Chunk format 1 video (found in The Vampire Diaries) contains frame IDs as well as frame names. Those were previously skipped over, but are now getting read correctly.
AVF videos with chunk format 1 do not contain the frame size in bytes like previously assumed. The data that I previously thought was the size is now simply skipped, and the size itself gets calculated on the fly.
Added generic methods for getting the pixel format and transparent color. Also added a method for loading an image directly into a ManagedSurface, which also loads the palette and lets The Vampire Diaries logo display properly.
Added support for chunk file format 1.0 (used in The Vampire Diaries), which uses frame names instead of ids and has a different id string. Right now the image gets decoded upside-down, and the palette is still unknown.
Nancy1 adds exactly 12ms to every frame of a secondary video, making it visibly slower than what the current implementation was playing. I'm not sure where exactly that happens yet, but this commit adds a (slightly hacky) fix for the problem
Rewrote most of the engine using a much more object-oriented approach and using more of ScummVM's common classes. This design deviates quite a lot from the original engine's, but should be more maintainable and extensible in the future.
Added code to make AVF videos seekable. Not tested on video type 2, but will almost definitely break it. Also updated class definitions to properly mark virtual methods.
Eugene Sandulenko
Orgad Shaneh
fracturehill
Walter van Niftrik