Practical time frame for completion of a project

Game Engine Game Engine Support and Discussion
1

Last week I started learning how to use Blender. My usual job is architecture but I’ve been tasked with using a game engine to build an environment that our client can walk around in and see the project in it’s completed state long before ground is broken.

My project is a fairly simple one but large. It’s slightly under 100,000 square feet with most of it open space. There are built in fixtures and equipment in most of the building. A large portion of the equipment is built of hundreds of small, highly reflective metal rods. In addition there are roughly 1200 light fixtures.

If I have any assistance in this it will be importing 3D models built in another program. All the assembly of the models, rigging the lighting and applying the materials will be done by me. Thankfully most of the equipment has already been modeled so I will be doing very little actual manipulation inside of Blender.

The rendering will be handled by utilizing roughly a dozen computers on our local network. Unfortunately only one of these computers will be dedicated to rendering, the rest are workstations being used from 9-5 M-F. The quality of the finished product does not need to be photorealistic but it should be as close as practical.

The only physics involved are doors need to open when clicked. The only triggers are light switches.

My question then is this: What would be a realistic time frame for completion of this project? OK, now that you’ve had a quick laugh at my expense, go clean up your monitor and keyboard and then a serious answer would be appreciated. I know this is a big project for a single noob to be attempting but the boss seems to think I’m some sort of genius so I must get it done.

2

I can promise that your boss doesn’t consider you to be “some sort of genius”.

This is what he’s thinking: “I’m a better man than this petty drone, because my salary is higher, and I’m his boss. Now, instead of hiring the appropriate number of people to actually build this project, I’m going to push it all on this one sap, and have him figure it out. If he can do it, great, I was the genius boss who was able to recognize one of his own. If not, I have someone to blame.”

So, now that we’ve cleared that up:

It all depends on how much you already know about blender, and how much you have to do with blender.

If the environment is already modeled, then you have a pretty good chance of putting things together, assuming that you start with a template of some-sort (for in game movement, and mouse-look functions).

Actually, you can have a look at a demo I made recently, which features a First Person setup: http://blenderartists.org/forum/showthread.php?t=193750

Overall texturing should be relatively painless, but an equal amount of time will be spent on lighting, which can be very difficult to do properly.

I would bake AO, and ray-traced soft shadows, into an additional texture layer, which would make everything look fairly realistic, with virtually no performance loss.

For things in the world that are liable to move: You should use dynamic lights, rather than a bake. Bake for static geometry, OpenGL lights for everything that moves.

And in the end, that mix should come across as fairly convincing, but, again, it depends on your skill.

That’s why it’s difficult to give a realistic time estimate, but I would guess ------ A month, from your current level, would be fairly reasonable.

3

Why isnt’ there a sarcasm/snark smiley on any website??? Truth is, I’ve pulled a couple of bunnies out of hats before so this is a mess of my own making.

It all depends on how much you already know about blender, and how much you have to do with blender.

Mostly I know how much I don’t know. I’m still having trouble getting the models to come in from Sketchup without being all sorts of jacked up. I have a thread going about that problem. Materials are ok but I’ve been seeing something about a 16 material limit and that has me concerned. I haven’t looked closely enough but I would estimate that by the time everything is modeled there will be anywhere between 50 and 75 materials used. It is a large building after all.

If the environment is already modeled, then you have a pretty good chance of putting things together, assuming that you start with a template of some-sort (for in game movement, and mouse-look functions).

The environment is still being modeled. It’s mostly down to placing the different items in the right place, laying is all the lights and applying materials to the structure. The movement is simple, you just walk and look around. Doors will open and lights will turn on and that’s it.

Overall texturing should be relatively painless, but an equal amount of time will be spent on lighting, which can be very difficult to do properly.

The good thing about the lights is there are only a few types of lights. There just happens to be a LOT of them.

I would bake AO, and ray-traced soft shadows, into an additional texture layer, which would make everything look fairly realistic, with virtually no performance loss.

I have no idea what you just said but I’ll look for tutorials on it.

For things in the world that are liable to move: You should use dynamic lights, rather than a bake. Bake for static geometry, OpenGL lights for everything that moves.

Yet another tutorial to look for. I don’t need dynamic lights or shadows so a bake sounds like the thing to use.

That’s why it’s difficult to give a realistic time estimate, but I would guess ------ A month, from your current level, would be fairly reasonable.

That’s not too bad. Now, that’s a month once I finish the tutorials I’m assuming. So 6 weeks is still more realistic.

4

I do a lot of archivis with blender, so I’ll add some comments.

Linked assets:

Any repeatable objects, like light fixtures(with lights attached), metal bars, soffets, windows, doors should be made and textured once in a seperate blend file, given a group, then relatively linked into the main file, where you can duplicate them as needed. This allows you to make changes to all of them by changing the source file.

As Social said, bake all the lighting:

Try to read as many tutorials on baking textures as possible.

Texture the entire building as if doing a render. Link in all your lights and other non-moveable assets. Once you can get a beautiful render of the place, you will create a second UV map and texture for the building, with each face using a unique portion of the UV texture, you can now bake the full render of the building onto the empty texture.

A little more info on baking:

You may want to bake lighting and ambient occlusion onto plain white textures to get the shadow map, and combine this with a second uv of repeating detailed textures (for things like wall panels, etc.)

And about the time frame, 6 weeks sounds fine for a beginner with competence.

5

Sounds good to me. I was worried about having to have everything native to the blend file.

Does the material limit still apply when using linked assets?

And about the time frame, 6 weeks sounds fine for a beginner with competence.

Confirmation of this number is good to have. Thanks.

6

Hey I’m also an archi turned bge user :slight_smile: I haven’t really modelled a archi project in the ge but I have used sketchup to build up some models. I found that when transfering sketchup into blender the best way is to separate the model as much as possible. The components and things should be bought in separately rather than trying to export the entire model it will normally drop stuff or bring everything in as one big mesh which will crash everything.
exporting your sketchup as in 3ds format also seems to carry texture info across so i somethimes add basic textures in sketchup and then I add normalmapping etc in blender to get the realistic look.

Then there are numerous FPS templates you can use to get the movement to work so you don’t have to worry too much abuot that.

All the best!

7

I’m starting off with just the basic walls, floor & roof and everything else will be linked. Everything from the doors, windows, telephones, toilets, etc.

The method I’m using to get the stuff from Microstation is to export it into Sketchup and then Export it as a .dae and then importing that into Blender. It works but it’s not very good