Page 1 of 2

Google Summer of Code 2009 Ideas

Posted: Fri Jan 09, 2009 5:29 pm
by Hos
Hi gang,

Below is a reposting of an email that I sent to bf-committers recently. If you have project ideas for Google Summer of Code 2009, you can either add them to the wiki page if you have the proper permissions, or post them in this thread and I'll try to add any reasonable ones to the wiki page on a semi-regular basis.



Leslie Hawthorne of Google has announced that her team has received some
funding for Summer of Code 2009, but it will probably be less money than
last year (meaning fewer students and fewer organizations involved).
While there are no guarantees that Blender (or any other organization)
will be involved, she has asked that organizations get the word out, and
start the search for would be applicants.

To this end, I've started a first draft of the "Ideas Page" for this
year. Basically, I've removed all of the ideas from last year, and
just put in some bogus placeholders, since the Blender 2.5 refactoring
will probably negate a lot of the ideas from last year. I think that
developers should start adding new ideas for projects to this page
now, and in March we can do a review of the accumulated project
ideas and pare them down to a reasonable set.

I've also added a section called "Requirements for successfully
completing a Summer of Code project", which lists the bare minimum
required to not fail the program. I've also done a small revision of the
application template, mostly to ask the applicant about any previous
work done with our bug/patch tracker.

The link to the new ideas page is here: ... 2009_Ideas ... mplate2009

The link to last years pages are here: ... 2008_Ideas ... n_Template


Posted: Fri Jan 09, 2009 6:58 pm
by pildanovak
My Idea:
Integration of libmv in Blender workflow to help Blender initial tracking support and libmv more attention from outside.

libmv is an opensource tracking library which is developed alongside and for blender by Keir Mierle and others.
home here:
Libmv developers targets are mainly camera-motion estimation, but tracking uses in CG are many. The purpose of the project would be a good design of a future interface of the tracking tools in Blender and inventing the workflow, adding UI code e.t.c. The practical feature result could be e.g. 2d tracking for 2d compositor(as found in Shake or Nuke) for use with moving masks and others, but if libmv moves to some advanced testing phase until then, it could be possibly implemented to do camera motion estimation. I don't know what the devs of libmv would think about such an idea though.

Posted: Sat Jan 10, 2009 7:20 am
by Hos
Hi pildanovak,

Reading the libmv site, it sounds like they are already trying to integrate it with Blender. Are you familiar with the developers, or know the status of this integration?


Posted: Mon Jan 12, 2009 5:18 pm
by Wahooney
Scripted Modifiers...

My job here is done.

Posted: Fri Jan 16, 2009 6:42 pm
by pildanovak
since this thread didn't flood with ideas(I wonder why), here's a couple of others, this time more realistic probably(the libmv idea was discussed as not too real for a student):

-Good working inset tool(macouno did a script for that).

-Better decimate modifier.

-Lod modifier - a modifier which can have n levels of lods, which are changed based on distance and on the editmode, viewdraw, render flags.The lods could be
a) autogenerated with some good decimation algo, with keeping uv/vgroups,
b)put in as linked meshdata - so the objects meshdata changes with changing distance.
Lod doesn't have to be a modifier only, it could work as a tool for whole hierarchies, so you can have lodded several objects together(like a house made of several meshes.)The thing should work with linked libraries, since memory savings are one of the priorities of such project.

-ARToolKit support in gameengine - already done as a plugin by ashid, could be put into engine and have own logicbricks. Blender wii!

-Porting some of the great methods from TopMod as modifiers:
there's plenty of great original ideas in this software, but it's not a production package. Using the algos as modifiers could make them useful in many cases.

Posted: Fri Feb 13, 2009 2:29 am
by ysvry
what i would like to see is blender as 3d webbrowser and editor.

Posted: Fri Feb 13, 2009 8:42 pm
by jyhuang
My idea: interactive image-based rendering

I was impressed by the illumination brush: ... r_embedded

Blend supports hdr file loading and rendering. However, it'd be nice if users can interactively correct certain rendering details after the initial rendering is done. Information such as lighting, BRDF, vertex positions is all available, therefore we can pre-compute the radiance transfer and store it to blender files.

Posted: Mon Feb 23, 2009 5:52 pm
by Hos
Thanks guys, keep them coming! (Both here, and directly in the wiki.)

The organizational applications start in about 2 weeks, so hopefully we'll be able to have the "Ideas page" updated with some of these ideas by then (after some developer review).


Hide backfaces in Xray mode

Posted: Sat Feb 28, 2009 1:37 am
by Azrael
A few have posted this one, hide backfaces in XRay mode (very useful for retopologizing a mesh) and IMHO a necessity when the retopo mesh is heavy.

Posted: Sun Mar 01, 2009 4:33 am
by dfelinto
pildanovak wrote: -ARToolKit support in gameengine - already done as a plugin by ashid, could be put into engine and have own logicbricks. Blender wii!
I would love to see that as well, but IIRC ARToolkit license is not 100% compatible with Blender.
It's free for non-commercial projects only.


Posted: Thu Mar 19, 2009 12:56 am
by IPBurbank
I have been hearing some buzz about blender including openCL which could greatly decrease render time. Here is the official site of openCL

This would be nice for rendering, the sequencer real time preview, and possibly it could be used for other tasks such as fluids and particles.

~not sure if this is a student project, but I figured I'd mention it anyway.

Posted: Thu Mar 19, 2009 6:43 pm
by nqe
Would expanding the current export and import plug-ins for the COLLADA format to include armatures and animation be a too small/big project?

Posted: Tue Mar 24, 2009 2:56 am
by razor85
Hi there, i was in the GSOC last year and i must say the program is really nice for us students. Last year i worked with Audio integration in Worldforge and this year i would like to work with Blender, most specific on the improvements (FBO, VBO) in the OpenGL core. How are the chances of this project getting accepted and in general for blender community, the importance of it?


Posted: Tue Mar 24, 2009 9:39 pm
by SirDude
nqe, I would say it would be too small in general. I do not know exactly how much work it would be though. I'm guessing it wouldn't be very hard.

Just you asking though probably indicates that you haven't really looked into it enough to really decide if its big enough to be a project.

From what I've heard the collada stuff needs some love. If you were to look at it and determine what exactly needs to be done, maybe do some test files for it and document all of the changes you would like to do to it, then that sounds more like a project. :)

Posted: Tue Mar 24, 2009 9:44 pm
by SirDude
razor85, I have a similar responce for yours, not enough details to really determine if what your thinking is good or not. Blender's GL stuff is pretty old and could use some work. If you were to do a project improving things it would need to insure that things still work on all of the platforms.

It's an area that needs work but you need to put some sort of a game plan togeather before we can really say ok this is a good project...