Good afternoon. I’m writing from Bogota - Colombia. I would ask for advice on the project I’m doing. This is a virtual tour of a proposed reconstruction and renovation of a temple in India. In pictures show modeling in Blender, and an original photo of the temple. my question is. I tried to do the modeling with the least amount of polys possible. when I try to run it with lights becomes very slow, I guess for all the calculations you need to do. The wall of the temple is the most complex because it has many inputs and outputs and that this changes the calculations. Someone could give me a tip on how to make the path is optimal in the average computer … or if you only can in those with video card. Thanks for your comments.
For optimizing, it really helps to have access to the blend file. However, I can try to give you a few pointers. First of all, the poly count shouldn’t be much of a problem unless you really wasted polygons in that scene. I don’t see any shadows in your rendered picture (other than ambient occlusion), but those can usually slow things down. What might be more applicable to your scene is making sure that it is broken up into smaller objects (not too small though) to allow for better culling.
There could be a number of reasons why your computer slows down with lights. If you are working with an older computer with a low amount of RAM and low-end graphics card, having many lights in a scene will cause lag in the Game Engine. Are you creating a virtual fly-through in the BGE or are you rendering out a video? Also what do you mean by the wall has many inputs and outputs?
first of all thank you very much for writing. Kupoman would like to clarify something. when I said that smaller objects … refers to reducing the scale of the design …?
first of all thank you very much for writing. fpsgod17 would like to clarify something. I work on the computer I have no problems, since it is a machine for this purpose. the problem happens when you run the executable virtual tour on other machines that do not have good specs.
Good night. thanks for writing. the computer has no problems … thereby. the idea is to take a virtual tour guided by the user at will. With the wall I mean the first image you send which shows the shape of the wall is not smooth but of various inputs and outputs … so are the four sides.
You could try adding occluder objects, they stop object that are behind them from being processed by the rastersizer. To do that you just have to create simple rectangles or planes and place them along walls. Then in the physics panel set the physics type to occlude and make the occluder objects invisible.
What I meant by smaller, was making sure to break up objects into pieces. This helps the frustum culling get rid of more geometry so less has to be rendered. I don’t know how your scene has been created so I can’t say for sure if this will help or not.
I think the (red) main temple has to much details.
The birds view usually includes a large area (long range). This means if you have a lot of details they will eat a lot of render time. On the other hand, you can’t see much of the details as they are just to small. A common way is to reduce the number of details. To provide the illusion of details you can use textures simulating them. Try to find a good balance between real details, and simulated details and what the viewer can see.
The first person, third person and frog views include a much smaller area but show much more details. Hidden parts can be removed (automatically by frustum culling, manual by occlusion culling[walls]). You need to present higher details in near range and less details in long range. It is hard to find a balance when having near and long range in one view.
A way to deal with this is the usage of LOD (level of detail) unfortunately we do not have an automatic LOD. For your architectural walk through I do not think you really need on. Just make sure to use low details from birds view and higher detail for first person view.
- Minimize vertices where applicable (especially when using boolean/path)
- Use duplicate with same data for similar objects (alt-d) (eg: windows frame, clutters, dome, etc)
- Break up objects by block (eg: building 1, building 2, etc)