[Translation] cryengine3 art guide

Modeling Guide

The following are some important principles that you must follow when creating an object model.

We often think about how to control multiple sides. Remember that a smooth profile is more important than a tessellation smooth surface. Let's take a look at how to apply a polygon to the following car model. The degree of polymerization of a smooth side polygon is very low, and the section that forms the contour, for example, there are many subdivisions at the fenders.

 

Add the edge to the appropriate collection that is retained. If necessary, perform an inverted (DAO) angle on the edge or add an additional edge. This is done to obtain the normalmap and smoothing model, in particular, when you need to make a high-gloss texture, you can add some extra edges to make the model coloring look smoother, as shown in the following car on the left side of the edge.

 

Keep in mind tesselation. Before exporting, always make sure to manually adjust tesselation to maximize the space utilization of each triangle. This will help us maintain the consistency of the coloring. The artist should ensure that the triangle belt is not damaged and avoid the appearance of a triangle fan.

 

Using as few coloring groups as possible, adding a new shading group usually results in the same overhead as the edge chamfer, using only one shading Group) because vertices are effectively separated into two copies by the engine in Different coloring groups. In addition, chamfering at the contour helps to color (it creates a highlight ). But avoid confusion, because an edge chamfer is not as obvious as the two sides are in different colored groups, it looks more natural because most sides are not % 100 sharp. This is determined based on specific art resources.

 

Always try to use the basic graph metaimage (box) box, (sphere) sphere, or (capsule body) capsules as the body of the touch detection, even some composite bodies using these multiple elements save the cost of using the original modeling bodies as the collision proxy bodies. Make sure that the color of these collision elements is in 3dmax, and consider how many types are required.

Remember that decals always performs two-direction projection (projected ). this means that everything needs to be taken into consideration from two aspects. Not all faces need to be overwritten by collision elements, there is no need to use a collision proxy to enclose areas that cannot be reached by some roles. Make sure they are in a smooth group.

 

Make sure that you use dummies as the root node for export during 3DS MAX export, these help bodies can maintain the readability and cleanliness of the file, and enable other artists to conveniently work on this file.

Work on the set layer and follow the naming rules.

 

Make sure that you use the dump to reduce the complexity of the model. The first layer of the DLE should not contain small details. Transparent objects such as glass should be deleted, the outline should be kept, and the complexity and number of materials should be reduced. Color changes can be compensated using vertex colors. The second layer of the dashboard should not contain some surfaces, so that there is no need to have too precise outlines to minimize the number of surfaces on most surfaces. The simpler the last layer, the better. You can use only one material without retaining the contour. You can also use some vertex colors to replace them. At least % 50 of the triangle surface must be reduced for each level of dashboard.

 

You need to apply the multi-edge overhead to areas close to the role as much as possible, which can be saved in some distant areas. Think about where these resources will be applied. If it is a background object, make sure that it is as simple as possible (low number of triangles/low texture solution)

 

Try to match the smooth group/material splits/UV boundary. The sharing edge must be ensured for each material splitting (material splits), smoothing group, and UV boundary splitting. When splitting in many different places, ensure that the vertex is split only once, and ensure that the number of triangles is not affected.

Similarly, use as few as possible the smooth group/material splits/UV boundary.

Avoid the impact on the back (meaning to disable the back rendering as needed)

Delete a polygon invisible to the role

Merging meshes into a uniform mesh can save drawcils

 

Texture and UV Expansion

The texture size applied to an object depends on the area size and quantity of the object displayed on the plane. For crysis, the screen resolution set by crytek is 800x600, assuming that the object fills the average screen height at any time (600/2 = 300 pixel ); therefore, we need to use a texture resolution of at least X. It would be better to use 512x512. The best case is that each pixel on the screen represents an element in the texture. Of course, this depends on the size of the object in 3D. You cannot directly match the texture space in 2D to 3D space.

Make sure that all texture ratios are constant. This means that all objects on the screen use the prime number equivalent to the same texture image per meter. In this way, you can ensure that all objects have roughly the same definition on the screen, and the texture is not always uneven.

Objects on the screen keep the same distance (for example, weapons) as the role, you can use a higher texture ratio because they are the closest to the role.

 

The texture is always square, and the processing speed of a square is faster than that of a non-square. If you have an object that uses a non-square texture, try to use a square texture, or even share a square texture with other objects.

Reuse textures. If you are creating a resource to ensure that the texture can be pasted, you do not need to re-expand the special UV. In addition, sharing the arts and sciences on different objects ensures the same texture ratio.

Do not overdo transparent textures. The more space on the screen is covered with transparency, the higher the rendering overhead. The use of alphatest in the transparent area will bring the second drawcall.

Remember to overdraw (overdraw ). Cut a polygon that is covered by a transparent texture to match its shape. This ensures that the less the transparent area appears in the screen area, the better. In this way, more textures can be stored in textures.

 

Merge UV to ensure that all UV vertices on each other are merged. This is very important for some mesh with a mirror, which can prevent vertices on the UV boundary from being split repeatedly.

 

Create a special map for an object ). Store them to the alpha channel of a difuuse texture and start gloss in the diffuse Alpha material. A good highlight texture will forge a lot of details. Sometimes you or even objects do not need a normalmap. Do not add additional textures, which is a waste of memory.

 

There are some creative ways to create various visual effects. At the same time, only textures are used (actually lightmaps)

1. Use vertex color, which is a simple method to add shadow information and color changes to the grid. Bake the environment block to the vertex color, and then manually add the color

2. Add another material, and you can get the difference by adding additional vertices. Use the vertex color to increase the contrast and color information to texture.

 

Depending on the sticker, You can provide some different textures. Instead of using the sticker in the editor, you can create them using the floating geometry option in 3dsmax to ensure that all the stickers are on one texture, in this way, you do not need additional materials. To merge textures, Alpha is required.

 

In some cases, you may need to add extra detail textures, such as an extra decoration or a piece at the bottom, which requires some changes. This is usually a good idea, but make sure that the main part of the texture is the texture itself. You need to add some extra Triangle Mesh on the mesh for texture tile. In general, this visual improvement is worthwhile. It also provides an opportunity to add changes through vertex colors.

 

Use textures to the maximum extent. Do not use a paved image to obtain as many details as possible and reduce the texture size.

Use all available texture spaces, do not use black as the background color of the initial texture (or do not use red in the game, because mipmaping), black is neutral, it does not give you an intuitive impression of how much space is not used. (In fact, it tells you to select a background color that will no longer be used in the game as a texture .)

 

Shaders

The rational application of memory, normalmaps is a common diffusemap or specularmap 4 times the cost. Therefore, normalmap should only have a diffusemap 1/4 resolution, avoiding the use of normalmap to create a specularmap

Fine normalmap is very expensive and should be visually compensated with detailed color textures as much as possible.

Use the diffusealpha shader generation option when the highlight texture does not contain additional color information. You can use the specular color in the shader to compensate for the color information missing from the highlight texture. By doing so, you can save 1 drawcall.

An environment map may be used in a map to ensure a uniform appearance and save memory overhead.

Avoid too many expensive shader, including skin shader, eye shader, wet shader and frozen shader.

Try to use Alpha testing for transparent objects (for example, vegetation)

Color is very important for some objects, such as certain items, cars, airplanes, and garbage bins. Color coloring means creating a grayscale texture for diffusemap. colors are colored by the diffuse color attribute in the material editor. All objects and surfaces can change colors and use the same material.

You need to determine which objects use the coloring function in advance, and then set it correctly in 3dsmax. By doing so, you can save a lot of memory.

General Comment

The level design should start from some basic proxy collection, which can begin by using some agent models created under solids or 3dsmax. Then all logic-related entites are placed. Then transfer the level to art.

Visblocking and occlusion in a level are very important. Therefore, the level should be a relatively small visual distance, and as few objects as possible should be displayed on the screen at the same time. The role should always see a very small activity area (Action bubble ). Including the excessive zone ).

If the game design requires more visible areas to be displayed on the screen at the same time, the complexity and complexity of the level should be balanced. To this end, the artist needs to know that objects within this range need to be used. (Small, closed indoor, large, outdoor, several objects on the screen, AI and screen Parameter options)