Here ya go:
https://blenderartists.org/forum/viewtopic.php?t=29451
scrolling, spinning and seperate alpha with seperate texture for the realtime game engine. A python library that is at the bottom or popup instead the top menu. Alpha scrolling and spinning with seperate texture for certain effects
I would like to see Blender recognise relative folder structures rather than absolute.
eg of absolute: c:/documents/pauls documents/destop/bla.blend
eg of relative: /desktop/bla.blend
I sometimes have to transplant my project to another machine, and then have to reattach all my textures because blender does not recognise the new folder structure environment.
You should pack your textures: it increases the size of the blend file, but this way you can import/“transplant” as much as you want 
On my wishlist is nothing too fancy, no high-end feature, just enhancement/improvement of what exists currently:
-shadows for particles (static/dynamic) toggable with a button
-halos taken into account in reflections/refractions
The only fancy feature I’d like is: for static particles, the possibility to make the particles “real”. I mean that the keys of the static particle is used as a path along which is extruded a n-sided polygon with a base dimension and a end dimension. What we’ve got then is… a true makefur or makehair plugin! I hope, then, that the Particle interaction options will still be available for the ‘path of keys’ so we’ll have animatable hair and fur using reflectors, force fields and vortex 
Ok here’s something that would be usefull, but more for coders.
- when you open a script in blender, and do some changes inside blender, you can hit a save button and saves the changes into the external script file while is called like “fiber.py’” or something.
- copy and paste function would be really usefull, like copy a bunch of text outside blender, and paste them inside blender, or inside blender copy and inside blender paste.
- Creating a script inside blender, and able to save it from blender to file with customable extension or from selection support.
- Coloring parts of the script.
- Error logger, the errors get logged in the console, but sometimes a script can crash blender and you won’t be able to find the actual problem. It should be handy to have a function to export the error to an error.log text file. I did this by my own when scripting scripts for the game engine. Really usefull.
These things are really usefull, like copy and paste, and saving changes to the external script file from inside blender. Most of the time I have to switch back to text editor outside blender to change the actual script file, and load again into blender. This waste a lot of time when creating a huge script. :-?
I would love it if the animation aspects of blender enjoyed a much more obvious part of the gui. I think if it was more intuative we would see more animations.
I also think that the mouse movement access to the move scale and rotate comands should be dropped and the box selection should be given preference(it took me about a month of blender usage to realize I had to press b. Until then I thought the feature was missing). Then again some people may love wiggling there mouse to make stuff happen.
I also find it anoying how when unrwapping something you don’t see the wire until the poly is alread selected. There’s probably some way around it, but I’m not aware of it.
BUT MOST OF ALL
I would LOVE an option to turn on dispacement mapping in the viewport, and also the option to convert the displaced mesh to a mesh… And to make that really kick arse; it would be great if vertex colours were available as a map. All of that combined would pretty much mean blender would have a z brush… Blender doesn’t already have a z brush… does it?
I’d like to see a better python scripting interface to make it more like IDLE (The editor that comes with python).
1)Color coding of certain key words.
2)An autocomplete function that displays a list of all possible words when you press “.” The way Blender’s python interface is currently set up, I’m having to do “print dir(Blender.NMesh)” to see lists of available methods. Autocomplete I think is the most important aspect here and would eliminate trivial bugs caused by typos.
3)Colored bars in the margin that show visually where an IF/FOR/WHILE statement starts and ends. I know Python relies heavily on indentation, but it’s not always easy to follow the indentation- especially if there more levels of indentation within an indentation.
4)Collapsable code regions like .NET has to allow for minimal cluttering.
I have a few suggestions:
-
Allow selecting joints in editmode without having to select the endpoints. Even if the selections in posemode were retained when transferring to editmode.
-
Have different select modes like edge and face select.
-
Micropolygons everywhere. In things like NURBs so you don’t have to set the UV resolutions. Most importantly in displacements. I hate having to subdivide my models up to even the limit of 6 (not sure why there’s a limit either) only to find that the model is still jagged.
I think what it needs includes changing the density of the subsurface based on the intensity variation of the displacement map. Areas where the disp map has similar colours don’t need a lot of polys.
-
More flexible materials, DOF, true deformation and transformation blurring, caustics.
-
Perhaps a choice to save in ascii format. If blend files do ever get corrupt, ascii files are much easier to fix.
I won’t add cloth and fluid dynamics to the list as they’re fairly complicated.
If there was integration with a Renderman compliant renderer instead of yafray, that would fix 3 & 4. I know there is a Renderman-aqsis build somewhere but it’s not official.
There seems to be a high dependance on matching Blender versions with yafray versions. At least with Renderman, both renderer and exporter try to match a standard which remains largely unchanged. Also, it means you can use a whole load of Renderers that support Rman shaders - 3delight, aqsis, bmrt, prman and more. Pixels3D does things this way.
Yafray still has shaders defined by fixed values so it won’t be able to recreate anything but programmable shaders can and they can also be exchanged online. So, if someone discovers how to make perfect glass, everyone can get the shader. There are RMan shaders all over the place:
http://www.renderman.org/RMR/RMRShaders.html
Even when certain features aren’t finished, you can sometimes use programmable shaders instead like for area lights, caustics or faked refraction.
Now, if the Blender material editor acted as an interface to making RMan shaders, people who weren’t familiar with programming shaders could work as normal.
I’m all for a better Blender workflow with more options for setting up links to specific saved projects file locations and more appending and datablock linking with options for multiple data control features across scenes.
I would like to see more features accessible via python. And I would like to see more developments in the area animation management features, dynamics, better morphed object animation library control, more mesh editing level animation tools. And character animation improvements such as dynamic properties for clothing, NLA and hair, etc. More controllers here , more controllers there, more controllers everywhere. Hehehe.
Since I make heavy use of Blenders modeling tools I would like to see improvements in vertex, polygon, edgelooping, translation, instancing and selection features. It would be nice to see some kind of Renderman compliant features as a standard part of Blenders interface with access to shader scripting etc. inside of Blender.
An improved more wonderful Blender scanline renderer that will bring Blenders scanline production and quality up beyond the best scanline renderers out there. Hehehe a little dream of mine.
But I will settle for whatever the marvelous coders come up with. Hehehe.
Blend on! And be marvelous.
I created a feature proposal (that gets biger and bigger per week):
Grouped by category, ordered by importance.
Particles:
1 - Set particles to be able to cast a shadow, and make the shadow darkness dependent on the ‘thickness’ of the particle cloud at that point. There is a workaround for this at the moment, but it makes animations impossible.
2 - Enable toon shading for particles.
3 - Particle self-interaction. I want gas shooting into a cloud to create a ‘mushroom’ effect, and smoke that hangs in a thick layer on the ceiling.
4 - IPO control of ALL emitter variables.
5 - Particles in-game engine. Painkiller can have it, why not us?
6 - Forcefield effect based on vertex and face location. Ie make a sphere from each vertex, or outwards along the normal of the face.
Game Engine:
1 - A GLOBAL PYTHON CONTROLLER FOR THE SCENE, executed once per frame (Flash “EnterFrame” style). Swarm AI is so goddamn hard to do when it’s localised…
2 - Up the lamp limit. 8 is so tiny.
3 - Investigate the possibility of using card-specific speed increases, like the nVIDIA “CG” shader language. In fact, do anything to increase speed.
4 - An INBUILT movie player, callable from python. Have options for scaling, whether to have letterbox or not, whether the user can cancel the cutscene, etc. FMV-ed is cool, but sometimes I don’t want to have my games fullscreen, and to have the movie then play fullscreen is rather jarring.
5 - An option for picture-in-picture. You could call the python controller, specifying the TL/BR co-ordinates of the PIP window, and what camera it is taking the view from. Could possibly work with request 4.
6 - More python controls. I want to be able to modify EVERYTHING in-game via scripting.
7 - Fix the “whitefaces” bug in shaded and textured mode.
8 - A python control to change the game-framing setting ingame. Ie so the user can choose their preferred resolution.
9 - Ragdoll? 
j/k I won’t need this for a few years.
Documentation:
CLEAR, DETAILED, online docs, not recycled 4-year old documentation. I do not want to be spending hours on forums and Yahoo trying to get an answer to the most mind-numbingly simple problems.
(To anyone who has bought the book - is it any better than the online docs? What version does it go up to? What’s on the CD?)
Modelling:
1 - Cloth/Softbodies. SOOO vital to some stuff i’m working on now…
2 - A seperate window for vertex key definitions. Flat lines in the IPO window are dodgy indicators.
3 - A “mark Antismooth” feature. By this I mean you select edges that you want marked as Antismooths, and press the “Mark Antismooth” button. Any edge marked as an Antismooth would NOT have normals interpolated over it, even when both faces the edge is attatched to have setsmooth on. This would be equivalent to smoothing groups in Max,
4 - Fiber-Gen/Maya 6 style hair. What I mean is, having the hair with particles, ‘combed’ by Bezier/NURBS curves.
5 - MetaNURBS. We crave the ubersmooth!
6 - More primitives. Max has ~15.
Mesh:
1 - Conversion back to NURBS. Set each vertex as a control point with max weight. The user can edit from there.
Sequence:
1 - Allowing Effect strips to be resised. It sucks having to split inputs to make things work. (This would involve an auto-detect feature that disables the effect strip when it is not straddling two inputs, turning it striped or something similar.)
2 - Being able to combine sound and video in the one file w/o the need for an outside tool.
3 - Autodetecting which strip should come first in a transition effect. I don’t want my second object suddenly appearing, fading back to the first strip, then suddenly reappearing again.
Materials:
1 - Having a switch to shuffle texture channels order up and down. Shouldn’t be too hard to code.
2 - Create a ‘cloudy’ option for raytracing. Ie more light gets absorbed and messed around the further the ray/photon goes through the material.
3 - Setting the Toon diffuse shading default to ~1, not down so low that you have to read Aner’s tute before the shader becomes useful.
UV Editing
1 - AutoLCSM, 3ds MAX style.
2 - Having UV mapping affect more than just colors - I want UV-mapping to be able to affect what procedurals can.
3 - Alpha maps for UV. Wouldn’t be that hard at all - simply require the user to create an image with the same name in the same folder as the source but with “_a” following the name and preceding the extension.
Objects:
1 - Adding more effects to the Effects panel, namely glow and trails. Glow should be a volumetric effect, with options for fadeoff (linear, exp, root, sin/cos etc), initial intensity and distance.
‘Trail’ would be like … ever played Gunbound? Watch the “Mage” mobile move. Another example would be the ‘electric punching’ effect in the FF7: Advent Children E3 trailer woods battle. Or weapon slashes in Soul Calibur II. Options would be life of the trails, style (glow or Vect), falloff (same as Glow), and material.
General:
1 - Better aligning tools. I want to be able to align to one axis only.
2 - Name-a-bilty. I want to be able to rename layers, vertex keys… (layer names can be shown in tooltips)
3 - A better .B.blend file. Include a four-way split screen, a Python screen (LHS 3D, RHS Text/Scriptwindow), one for UV Editing(LHS Potatomode 3D, RHS ImageWindow), possibly more.
4 - Global Time Scaling - being able to define a new framerate, and have the IPO timings scaled to suit (like Magpie).
5 - MP3/OGG support. Esp ingame. Distros become a lot easier compressed.
6 - An automatic speech analyzer that makes a phoneme list and applies it to a specific object. Request 2 would have to be implemented first.
7 - C4D-style Object Expression tags. Simple algebra-style text that creates a lot more versatile constraint system.
8 - Tooltip on the “Quality” slider in Format Buttons (F10) to indicate which value means best compression and which means best quality.
I hope at least half of these will be addressed by the next release.
After just getting 2.34 I have this to say
-blender needs a chamfer function, would be good for rounding
-weigh subdivision by verticies rather then edges
-per-pixel weighting (similar to what Z brush has with displacement)
-dynamics in animations or for setting up physically complex stills (ie. a pile of boxes not stacked)
-have a robust polygon reduction tool that would be way better then the decimator
-true face select
-option to leave verticies that would not be visible invisible during editing.
and if possible
-fuse Yafray into the blender internal renderer.
EDIT: almost forgot
a tool that makes meshes as clean as possible like making quads where possible, rearange faces, turn crossing lines into face edges, erase overlapping faces ect…
im a total noob to blender… so i dont know all its features… but neing i AM a noob… i can offer some pointers on making the system more user friendly.
The universal UnDo… that was my first insistance but someone already hit it for me… i dont know how many times ive gone through a “test” change and didn t like the result… only to have to reload my entire file to undo it… blender is a proggie of trial and error… and no undo other then edit mode is a BAD thing.
a setting for colision detection WHILE editing. i know it exsists in other programs… i want to be able to grab a few verts… and be able to set it to detect OTHER objects… and mold/wrap around them… it would make animation SO much easier when it comes to getting characters individual fingers or tongues to wrap around things instead of spending 20 min to make sure that each vert is curving int he right direction.
Unfortunately you shouldn’t overdo the undo in a program, you have an undo that covers EVERYTHING in the whole project like in other apps and it takes a ton of memory from your computer.
One more thing, make a better converter to convert tris to quads, the current one is a mess, do this.
-subdivide the default cube about 2 to 3 times
-convert to triangles
-convert to quads
instead of seeing a nice grid of quads like you would expect, it mixes in some big four vertice triangle shapes and some triangles are still there.
In 2.34, you should be able to do this by pressing the ‘relative paths’ button in the file selector when you’re loading a file up (like a texture, etc). It should then store it relatively, within the .blend file.
sorry if this noob write here , too…(my suggestions ar charactrer animation related)
-
world space coords joint pinning (glueing the joint to world space wherever it is, and having IK act differently from that moment(that is, according to the fact one joint is “glued” in the point it is in certain frame.)) .Toggleable at any frame. This without the need of making several skeleton/branches but a continuous armature (needed by several game engines) or the need of using empties or other objects.
-
Ability to always do FK in a branch that already have IK solver applied to its termination, or any other constrain. That is, to rotate freely in any angle, or better said, just as if it had no IK. (like in Character Studio)
-
If it’s not possible to do pixel based weight painting(like Deep Paint 3d texture 3d painting, or Zbrush when in the texture painting mode, not in vertex paint) instead of vertex color based one (in low polygon models the painting is more coarse) , I’d then suggest a better workflow in weighting proccess. I know there’s an script for some of this already.
-
A “ball” gizmo (think about Maya’s, 3 rings, each for each axe, forming a “ball” in place, usually at the pivot point, the joint, of the bone) for FK rotating the bones(and as I said, allways rotation be possible, at least when only ik solver are applied to that chain), it lets you pose the character with a lot of accurateness. Indeed, imho, this and planted key (joint pinning that I explained in the first suggestion) are my favourite features in Max Character Studio.
[ I don’t know if being as noob (in Blender) as I am, am allowed to put suggestions here, but well…
I know they all sound to sci-fi, and may not be near to possible to do in Blender with the existing code (Harkyman told me about it) but as this seems to be a vault of suggestions, posted here. Sorry if they’re to verbose, I wanted to explain well. I’m using other package now, but I think these would really benefit Blender. ]
Congrats to all Blender coders for this huge effort.
-
Edge and Face modes. For both selecting and modelling.
-
NGons. I’ve read the arguments for and against these and I just don’t buy the against argument. Restricting faces to 4 vertices forces the modeller to deal with sculpting on a lower level than is necessary. Any experienced modelling knows the pros and cons of many sided poly’s and will work accordingly, eradicating these towards the final work. While reaching that final work it is a great benefit to allow NGons to exist to aid workflow.
-
Better fill alogrythms. I make a bunch of edgeloops, select all verts, shift+f and I get a big mess full of Tri’s. Quad’s rule. Death to Tri’s.