Collada plugin for After Effects

>More info here
Plugins and Scripts > After Effects

Collada Plugin for After Effects - Import

Import Process:

collada import ui

Collada import options dialog

Set options as documented below and press the "Import" button to open the file dialog. After the command window closes push the "Continue" button.

Push the Import button to open the file dialog and choose the collada file.
The status of the import will show in the command window.
The plugin does not import geometry by default, but it still has to read all the geometry in the collada file. If there is a lot of geometry it will delay the start of the actual import process.

Push the Continue button after the command window closes. The command window may just flash on the screen for a moment depending on the complexity of the scene file, geometry and number of animation keys. The status of the import will show in the info panel.

A new folder named "Collada Import" containing a new composition with the same name as the imported file will appear in the project panel when the import is complete. If the null sizes are too small or too large you can re-import with a manual scale factor. A larger scale makes for smaller nulls in the view and smaller scale values make larger nulls. The imported scale shows in the info panel immediately after the import process. It will create footage items from images referenced in the file.

Auto Scale the scene.:

Will attempt to automatically scale the collada scene based on the positions of all the nodes so that the 3D nulls will have a good size in the viewport. v1.1x will choose a scale of 1 or 100.
Uncheck and type a value for manual scaling.
Auto scale only looks at top level scene items to determine scale and is only a crude approximation.
The calculated scale value will display in the info panel when import is complete. This can be used as a starting point for a manual scale import.

Use comp frame rate.:

Will use the frame rate value of the selected composition in the newly created collada comp.
Uncheck and type a value to override or check "Guess" to have the importer try to guess the frame rate from the collada file. There must be a key on every frame of the Collada file for the guess to work. If no comp is selected in the project tab or the timeline, the edit box value will be used. The Collada file frame rate must match the frame rate of any corresponding footage or there will be synchronization problems.

Comp size from active comp.:

Composition height will match whatever is selected in the Project tab. If nothing is selected in the Project tab then the height will be read from the active comp in the timeline and if nothing is selected in the timeline then it will read the height in the input field.

A square comp will be created for the collada data imported. To override the comp size uncheck the box and type the desired value.

Include joint nodes.:

Check to read in skeleton joint nodes.

Include geometry nodes as nulls.:

When checked will read the transform nodes for geometric(mesh) objects.

Faces and Vertices(experimental).:

Experimental feature in version 1.1x and up, when checked will read mesh data. It will never go beyond the experimental stage because of performance issues. Each mesh face translates to 2 layers in After Effects and the number can grow to unmanageable levels very easily and bring After Effects to a crawl. Very simple geometry is the key.

Works best when geometry is composed of one square face, with a texture applied, facing the positive Z axis with no rotations. Translated, scale and rotate into position and the face will match up in After Effects. This has only been tested in a Z-up 3D application(trueSpace).

If Vertices is checked then 3D nulls will be created for each vertex of each face of the mesh. So a cube won't create 8 nulls but will create 24 nulls. A script is available to help isolate the redundant vertex layers. If Face is checked then a solid or image layer is created for each mesh face. The solid layer will have a size of 256 x 256 and the mesh color will be read in. One solid will be created for each mesh in the scene. If the mesh has a texture applied then a footage item will be created for it instead of a solid. Each solid/image layer corresponds to one face of the mesh and will have a alpha track mat to shape and size the result. The solid layers have cast shadows turned off.

Face import only works with simple box and rectangle geometry. All faces should be made of right angles only. The shape of the faces in the mesh uv map must match or be scalable to the shape of the actual face and it must fit in the texture space of [0 1]. The image that is imported can only be replaced with another image of the same dimensions. It is best to have all geometry textured and uv mapped. This prevents the need for a solid layer for each face. Triangle faces are not read in properly.

Requires that that the "Include geometry nodes as nulls" option is checked.

Click this line for the geometry import overview

Add corresponding 2D nulls. and Hide 3D nulls.:

When checked will create 2D nulls linked to the imported 3D nulls via expressions.
Check the hide 3D nulls will make them shy and turn on shy for the composition.

Include lights:

Check to import lights from the collada file. Lights import with cast shadows off. The easiest way to brighten a dark import is to add an ambient light.

Brute force animation.:

The After Effects importer is designed to read in collada files that have simple xyz rotations and xyz translations and will take into account extra translation nodes for pivot point representations. Some collada files have many more translation and/or rotation nodes so a special import mode called "Brute Force" is used to read these files. It works by creating a separate null in After Effects for each transform node defined in the collada file and parents them based on the collada scene translation/rotation order. The Brute force imported data will also be exported in a kind of pre-digested form that can be read by programs with fixed rotation orders that don't match the rotation order of the original collada file.
The default behavior and bruteforce give slightly different results - default shows the pivot, brute force shows the result of the pivot but can be made to look like the default case.

The names of the scene items should be kept short. The brute force method increases the size of the layer names and could run past the maximum size allowed in After Effects(CS3). The names need to be unique for this option to work.


  • The collada file frame rate must match the frame rate of the new composition.
  • Animation keys should be baked(a key frame for every frame) in order for the animation curves of the collada file and the result in After Effects to match up.
  • Camera target and spotlight target information does not translate well and these items should be baked and converted to simple rotation/translation type animations before exporting from the 3D application.
  • Matrix style transforms are ignored. Matrix node transforms are an option for some exporters.
    This plugin only reads the more common rotation, translation and scale nodes.
  • A z-up collada file will import standing on it's side relative to After Effects' y-up world. The top level layers can be parented to a new null layer and the null can be rotated 90 degrees to make it stand up. A script is available that will create the null and do the parenting automatically here.
  • Lightwave versions prior to 10.1 export the camera aspect ratio incorrectly. A script to fix the problem inside After Effects after importing is available.
  • Lightwave does not export camera fov, light color or spot angle animation. There is a Python script that can be used to add these animations to the collada file for Lightwave 11.
  • LScript's are available for earlier versions of Lightwave to tie these animations to nulls that can later be copied into the proper animation tracks or pick whipped for these items. Reverse the process by adding nulls in AE and whip picking back to the fov, spot angle and color animations to export for programs that don't read these values from collada files.
  • Non-brute force import will read the first extra translation node as a pivot and ignore all others.
  • If a node has more than 3 rotations the importer will stop with a message saying to use the brute force import option
  • Don't right-click in the project view to load the collada file since the selected composition becomes deselected. Use the file menu to access the import function. You can use right-click , but it will not read frame rate or comp height since the active comp will become inactivated.
  • The faces option only works reliably on 4 sided polygons with all interior angles equal to 90 degrees. It works best on box shapes. Faces will often import with incorrect orientations. The importer is hard coded to limit the face count to 512. If the faces appear dark after importing they can be lightened by removing all the lights, adding more lights or changing the material properties so that light interaction is turned off. 2 layers will be created for each face. One layer for the image or solid and another that is linked as a track matte layer. 4 sided faces import as simple rectangular shapes. Faces with more than 4 vertices or interior angles not equal to 90 degrees will have their shapes defined by a layer mask in the track matte.
  • The faces option needs UV mapped geometry to import image based footage. Best results come from meshes with a single UV map and as little distortion as possible.
  • The importer will not read footage items that have spaces in the file path.

June 9, 2010