Welcome, Guest. Please login or register.

Login with username, password and session length

 
Advanced search

1411274 Posts in 69323 Topics- by 58380 Members - Latest Member: bob1029

March 28, 2024, 03:41:07 AM

Need hosting? Check out Digital Ocean
(more details in this thread)
TIGSource ForumsDeveloperArt (Moderator: JWK5)Methods for deciding on Polycount?
Pages: [1]
Print
Author Topic: Methods for deciding on Polycount?  (Read 934 times)
Lance of Longinus
Level 0
***



View Profile
« on: April 22, 2018, 11:23:00 AM »

When it comes to 3d modeling, the thing that gives me the most headaches is how to decide on the maximum of polygons for each type of model.

Most information about that topic that I found on the internet seams to be about older games and is usually not very clear. While I understand that when you have many objects in a certain area, it is a good idea to give them less polygons and when you have a large object close to the player, it is better to have more,
I have no idea on how to make even the vaguest estimation of the actual polycount.

Is there any way to roughly get an idea on how many polygons to use in order to strike a good balance between model quality and computer performance?
Logged
Richard Kain
Level 10
*****



View Profile WWW
« Reply #1 on: April 23, 2018, 01:51:42 PM »

The maximum is not what you need to be concerning yourself with. What you should think about is the minimum number, not the maximum. The maximum is theoretically infinite, depending on the hardware you're running it on. The minimum is a harder limit, and one that is much easier to predict, as it ties more directly into your personal optimization.

And as far as that minimum is concerned, what you need to focus on is level-of-detail, and providing different levels of rendering for different screen sizes. For objects that are never going to be the focus of a scene, you can get away with minimal LOD work. Just throw in two or three different levels, based on what looks decent from different screen sizes. For major characters or objects that ARE going to be heavily focused on visually, you will probably want 4-5 LODs, with the top ones having high enough polycounts to keep the edges reasonably smooth, while the lowest levels can be just enough polygons to keep the edges coherent.

What you need to ask yourself, is how big will the model be, compared to the overall screen height, at any given time, and what LOD can I get away with for that height? That is how LOD usually works, and what you need to think about. The biggest issue with most 3D models and polygons usually involves the edges, or silhouette of the model. That is where those low polycounts are usually the most noticeable. If you zoom in on your model regularly (so that it is very large compared to the total height of the screen) then you will need a higher polycount so that the edges of the model still look fairly smooth. When you zoom out on a model, it renders as a much smaller number of pixels, and you don't need as many polygons to have a convincing silhouette. It's all about the edges.

All of these factors are subjective, and they vary from model to model. A model made up mainly of hard edges and angles isn't going to need nearly as many polygons, no matter what LOD support you add. It will look the same from almost any zoom value. A smooth character with a lot of rounding and soft edges will need plenty of polygons in order to maintain its appearance.
Logged
Lance of Longinus
Level 0
***



View Profile
« Reply #2 on: May 19, 2018, 06:49:27 AM »

Thank you for your answer.

So if I would use these two models of an oil drum to the left for far distances and in the middle for close distances and give them two seperate normal maps based on the high resolution model to the right, this would improve performance?



Or would I have to make all models use the same normal map and have a more similar geometry?
« Last Edit: May 19, 2018, 08:54:29 AM by Lance of Longinus » Logged
Pages: [1]
Print
Jump to:  

Theme orange-lt created by panic