Low poly models is somehow a large term because it depends on lots of specifications, the first question that comes to mind is, how low are we talking about? is it 5k tris? 20k tris? then what do high poly models mean?
In this article, we will define the term itself, where it is used, and real-world examples and metrics used in the industry to fully have an understanding of what low poly models mean.
Check out this article if you’re interested to know more about The Game Development Process
Definition of Low-poly Models
The term itself refers to meshes that are very low in polycount, and it is generally used in the gaming industry as it is a must to have meshes that have a very low poly count in order for the game to run smoothly with no issues.
A 3D mesh consists of 3 main components:
- Vertices ( one point )
- Edges ( a line consisting of 2 vertices )
- Polygons/Triangles ( 4 vertices or 3 vertices / 3 edges or 3 edges )
Below are images representing each component you have a visual understanding of these elements as these will make you low poly models and your high poly models and it’s essential you know these elements very well.
Why are Low Poly Models Important?
Nowadays the gaming industry has developed quite a lot and with it of course technological advances in handheld devices, for example, cellphones, tablets etc.. that’s why games can afford now 3d models that older game developers couldn’t even dream of implementing in games.
Nonetheless, game engines still need optimized 3D models in order for the game to run on devices, and in order to do that 3D artists need to create low poly versions of the models following a TRC (technical requirement checklist) made by the tech team.
Benefits of Low Poly Models
There are lots of benefits to having low poly models for your game but I do also argue that there are also downsides to it as well concerning the visual quality of certain assets and how far can we push for quality though for game production I think the benefits outweigh the downsides by a lot.
Let’s dive deep inside some of the benefits of low poly models:
The higher the polycount the slower the game would be the reason is when the game loads on a mobile device first the game needs to be downloaded.
Now some packs are downloaded at the beginning and some at runtime( while playing) and the bigger those files are the more space it will use and the more RAM memory from your phone it will consume.
For that reason, you notice while playing some games on your phone it gets heated and the battery dies out very very quickly, this is also true if you have a lot of apps open or lots of chrome tabs open, because this uses lots of RAM (Random Access Memory) in other terms your device short term memory.
So in order to reduce the game size and not consume so much memory we need to reduce the size of the 3D models that make up the majority of our assets.
That will make our game not only load faster but also have a very fast FPS (frames per second) which means there would be no lagging while playing the game and the player will have the most pleasant experience.
Compatibility with More Devices and Platforms
This is a very important point that will broaden our target audience, let me explain why. If we make a game that looks amazing but only users that have an iPhone 14 or above can play it because of its size and FPS, well that won’t set us up for success, won’t it?
The more device coverage you can have the bigger the player base hence the more potential revenue you would be making in the long run, and in order to cover low-end device you need optimized low poly 3d models that load fast and runs fast.
Easier to Produce
As a 3D artist myself I can tell you that creating 3D models for movies and advertisement is way way harder than creating low poly 3d models for games.
The reason is simple, creating a high poly 3d model involved 3d sculpting, retopology, displacement maps, shaders, look dev is harder to manipulate, etc…
While creating a low poly 3d model isn’t as complicated! you can even skip the 3d sculpting phase if you have a good base mesh and the style you are after is hand-painted textures you don’t even have to do any bakes.
Simply tweak your base mesh and paint your textures and that’s it! I will explain more about the process next.
Creating Low Poly Models
To create a low poly model there are several ways to tackle it, it depends on the style of the game, how low you want to get, the tools available for you, and how much time you have. Let’s discuss 3 of the most famous ones below.
Low Poly Models Using Basemesh Method
This is the oldest way to create a low poly model, it was present even before sculpting software was famous, actually, at the time this method was the only method. it consists of the following steps:
- Create (poly by poly modeling ) or use an old Base mesh
- Tweak it in Maya/Blender to conform to the new asset you have
- Sculpt over it if needed ( depends on the style )
- Bake your maps in Substance painter ( depends on the style )
that’s basically it if you are tight on time you would re-use an old basemesh you have or create a new one and use it for several assets on the project.
Low Poly Models Using Retopology Method
This method emerged with the rise of 3d sculpting software such as Zbrush and Mudbox (ahh good old days) and it consists of the following steps:
- Sculpt in Zbrush or Blender
- Retopology in Topogun or Blender
- Bake maps in Substance Painter ( depends on the style )
Now, this is a longer process than the one before but this depends on the needs and the artistic direction set out by the Lead artist.
Low Poly Models Using Decimation Method
I don’t recommend using this method except for small props and side objects because this is basically an automated process and it gives out non-planned results, its good for small rocks and sticks, etc.. the process is as follows:
- Import your mesh in Zbrush
- Tools – Decimation Master
- Click Preproccess current
- Click Decimate current
- Repeat the 2 last steps until your mesh is as low as you would like it to be.
I am going, to be honest with you, I never liked optimization as a creative person because it generally brought down the level of quality of my assets.
But with time I came to understand it and be friend with it, as I know how frustrating a game that is lagging and not running the way it should be running for me outweighs great-looking assets.
Here are some game optimization tips I picked up over the years:
- Reduce the polycount ( Max Characters Max 7k tris, Buildings Max 5k tris )
- Delete all hidden polys that are not seen by the camera view
- PNG textures ( Max 512 for characters and buildings )
- Atlas textures
- Use LODs (Level of Detail) (low, mid, high depending on zoom levels)
- Bake Lightmaps and use lighprobes
- Reduce the number of UV islands as much as possible
- Combine similar meshes when possible
I will update this list with more tips as I gain more experience so you guys can benefit from my journey.
Best Practices for Using Low Poly Models in Game Development
Quality vs Performance
There is no hiding from this one, and there is no reason to hide it’s just about setting up your priorities and what you want to achieve, do you want to have a great-looking game that doesn’t run well? or do you want to have a crap looking game that runs super fast? OR do you want to have a balance where you have a decent-looking game that runs fairly well?
Life is not black and white at least not in this case, it’s just about balancing and making smart decisions, it’s always extra work to optimize but that’s needed work in order to have an overall good experience remember a great game is a combination of all elements working together and providing a good experience to the player.
Usage of LODs
Let me start by saying don’t leave LODs till the end, because then it will be like creating the asset from scratch again. Have in mind while you creating the asset that this will be High and then I can break this down into 2 more levels this will make the process easier for later.
For those who don’t know what LODs are, it stands for Level OF Detail, each 3d asset created in the game has a certain poly count LOD is creating different variations of that asset polycount, one would be for example 5k tris and one would be 2k tris and one would be 500 tris.
Then based on the zoom level of the main camera view the right assets would be loaded, for example, if the camera is zoomed out it doesn’t make sense to load up the 5k tris model while it’s just 4 pixels on the device screen right?
Low poly modeling isn’t going away any time soon but compared to the limitations we have 10 years ago where a low poly model of a character needed to be 120 polygons we have indeed progressed a lot.
Optimization is a crucial step of the asset creation process it shouldn’t be neglected and it should be integrated well at the beginning in order to save time and be consistent with the art style of all 3d low poly models in the game.