Project Workflow (Im all over the place)
polycounter lvl 8
So with quite a few of my projects I constantly find myself off the general route.
The workflow I would like to have is:
Block out
Highpoly
Lowpoly
Placement
However I keep getting muddled up with technicalities...
For example I am creating a town in UDK on a landscape, I want maximum control over how everything is placed and looks.
Is it best to work 90% of the time in your modelling package and model everything to eventually move it across to an engine?
Or is it more wise to build small pieces and glue it together in the engine?
The problem with the latter is optimisation, to get fewer drawcalls it's best to merge chunks together to fit into less drawcalls per export.
If I have already built it in an engine then things won't be optimised as much as they could be to if I built it completely in a 3D package and controlled the exports with optimisation in mind.
So I am finding myself constantly switching backwards and forth in a very clumsy fashion... It gets the job done but it's a mental pain in the ass thus taking far longer than it should, hit many creative blocks.
The workflow I would like to have is:
Block out
Highpoly
Lowpoly
Placement
However I keep getting muddled up with technicalities...
For example I am creating a town in UDK on a landscape, I want maximum control over how everything is placed and looks.
Is it best to work 90% of the time in your modelling package and model everything to eventually move it across to an engine?
Or is it more wise to build small pieces and glue it together in the engine?
The problem with the latter is optimisation, to get fewer drawcalls it's best to merge chunks together to fit into less drawcalls per export.
If I have already built it in an engine then things won't be optimised as much as they could be to if I built it completely in a 3D package and controlled the exports with optimisation in mind.
So I am finding myself constantly switching backwards and forth in a very clumsy fashion... It gets the job done but it's a mental pain in the ass thus taking far longer than it should, hit many creative blocks.
Replies
This village I am talking about will feature 10-20 unique buildings with quite a lot of duplications and modularity. However I would like vast variety in elevation to the village, I would also like dedicated paths, stairs, streams running through it and perhaps others that would seem far too mentally clumsy when deciding, Engine or modelling package.
To compare, Epic Citadel in UDK, the paths the stairs and nearly everything is slotted very perfectly into each other as if they were modelled directly next to each other, stitched together and then broken apart. How would that workflow be done? 90% modelling package?
If it's 90% modelling package then are you not limiting yourself to the final look of it?
Or do I need 2 versions that are 100% identical in both UDK and 3dsmax?
I know I am most likely over complicating things but it feels like a creative block most of the time when trying to get my head around the 'perfect' workflow (You always want the best workflow possible).
Sorry for double post and a huge post, hope someone can help me clear my head up!
You are correct that making large chunks reduces drawcalls. However it also wastes memory as the large chunks usually take up more memory space than the smaller pieces.
Also a large chunk won't ever get culled. So all the polygons are always being drawn which can lead to problems.
The trick is to pick chunks that are the right size for your game.
You don't mention what kind of game it is? mostly this is because a different camera view or different specifications determine where you want the most detail and what kind of tools you want to use.
epic has multiple pages talking about workflow availiable here;
http://udn.epicgames.com/Three/GettingStartedOverview.html
http://udn.epicgames.com/Three/GettingStartedLevels.html
Workflow isn't important. The end result is important. Obsessing over it will get you nowhere. There are many different studios that all use different workflows and still produce good games despite differences in the way they work.