manuelbastionilab header image

[Back to development log]

More details in roadmap for lab 1.7.0: new expression system, new GUI, new topologies and anime characters.

18 March 2018

ManuelbastioniLAB 1.7.0 will be a rich milestone release. The goal of this release is to greatly increase the quality of lab models, adding new models, new topologies, new GUI, better expression system and more. This brief article quickly introduces the main points of the roadmap.

New humans and anime models

One of the most challenging feature is the implementation of a complete tool to create anime and manga. It has requested many months of work in order to identify and understand the key features used in Japanese illustrations, evaluate their feasibility as tridimensional models and classify the main types.

The result of this research is an impressive modelling work to do: 140 anime base models to cover five base styles, implemented using new meshes.
Lab 1.7.0 will include the first two styles, that I called "Stylized" (the standard style often used in many anime) and "Realistic" (a more detailed anime style often used in games and male-oriented anime). Total: 56 anime models to do for lab 1.7.0.

Also, since this is the modelling-focused release, it was decided to implement the new generation of human base meshes too. The new human topologies will greatly increase the details and the anatomy accuracy, but as in previous models they will not require a CPU brute force: the models will be optimized for subdivision-surface, preserving the quality and still being "human readable": in other words max quality with low poly count. A couple of preview images from the gallery shows the quality of new models: torso muscles and human ear.

New GUI and creation workflow

This great number of new models will also require a different phylosophy in the design of interface. In the current lab the user selects the base character choosing it from a list of available models. Clearly this will be no longer possible in 1.7.0, because it will include more than 60 different base meshes. The new GUI will be more intuitive and simple: it will just ask the required attributes: male or female, overweight or underweight, human or anime, etc. Then the system will generate the character according to the selected parameters.

Optimized topologies and UV projections

The model topology will be optimized for the selected character. For example the mesh for an overweight human will be simpler than a mesh for a bodybuilder human, because it doesn't need the same amount of muscle details. This approach will produce better results with less polygons and also will automatically use the appropriate UV projection and textures. On the other hand it will limit a bit the range of morphing, since the system will automatically prevent the creation of unoptimized meshes: it will be not possible, for example, to start from an overweight character and morph it in a bodybuilder.

Experimental releases

New topologies also mean new high quality textures, improvement in shaders and new morphs parameteres remodelled from scratch. Obviously, this big amount of modelling work will be gradually distributed in versions 1.6.2-1.6.9. The new models will be released version by version as "experimental-incomplete" features, since they will have only the basic rigging and morphs. Advanced features as automodelling, measures, IK ecc will be gradually added in order to reach the completion in 1.7.0. All 1.6.x versions will use the "old" GUI, so during this release cycle the user will see a long list of characters: as said, this is a temporary solution and a more usable GUI will be implemented in the final 1.7.0.

Face rigging and shapekeys

Another important node in 1.7.0 is the technology that will be used for face animation. Shapekeys or face rigging? After a long discussion on the social page and the Blenderartists thread, the current decision is to implement both. The main system will be the rigging, with an option that will bake the face poses as shapekeys. The evaluation of feasibility is still in progress: the determination of number and placement of feature points, the way to add details and other technical issues will require some experiments and new features to code and test. Another thing that is still under evaluation is the use or not of Blender drivers. The support of after-creation expression tool is also under evaluation.