SE Block Tools for Blender

The first layer contains the model of the fully-built block. You don’t need to restrict yourself to only one mesh-object on this layer. If it’s useful to split your block into several objects — probably to apply modifiers to some of them — by all means do so.

The second layer contains the simplified meshes that represent the block’s collision shape. If you don’t provide any collision meshes the resulting .mwm file will not contain collision data and SE will default to a simple collision box that encompasses the volume of the whole block.

Havok, the physics engine of SE, requires some settings to do its work. You provide those on Blender’s Physics tab by enabling Rigid Bodies for your collision meshes. Only the settings Shape, Mass, Friction and Bounciness (Restitution) are exported to Havok.

Of those settings Shape is the most important because it has a significant performance impact. Whenever possible try to only approximate the shape of your block. Use primitive shapes like Box or Cylinder. Use Convex Hull only if necessary. Avoid Mesh collisions.

The third layer defines the mount points of your block. The meshes on this layer are not directly exported but are used instead to calculate the mount point definitions .

This works by creating rectangular faces that use the material MountPoint and are aligned to one of six sides of the block. To quickly create the material and an object that has one mount point face for each of the six sides use the corresponding button on the Tools panel of the 3D view. Duplicate, resize and move those faces in edit mode to create more mount points for a side of the block. You can also split out faces to several objects if you want to apply modifiers like 'mirror' or 'array' to some of the mount point faces.

For mount point editing it is a good idea to switch to ortho-view (Numpad 5) and turn on 'snap to grid'. You can also tell the add-on to show the block’s bounding box by enabling the option on the scene tab . It’s dashed, orange lines represent the space your block occupies in-game. This should help you to position the mount points correctly.

If you don’t provide mount points none will be calculated. The game defaults to full mount points on all sides of the block in that case.

Layers 11-13 are used for the construction stages of the block. There is one layer for each stage of construction, from earliest to latest. If you change the number of construction layers you increase or decrease the number of construction stages. A block is allowed to have no construction stages at all.

By default the construction stages reuse the same collision shape your main block model uses. It’s possible to change that using different export settings. But keep in mind that can be dangerous to a ship or the player if a block changes its physical dimensions when it is welded to completion.

Layers 6-8 are used to model versions of your main model that have progressively less detail. The game can display those models when the block is too far away to see the small details of your main model. This way less triangles need to be rendered which is good for performance. The default distances at which the models are switched are 10, 30 and 50.

The internal name of this texture is DiffuseTexture. By convention Keen names texture-files of this kind with a suffix of _de.dds. The RGB-channels of this texture represent the diffuse color of a face—​just like with any regular digital image.

The A-channel of the texture does not represent the transparency of the pixels, though. Instead it tells SE how much emissive light a pixel gives off in the dark. Strangely a value of 0 means full brightness, a value of 255 means no emissivity. In theory the light given off could be different from the diffuse color of the pixel but in SE it is always the same as the diffuse color so there are no extra channels for that.

If instead the texture-filename ends in _me.dds the A-channel serves a double purpose. Here the values in the range 115 to 140 represent a mask that controls which parts of the texture can be recolored by the player.

Recoloring uses a HSV color-model in which the player-chosen hue replaces the original hue. Saturation and value are added and clamped at a maximum of 100.

If the value of the texture’s A-channel is not in the range 115 to 140 it still controls emissivity but the values have a slightly different meaning.

Internally named NormalTexture this kind of texture has a suffix of _ns.dds and contains no color information at all. The RGB-channels contain a standard DirectX normal-map. The A-channel tells the game how much of a material’s maximum specularity settings it should apply to a pixel. A value of 0 means "don’t apply any specularity", a value of 255 means "apply full specularity".

The ColorMetalTexture (_cm) and NormalGlossTexture (_ng) correspond with the common metal/roughness workflow where glossiness is just the inverse of roughness. Here the normal map uses the OpenGL convention, i.e. in comparison to the DirectX convention the green channel is inverted.

The AddMapsTexture (_add) combines additional greyscale maps in its four channels:

The AlphamaskTexture (_alphamask) is optionally used to make parts of a texture transparent. It has a single greyscale channel which really only differentiates between two states: fully transparent below a value of 128 or fully opaque at 128 or above. This is most useful for things like foliage and small details like facial hair. If you want to use this texture you need to also set the technique of the material that uses it to Alphamask.

Materials are what you use to tell SE which texture-files to use for the faces of your meshes (and you use UV-mapping to tell the game which part of a texture to use for a face).

In Blender there are several types of materials depending on the renderer you have chosen. If you choose Blender Render as the renderer you can switch between simple and node-based materials. If you choose Cycles as the renderer you can only use node-based materials.

This add-on needs you to use node-based materials because it sets up several nodes to preview the textures in Blender. It also requires the use of Cycles because the nodes it sets up are meant for this renderer. So to set a material up for use with SE first switch the scene to use Cycles Render and then use the button on the Material property tab. You need to repeat that for each new material you define.

By doing that you get six texture-slots for the different kinds of textures SE expects. Point them each to the texture-file corresponding with its name.

Specularity is a DirectX 9 mode-only setting and is configured per material. You do that in the "Space Engineers" panel of the material. The values you set there are the maximum values for that material. How much of those maximum values are actually applied to a pixel of a face is determined by the A-channel of the normal/specular texture.

The effect of the two values Intensity and Power are hard to describe so it is best if you play with them and watch the results in-game. Just know that at high values (30.0 / 30.0) the game’s specularity shader also starts to mirror the environment like smooth metal would do.

Glass materials are handled specially by the game. They have some extra settings that are not contained in an .mwm file but are instead stored in a file called TransparentMaterials.sbc. It makes no sense for this add-on to support editing the content of this file because the settings cannot be viewed in Blender, anyway.

What you get if you change a material from Normal to Glass are the settings that are stored in the .mwm file on export. For the meaning of these settings please refer to Keen’s tutorial on transparent materials.

Normally you are free to choose the name of a material. But there are some visuals in SE that are only enabled if the material of the face has a special name

SE uses empties to represent predefined points of interest on a model like terminal keypads, cargo ports or the location of the thruster flame. The purpose of each empty is determined by its name — i.e. detector_terminal_1 or detector_conveyor_1 for example.

When SE "mirrors" a block on a mirroring-plane what really happens is that a copy of the block is placed on the opposite side of the plane at the same distance to the plane as the original block. Then the copied block is rotated according to the mirroring-settings for the local block-axis that is currently pointing towards the mirroring-plane. So there is a maximum of three independent mirroring-rotations per block.

You can model these rotations in Blender by adding and rotating three specially named empties to the scene. Fittingly their names are MirrorLeftRight, MirrorTopBottom and MirrorFrontBack. They need to be placed on the layer that is designated as the mirroring layer in the export settings. By default this is the fourth layer. Use the corresponding button on the Tools panel of the 3D view to quickly do so. It’s okay if Blender adds a .001 suffix to the names, the add-on can handle that.

The rotations you set on the empties are only valid for export if they are multiples of 90° — so it’s a good idea to turn on snapping when you rotate them in the 3D view. Even if you adhere to that rule some combinations cannot be used because SE does not support them. To check if a rotation is valid watch the Mirroring property on the Data tab of the empty.

This node uses a layer mask to filter out objects that are not on the marked layers.

This filter matches objects by their name. You can either choose an exact name or switch the node to use a regular expression. You can also invert the match so only objects get included that do not match the pattern.

This filter works just like an Object Name filter, only on the names of groups the object is in.

This node has two input sockets and selects between them depending on the currently exported block-size. This is useful if you have objects in your scene that should only be exported for one block-size but not for the other:

This node adds one output-socket for every layer you select. Each of those sockets only provides the objects from the corresponding layer. This is useful for sequential groups of meshes like construction stages. The node also implicitly numbers the output-sockets starting at 1. The number is available for text-substitution as parameter ${n}.