes the program to create a much higher number of polygons than necessary. An experienced 3d modeler might notice the problem, but might have a hard time to convince the program to use exactly the number of polygons they want and eventually just give up and have the program have its way to meet the deadline.

A good 3d modeling software for games might also provide tools to automatically reduce the polygon-count of a 3d model without notably affecting its appearance while another software might offer no such tool or only tools with worse results.

*) In practice, a 3d model is often a lot more than just polygons. Different kinds of textures, animation bones, etc.. But everything I wrote about polygons also applies to these aspects.

\$\endgroup\$
  • 1
    \$\begingroup\$ The higher you go on abstraction ladder, the higher the overhead can become when using inappropriate software+specialist. (In)efficient UV mapping, texturing, wasteful(less) skeletal animation, etc.. \$\endgroup\$ – Kromster says support Monica Jun 1 '14 at 7:47
  • 1
    \$\begingroup\$ Also consider the format used. If a model format allows the entire model to be drawn with a single draw call it's going to be faster than a model format that requires 1000 draw calls. This also applies to loading times; a model format that can be loaded directly into a vertex buffer will load faster than one that needs lots of parsing and intermediate representations. \$\endgroup\$ – Maximus Minimus Jun 2 '14 at 14:32
1
\$\begingroup\$

No

but that answer has a number of caveats, so I'm not really disagreeing with Phil, just looking at it from a different perspective.

In particular, the model must be identical in each case (eg. a 3d model from Maya and an identical model exported from 3ds Max are treated exactly the same by the game engine). However different modeling tools tend to nudge you toward different approaches to modeling, and thus a newbie who doesn't really know what he's doing is likely to create models that aren't technically identical, even though to the untrained eye they look the same.

For example, a low-pol model that simulates detail using a normal map will be much more efficient to render than a high-pol model with that same detail apparent in the polygons. The visual appearance may be the same however (indeed, the whole point of normal maps is to transfer detail from a detailed high-pol model onto an efficient low-pol model).

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.