[Bf-committers] EU have a dream? Blender in 2012!
letterrip at gmail.com
Sun Feb 4 15:08:18 CET 2007
Certainly worth pursuing in my view.
Where Blender should be in 2012 is a worthwhile and interesting
question, it is crazy to think that is only 5 years off.
We have need to improve in breadth and depth of our tools; and scale
our ease of learning and ease of usage from total novices to the top
end professionals of a variety of fields.
SubDivision based - ngons; tool piping; improved robustness in general
- with Briggs work and the tool api refactor - that will bring us a
long ways in achieving generalized improvements in tool flexibity that
are desireable. This includes a richer functionality for
Sculpt based - there are lots of great ideas for improvement - Silos
allowing of editing the base model and generally good edit tool
integration; a number of interesting ideas from Mudbox and ZBrush.
For organic modeling it seems pretty clear the industry is seriously
picking up momentum for this.
UV unwrapping - brecht keeps us on the cutting edge here. Switching
to editmesh (or its successor) so that a richer toolset for
manipulating the mesh after the fact is desireable. Better full
automatic unwrapping so that naive users can 'just paint' without
having to learn tools is probably a longer term goal. Asset transfer
in general (including UV maps but also animation related information
etc) needs to be addressed.
Paint Based Texturing - I know that brecht has a number of ideas in
the near term to improve texturing. We should consider vector based
texturing in the longer term so that textures can be resolution
independent. We should also consider better integration with the open
source painting and texturing tools. Possibly look at GEGL and other
dedicated paint related libraries. Layers and other more advance
functionality should be considered - ie do we want to be a legit
competitor to BodyPaint 3D?
Procedural Based Texturing - I'd suggest we have a look at the various
procedural texturing tools - specifically DarkTree and Genetica but of
course also the major 3D packages. Right now I suspect we have things
a bit too hardcoded per texture for the non node based textures, and
for nodes there are probably quite a few potential parameters we
haven't addressed. Again there might be some really interesting stuff
Animation - thus far we have completely ignored motion capture - while
it currently is mostly larger studio - in five years time facial
motion capture and possibly even body motion capture could be common
at all levels of production. Another big area we've only scratched
the surface of is AI based animation. Here are some things I suspect
will hapen 1) crowd AI will get a lot more sophisticated 2) AI might
start doing a lot more of the work on main characters and secondary
characters. Other areas of interest is control systems - we have no
real method for taking advantage of MIDI input - which can be very
rich control surfaces. Also things like doing previz using more
sophisticated game controllers is going to become prevalent - possibly
even doing digital puppetry for significant portions of final or close
to final animation.
Of course there are many other more basic animation things like
improving our skinning, piecewise interpolation for curves, etc.
Mostly 'catchup' type work.
Particles - jahkas work will represent a tremendous improvement. What
further developments will be needed after he finishes (asside from
volumetrics) are unclear.
Simulation - a more unified system, improved baking and caching -
selectively resimulating parts of a simulation. Threading for all of
Rendering - in five years time a lot of the rendering could be done
using sophisticated gamengine. Thus I think Ogre as an alternative
render engine as well as general integration is going to be important.
For the existing internal engine there are obvious areas of catchup -
in particular volumetric rendering is a clear that we need to address.
Speed and memory consumption improvements will always be in demand,
and with hardware speed increasion people will want to use more costly
features such as caustics and SSS in animation renders.
Compositing and Non Linear Editing - there are significant
opportunities in both those areas - if we choose to head in that
direction it depends on how broad or narrow we want our targets to be.
Ie if integration with live action is needed/desired than things like
matchmoving and camera stabalization will need to be addressed.
Scripting, APIs and Plugins - python is popular, but should we open
things up to additional scripting? Also where do we want plugin
Customization Education and Learning - Script recording and playback
(ie integrated tutorials), remote guided/interactive learning with
Verse, Migration tools. Pipeline integration. We need to be as
easily learned and integrated into others pipelines as possible.
Of course that ignores other questions such as - what markets do we
need/want to be a major player in in 5 years? Ie do we need to have
NURBs focused on for CAD users? Do we need better integration and
output for illustration users?
So those are my thoughts of where we should be heading over 5 years.
More information about the Bf-committers