And about the usefullness of the retopo tools, if you are not going to sell your models (many employers don’t care about the original base sculpt, but they do care about a low-poly mesh + normal maps + displacement maps + textures + other maps/textures) or you are not using them as a base for animations (like Sintel) or gaming (like many of Endi works), the probably you can just ignore them. Not everyone needs them, i agree, but still are needed.
But whatever goes in or not, Congratulations to the Blender Foundation for being accepted in the GSOC again.
improve visualization so that with xray on the backfacing faces are culled; similarly backface culling for wireframe see this video for clarification of the problem - http://www.vimeo.com/8790166
‘slice and connect’ retopology method (similar to that offered in 3d coat)
All the ideas are really cool. I personally like the UV mapping tweaks and the retopology tools - that would be amazing.
I do feel though that the BF should take some time, especially as 2.5 is getting out of ‘Beta’ phase, to finally get B-Mesh into trunk and work out some plan to get the render branch finished. Funky features are great, but Blender needs to concentrate on getting the basics down a bit. Modelling and rendering have been ignored too long.
I know Ton says he wants to concentrate on that, but progress so far on each of these fronts looks like it needs a lot of help
we are done with beta by siggraph at the latest, probably a lot sooner. Everything can begin merging to head after that.
I know that revising the shader system is one of tons next targets and that the Mango project will need photoreal rendering. It isn’t clear to me how far off bmesh is from mergable - it always seems like a month more of work
i hope they work of fixing/developing the plugin system for the video editor. keyframeable effects with a good plugin system for developers would help replace half of the uses for after effects, which is the only program i’ve missed since moving to all Linux. The video editors effects are mostly all keyframeable now and it’s awesome but there isn’t a working plugin system in 2.5 so there are only a few effects.
libmv did not get accepted as a seperate GSOC project? As far as I remeber I have read something about an application from their side. Are there plans to accept a libmv integration project as a Blender GSOC?
I partially disagree.
I am not sure that everybody knows Re-projection is in 2.56.
Or edit externally button in Image Editor, that have the same kind of effect but with Image Editor angle.
Blender have good procedurals, Texture nodes and there is Textures_tiles patch.
I would like to see ability to easily rotate mapping/texture space and Source Image in texture mode in trunk, too.
But I am not sure that merging this in trunk would be enough for a GSOC.
Complete Scalable Texture support is very promising idea.
Except this, I think polishing texture workflow should be done with a re-design of Image Editor rather than with new tools.
If it could display textures as layers and stencils as masks and could be used for re-projection; Gimp would not be often called.
I am not sure that it is obvious for everybody.
But Blender can potentially use all select tools from edit mode for UV Unwrapping.
So I would appreciate to read small descriptions of requests to understand what to look at in videos like it was done for retopo.
We can make textures with Free Open Source Software.
We can’t make tracking with Free Open Source Software.
Even if it is obvious that efforts will be made on Libmv integration, it is good to remind this.
So, I would give the priority to Camera Tracking Integration, Relationship Editor and things that blender can not do.
I would like to see new compositor quickly finished to be able to use others projects like phonybone’s nodes branch.
Probably, linking for modifiers is to important for a student.
Maybe, they can create new modifiers :
I would like to have dupliverts, dupliedges, duplifaces, particle display of dupli group in one modifier.
Particle Instance modifier need a Particle System.
Maybe, it could be renamed Instance and give choice to use object vertices with an “modifier requiring original data” warning.
This warning used by multiresolution could be used for new modifiers.
But I suppose that modification to modifier stack is freezed because of phonybone’s branch.
It is not easy to say what little improvements can be added when bigs have not been totally done.
Maybe, some todos and blenderstorm ideas could inspire others GSOC improvements.
Integration of libmv would be a huge project, and I know it been rejected before because it’s too complex for Gsoc.
I’d like to see polishing of the texture painting workflow aswell. Also, new functionality like for example Layers, along with speed improvements (GEGL, OIIO integration maybe?) would be extremely useful but would require a lot of skills from the student…
Simple way to resymmetrize. Right now you have to cut the model in half, in Edit mode, then glue it together by hand, and fix by hand the seam in the middle that gets distorted. (also if there are layers, some symmetric and some not, this would not be an appropriate method for keeping the symmetry only on the symmetric layers, they would all become symmetric.)
I like the idea of the paint tools capabilities to be streamlined, and a complete real-time GLSL preview in the viewport of all maps, including bump, specular, and shadows, in high density models.
uv tool improvements
symmetrical seam marking
symmetrical layout of mirrored geometry
symmetrical selection, etc. in the uv editor
overlapping of mirrored geometry (so that mirrored parts of a mesh can share the same image textures)
improved stitching and ripping operations
additional selection operations such as selection of the matching seam
Talking about symmetry, I think there was a problem with normal maps and the Mirror modifier. I believe it was that the mirrored side (3D geometry) uses the map exactly the way it is in the UV map for the original side, so the normals are not mirrored as they should.
Finish the Material Presets we were going to build in to blender. I would love it if it sensed if you had yafaray or lux installed and tailored the materials for that as well. Even you die hard “have to do everything myself” folks would at least start with a material that was close to what you wanted and tweak if you had it. It would save ALL of us a TON(pardon the pun) of time. Also this could be done entirely as an add on so no worries that it wouldn’t get merged in to trunk. I have thought about doing this add on myself, but I haven’t had time and didn’t want to step on anyone’s toes if it was still a project somewhere.
I think texturing/retopo are two areas that are meat and potatoes type improvements, they are important in a general modern workflow, hopefully some students pick these areas. To go beyond UV unwrapping/external image editors and do all texturing directly on a model would be pretty cool, I think thats the next big leap, but probably too big for a GSOC! Speaking of which, I thought someone was working on a layer system last year…
in some points i agree with you but retopo looks nice right now, i proved it long ago and work perfect to me
The texture System is a good point , layers for sculpt and painting would be fantastic
