If you load a model from G3D, DAE, FBX or X3D with embedded animations, Gorilla3D will automatically load and add all animations to the model.

Animations are handled by the TGorillaAnimationManager inside of a TGorillaModel component. By the animation manager you are allowed to browse through all available animations and to playback or stop those.

Loading of model-animations during import

Setting up a TGorillaLoadOptions record enables to import animations directly with importing the model itself.

Since 0.8.4.+ you are allowed to use this way for DAE or FBX files.

var LOpts : TGorillaLoadOptions;
LOpts := TGorillaLoadOptions.Create('C:\Bellydancing.dae', nil);
// forbid to import light instances (optional)
LOpts.ImportLights := false;
// forbid to import camera instances (optional)
LOpts.ImportCameras := false;
// setup automatically loaded animations
LOpts.ImportAnimations := true;
// setup an array of fixed size with filenames
System.SetLength(LOpts.AdditionalAnimations, 2);
LOpts.AdditionalAnimations[0] := 'C:\Idle.dae';
LOpts.AdditionalAnimations[1] := 'C:\Walk.dae';
// limit the size of loaded textures (optional)
LOpts.TextureLimits := Point(512, 512);
GorillaModel1 := TGorillaModel.LoadModelFromFile(GorillaViewport1, LOpts);
GorillaModel1.Parent := GorillaViewport1;
// get first animation found and playback
// CAUTION: it is not the first animation you have imported. It is the first animation could be found in animation-name-hashmap.
var LAnim : TGorillaAnimation;
LAnim := GorillaModel1.AnimationManager.Current;

Post-loading of model-animations

In case you separated the model data from animations, you can load animations after loading the model.

  GorillaModel1.AddAnimationFromFile('Bellydancing.dae', GORILLA_ANIMATION_CACHING_DEFAULT);

This is only supported for DAE files!

Playing model-animations

Play/stop current model-animation

The “Current” property of the AnimationManager represents the currently active animation or if not set yet, it will automatically select the first animation from the internal animation map.

You can simply start and stop animations by the provided methods.


Iterating model-animations

To iterate the animation list and playback you can use the provided methods PlayNextAnimation() and PlayPreviousAnimation().

procedure TForm1.FormKeyUp(Sender: TObject; var Key: Word; var KeyChar: Char;
  Shift: TShiftState);
  if not Assigned(GorillaModel1) then
  if (KeyChar = 'e') then
    // next animation if exists
    if (GorillaModel1.AnimationManager.Animations.Count > 0) then
  else if (KeyChar = 'q') then
    // previous animation if exists
    if (GorillaModel1.AnimationManager.Animations.Count > 0) then

Playback by model-animation name

At the moment animation names are set by the order of adding. So the first animation is called “Animation1”, the second “Animation2” and so on.

You are allowed to browse an animation by its name, by using the provided hashmap component inside of the AnimationManager.

procedure TForm1.Playback(const AName : String);
var LAnim : TGorillaAnimation
  if not GorillaModel1.AnimationManager.Animations.TryGetValue(AName, LAnim) then
    LAnim := nil;
  if Assigned(LAnim) then

Model-Animation caching

To animate models with a large number of vertices is very slow. Using the same animated model multiple times is also extremely slow. Due to those performance needs, Gorilla3D offers some caching techniques for model animations.

As you've seen in the previous examples for loading a model, there was a supplied parameter for animation caching: “GORILLA_ANIMATION_CACHING_DEFAULT” (Gorilla.DefTypes.pas)

Gorilla3D offers a configurable caching mechanism to suite your needs.

List of animation caching flags:

Flag Notes
TAnimationCachingFlag.AllAttributes All attributes (position, texcoord, color, normal, …) of each vertex are cached in a framebuffer object. This may be the fastest method, but memory usage is enormous. If possible use TAnimationCachingFlag.VertexPosition instead.
TAnimationCachingFlag.Compressed Compressed caching is recommend for large models and long animations, but will not create a framebuffer directly in GPU memory. Cached frames will be compressed inside the application memory. On MSWINDOWS x86 we use LZ4 compression for enourmous performance boost. On all other platforms we use ZLib compression algorithms by default.
TAnimationCachingFlag.VertexPosition Cached frames will ONLY contain the vertex position. The rest of the vertex data will come from a static mesh buffer (not animatable). This is the recommend setting for caching, because in 95% of all cases you only transform vertex position on an animation skin.

The constant value “GORILLA_ANIMATION_CACHING_DEFAULT” uses only the VertexPosition flag as default setting:

GORILLA_ANIMATION_CACHING_DEFAULT : TAnimationCachingFlags = [VertexPosition];


It is possible to combine compressed and vertex-position caching flags to reduce the number of data compressed. Otherwise on compressed caching, full vertex data will be cached.

GORILLA_ANIMATION_CACHING_COMPRESSEDEX = TAnimationCachingFlags = [TAnimationCachingFlag.Compressed, TAnimationCachingFlag.VertexPosition]


The caching framerate used is currently fixed to 60 FPS. But in future you should be allowed to configure this setting to reduce memory usage.

Next step: Materials