Ogre Forums

I'm curious to find out how the other artists that visit these boards view the creation of models etc. for modern game engines .

Undoubtedly the finnished results are quite spectacular but I find that the nature of creating these assets has become increasingly labour intensive and it's having a very negative impact on my productivity .


I first started making mods and maps for games as a hobby about 10 years ago and soon discovered that I enjoyed making this stuff more than I enjoyed playing the games themselves. Now I find myself working as the only artist on a small indie project and I'm really struggling to cope with the workload as everything seems to be so much more involved with high polygon models, shaders, normal mapping and the rest of it . In the past I was able to create a couple of models per week (more if I cut corners) now I find it takes me in excess of 2 weeks to create just one game ready model. I find the work is more demanding and I find I tend to "burn out" quite rapidly, it's just not as enjoyable anymore, although I'm quite proud of the results I have achieved I do have to ask myself if it's really worth it.


I wonder if any other artists here can identify with this, or is it just me ?

I am not an artist, but you are completely true.

I read than it takes 3 weeks to model a complete guy in a modern FPS (because they model even the guts and their physics..).

On the contrary, when I was youger I could do great things with duke nukem Build editor only, and it was done in few hours (including basic programmation!).

Nowadays, more and more details can be visible. That is one of the reason why there is so much more to do than before, from an artist point of view.
Many techniques are also available, and that can make things always harder.

One solution would be to encapsulate everything with higher levels of semantics. Example : the soulcalibur IV character modeler. But programmation is violent (if you includes lods), and it needs a lot of base graphical data.

Another is to be able to make procedural data. This is widely used for games, but not for simulation.

For static scene, I think that 3D scanning can be a good solution to accelerate production. There are lots of algorithms that can help. 3D virtual sculpture (with force feedback) is said to be a great asset for high poly modeling. But then you still need to declare the different kinds of material (for the physics, and/or the visual).

For human animations, lots of programmation (physic+IA) on top of mocap system can help to do the job done, but human action is still a lot needed. Puppeters could also be an intermediate solution.

My point of view is that new technology doesn't have to impact your productivity negativly at all. E.g. you can spend less time on painting colour maps (no need to draw lighting by hand) when you're using normal maps. You don't have to keep poly count that much in mind anymore. Which again enables a modelling style that lends itself more to re-using parts and working faster. I'd say my productivity has increased.

Of course you'll spend a lot more time on a single object or character if you want them to have the same level of detail as say Unreal Tournament 3 or Crysis. But this isn't because of technology, it is because you want to compare yourself to titles that are created by 50+ artists working for several years. Which doesn't make sense if you are doing indie games

In essence: people should get it out of their head that the advance of technology means that one has to create art with ever increasing complexity. There is an infinite number of interesting and efficient-to-create art styles to explore - now that we aren't limited anymore to just textured polygons.

it really depends in some ways it can be easier with modern tech. You don't draw the light shadow and bumps etc. You just draw local colour map, grey vertical height, and do a quick specular which can be made easily with a filter. And have nice next gen models that look cool.

It's when you work low poly, subdivide to high poly, rework, texture and then bake to low poly with normal map and then tweak the normal map and clean it up. Thats when things get annoyingly slow.

But so long as you have access to good shaders its for the most part easier. Especially with a good deferred rendering setup that makes lighting and shadows a nobrainer.

Evak wrote:But so long as you have access to good shaders its for the most part easier. Especially with a good deferred rendering setup that makes lighting and shadows a nobrainer.

What do you mean with good deferred rendering setup? Do you mean the
deferred shading techniques? Subdividing calculations among several buffer etc?
How is this more efficent in terms of content creation?

I think my biggest handicap is my lack of familiarity or experience with all these new techniques, I barely understand shaders at all , my programming partner has a slightly better knowledge of these things but we still haven't been able to get things like normal maps working correctly.


my models usually start life as a high polygon mesh I then build my low polygon model around it before feeding everything into ultimapper (I'm an XSI user) to generate ambient occlusion and normal maps, the occlusion map I use as the basis for my diffuse texture.

Admittedly texturing is a great deal easier with all the shadow detail baked in but this is offset somewhat by the fact that I'm now using more and higher resolution textures for my work, so that instead of using a single 512 x 512 texture I'm now using two 1024 x1024 textures per model, the volume of work has increased even though the work itself is easier.