Before you start modeling, you need to know a few modeling concepts that are very specific to Quake.
Read this section carefully, since it contains valuable tips and tricks that allow you sucessfully create a model for Quake.
Monsters perform a number of actions. These actions are modeled as scenes. Scenes are sequences of frames that sometimes are played in a continuous loop like when a monster is running or standing. Sometimes a scene is played once, like when a monster is dying.
In general monsters have the following kinds of scenes :
Most monsters have different attack, pain and death scenes, to make the actions look a bit more realistic.
It's amazing to see with how little frames a run scene is animated. A soldier that is running has a run scene of no more than 8 frames. The reason for this is that scenes are played back with a fixed rate of 10 frames per second. This would make a run scene of 8 frames which depicts a soldier taking two steps take 0,8 seconds, which is realistic for running.
If you insert too many frames into a fast action scene, it will look like a slow motion replay. On the other hand sometimes you need to insert more frames to make the animation go slower. For example to prevent a standing monster from looking like it has a violent shaking affliction, you need to insert a lot of frames. Assuming the monster is looking around and turns its head every 10 seconds or so, this would mean you'd need 100 frames to show that.
For models like torches that stay in one spot and that don't interact with the players, you can turn the scene into a framegroup. A framegroup is a special scene where you can specify a duration for every frame. Framegroup animations are handled on the client side, you might have noticed that during network play when the server stops you still can see the torches burning. Well that is because the animations are played on the client side.
The structure of Quake models limits the things you can do in animations. There is a single list of triangles for all the frames in the model. This means that every frame must have the same number of vertices. What you cannot do is for example let the monster swing an ax in one scene and remove this ax in another scene. There is a trick to workaround this limitation though. To make an object disappear you need to scale down the object to a minuscule size and put it inside the body of the monster. The original Quake models use this trick very often.
You can have more than one skin in your model, but there is only a single list of skin vertices that map the skins onto the 3D frames. This means that all your skins need exactly the same layout and the same size.
Objects and pivots are concepts that are not originally found in Quake models, but that are introduced by qME. An object is used to group vertices and triangles in a model. For example an ax is an object and a monsters head can be an object etc.
When importing an original Quake model, the model is scanned for groups of triangles that are separate from each other. These separate groups of triangles are turned into separate objects.
Objects are organized in a tree. At the root there is a root object that has other objects as its children. Applying an operation like scaling to the object will apply it automatically to the objects underneath it.
Pivots are kind of like bones for your model. For every object there is a pivot in every frame. This means that when you have for example 10 frames and you have 3 objects, you have a total of 30 pivots.
A pivot controls the position and orientation and scale of the vertices in your object in a single frame. Just like bones are attached to flesh in your body. Moving a pivot means moving the object in that frame and thus the vertices.