Google Summer of Code Wishlist
Preparations for this year’s Google Summer of Code have begun and the organisation is asking for your input!
Hi all, I’m working on our GSoC application again this year. Part of that is a wishlist.
Would be interested in where you think Blender needs the most ‘polish’ still for your daily work needs.
When you respond do ‘Area – Suggestion’
Ie ‘Sculpt Modeling – faster multires level changing, higher poly count, faster smooth brush’Here are the categories. You can also provide links to relevant research papers if you are aware of any. Note that these won’t necessarily go on the ideas page that we submit for gsoc, but they will be added to the pool of things we consider.
Link
- GSoC Wishlist (BlenderArtists)
Integrated user-fed mesh/material/whatever library..
+1 to that!
I hope to get something like spotlight in zb.
I found this video of “vpaint” but i don’t know if it will be implemented
http://vimeo.com/13812115
That’s my #1 as well: PTEX / Vpaint. Hopefully now that Bmesh and Cycles are integrated, that can be possible.
2 and 3 are fixing hair/fur animation and bringing back reactor particles through the new node-based particle system.
Reactor Particles, yes please. Also with blender 2.5 there was a promise of a node based particle control, I’d like those two very much
and this is NOT in the trunk because….”?”
It’s not in trunk because it only works with VPaint and Ptex. It’s of limited usefulness with VPaint (would be more useful with texpaint, but it’s not clear to me how to code it for texpaint), and Ptex isn’t in trunk.
You’r the one who code it right… Nice work, I’m sure that all the blender community is proud to have persons like you in here =) I’m talking as an artist, and for me it seems much more intuitive way of texturing things, I may tell you that the only thing that I do not do inside blender is texture painting, so i go to zb doing the painting, then I go to blender again and bake the color information of the high poly models. Can we just have that and then bake the color information into a texture?
I’h love to see a system similar to Cinema4D’s XPRESSO. I think there was something similar put together a while ago named nodes for/on everything or something. Saldy it’s been a long time since I heard or read anything new about it..XPERSSO is a node based system that creates all shorts of objects interactions and it’s one of the few reasons I still use Cinema 4D slightly over Blender today. Since a node system (for compositing and materials) is already there, it would be super nice to have a similar system in Blender. Simple but with endless posibilities!
Not exactly essential, but i would love to have some destruction features. Something similar to ray fire. Phymec tools looks great, but i would like to have something a little more official in development. This would complete blender for me personally :)
Ton Rosendaal plushie Toy
How about a drag-and-drop material library that is connected to an online repository to easily upload and download new materials?
Sub surface scattering for Cycles would be nice, but maybe that is too difficult to implement properly for a CS student?
library with online repository: +1 (integrate model repositories as well?)
I was thinking about what this would entail a few months ago, definitely not an easy task. And it would be better if it were instead reach out to existing material websites, which may or may not function on add revenue. Definitely something I want to see integrated into the project manager when it gets going though.
As some other user said: unlimited clay when sculpting, new brushes and dynamic retopology. Also, I’d like to see the option to animate cloth pinning.
Auto Retopology like that in 3D Coat
Sorry should have added……… PLLLLLLLEEEEEEEAAAAASSSSSEEEEEE!!!!!! ;)
Multicore support for baking :)
Sometimes baking takes more time than rendering. I just have a 3 core CPU but I guess people with 4 or 8 cores makes it crazy to see their cpu at 20% but baking takes hours :D
I would imagine GPU support would be significantly faster than multicore support. :)
If I ever get that dual 12-core processor (24-core, 48 virtual cores if hyper-threaded) computer of my dreams someday, I’d hope to use CPU rather than GPU! But hey, why not both CPU and GPU? :D Nothing’s faster than hybrid-accelerated CPU+GPU support! Of course, for maximum support for all users, I’d hope it’d just be optional between either CPU, GPU, and CPU/GPU.
Actually, a GPU might still outpace a CPU even with 24 cores (my understanding of virtual cores is that they do not contribute to faster computations, they just divide the use of a core’s time between two processes more easily, correct me if I’m wrong as I tend to be). Current hybrid accelerations don’t work well due to overhead of communicating between the GPU and CPU, but hopefully that paradigm is changing.
But anyway, for an intuitive idea of why a GPU might be faster than a CPU, let me give a quick example (I know you might already understand this but some other people who read through here might not, so I hope to help them too). Take for instance, a general case where a CPU core is 4x faster than the GPU core (not accurate but this is for intuition again!). For your GPU, you have 128 cores, but again, each slower than the CPU.
Now, lets say you need to computer 1024 things, and we can do any combination of them at the same time. Computing 1 thing a second (intuition!) your GPU will do it in 8 seconds (1024 things/128 cores). By comparison, you CPUs will do it in over 10 seconds! (1024 things / ( 24 cores*4 times as fast)). If we start adding more things to compute, it will only take longer!
But the key is is that we can do each of those 1024 things at the same time, which is not always the case. if we can only do one of the 1024 things at a time, the CPU blazes past the GPU (still slower overall because only 1 core can be used). And granted, this is a gross, oversimplification, but I hope that it is apparent why a CPU might not be as fast as a GPU even when it seems like the CPU is huge. :)
I don’t thing a GPU could bake a simulation. GPUs can do some things faster than CPUs but they can’t do everythink. CPUs are more flexible
Ah, I assumed textures.
In all technicality, GPUs can do anything a CPU can do (http://en.wikipedia.org/wiki/Functional_completeness), the question is which is faster in the end. I’m not sure that the current physics setup in Blender would allow for efficient use of the GPU; but I know that GPGPUs (General Purpose GPUs) killed physics cards as a possibility for future “standard” hardware and thus should be good for certain simulation algorithms.
Thanks for the clarification. :)
Improved arch viz interface with the ability to work in feet and inches and much larger environment limits. Essentially, I would like to be able to import and build on life sized architectural models in real world imperial units inside of Blender. Right now, any model I import from Autocad has to be scaled way down to even fit within the limits of the Blender environment, and Imperial units are something to avoid. Of course, I’m fairly new to Blender so it’s entirely possible I’m just ignorant about how to make this work as fluidly and intuitively as the 3DS Max/Autocad relationship I was used to.
OK, OK – How in the world do you import from AutoCAD – The best I can do is open up v.2009 and use the 3DSOUT you can download from autodesk….. is there an easier way???
On a side note, I always have to scale my models down 1/12 so that my models come into Blender in decimal feet units. I also move the model close to 0,0,0 so that it is not off in no-mans land. Decimal feet units does suck, but I try to do all my modeling in AutoCAD first so I don’t really have to mess with the units.
I export to .stl and import into blender that way. Everything is unioned together, and the topology sucks, but you can work around that with a little bit of effort. Exporting separate .stl files for each component also makes it a little easier.
Blender 2.49 imports directly DXFs. That’s the one I use. Proper Imperial units, I would like that.
It would be easier if you would understand that 3d software only work in the units they have and are playing to have real world units in order for artists to have reference points. If for example 1 blender unit would be 1 meter doesn’t mean that 1 3ds Max generic unit equals 1 meter, in fact they’re not the same.
Working with OBJ’s between blender and max requires rescaling because of the way the two softwares handle the units. In blender a 1m^2 plane when exported as OBJ and imported to 3ds Max with metric units results in a 1cm^2 plane. As such you have to export the OBJ with an export scale setting set to 100.
There’s no such thing as absolute real world units in 3d software, they just make it easier for the user to measure things. The reason 3ds Max and Autocad work well together regarding units is just that they’re set up to interpret them the same way.
“The reason 3ds Max and Autocad work well together regarding units is just that they’re set up to interpret them the same way.”
I’d imagine the fact that they’re made by the same team and have a common ancestor probably also has something to do with it.
Autodesk, like adobe, has an appalling amount of incompatibility between most of their software packages.
Ability to use render output as a texture node.
Faster viewport
Here’s my top four choices:
1. a caustics map baking feature.
2. iOS support for the Blender Game Engine.
3. Musculature control (tension control) of weight-painted meshes to simulate muscles flexes easier.
4. A Particle Brush System: A particle painting tools for Texture Painting using Dynamic Paint and particle sculpting tools incorporated into Sculpt Mode (both instance are using particles as a brush).
If iOS, then android too ;)
YES, for this GOOGLE (and Android…) summer of code :
1__ It will be AWESOME if Blender can EXPORT the GAME made in BGE (Blender Game Engine) for Android
(at least Android !!!?), after may be iOS ;-))) ?? (IT WILL BE GOOD FOR the GOOGLE COMPAGNIE…)
2__ And add : NEWS LOGIC BRICKS = PYTHON CLASSES (to use too in Python scripts…!) : N – FINGERS for the
multi-touches on the game-screen, interactive home applications and interactive public applications, …
3__ First for BGE and after for Blender Render ;-))) :
3a__ A FULL IMPLEMENTATION (ALL inputs and outputs) for microsoft – KINECT
Like : NI MATE ( http://www.ni-mate.com/case/case-study-1/ )
3b__ AN INTUITIVE AND AUTOMATISED-MODULE or an SCAR_Logic_Brick :
SCAR = SIMPLY-CARACTER-and-AUTO-RIGGING in the game
Like : MIXAMO ( http://www.mixamo.com/c/auto-rigger )
After, i think, it’s easy to use this “module” into Blender Render for REAL-TIME CAPTURES… !!!? ;-)
All, have a GOOD DAY, and BGE Imagination !!! ;-)
Sévérino,
Sculpting!! Unlimited Clay, Better Sculpt performance with High density meshes, and Polypainting in the MultiRes modifier.
Linux needs a good Sculpting tool, and 3D Coat is really Artist Un-friendly (unfortunately).
3D coat is artist unfriendly? I find it quite artist friendly and easy to use with a low learning curve.
The real problem 3D coat has are the fanatic christian developers with their “please-don’t-create-lightly-clad-girls-with-our-software-or-you-will-go-to-hell-TOS” unfortunately.
Whoa, I had no idea. Just read the “Our Voice” page and.. whoa…
Yeah I know.
But that’s nothing to past activities they did. They refused to sell this software to a costumer who had a sexy girl in lingerie in his portfolio and sent him a mail to proselytize him.
I knew they where Christians, but didn’t know they where crazy, lol.
Anyways, my attempts at 3D Coat (and there have been many) have always ended in a bit of frustration. Crappy UI aside, 3D Coat simply doesn’t have some sculpting brushes I need (in either Voxels or Surface), the top being Crease and Clay. Yes, they have a Clay brush, but it doesn’t work like equivalents in Zbrush/Sculptris/Mudbox. Crease brush (my most used brush) is no where to be found, and I hate having to alternate between Carve and Pinch.
That being said, I commend the technology work they’ve done. PTex texturing, Multi-Res Voxels, and Auto Retopo are all excellent features. Autoretopo is worth the price tag alone. Still, I would love to see open-source Sculptris tech in Blender.
I agree with you about the Sculpting Tools and their UI. That’s not really convincing and not nearly advanced as Zbrush.
What I really appreciate is 3Dcoat’s texture painting. That is better than Mudbox’s texture painting IMO.
And Zbrush does only polypainting which means you need high density meshes.
Yes and the retopo tools are nice as well. No doubt.
But I still don’t like that these guys bother me with their christian beliefs. I respect the religious beliefs of people. But I hate it when someone tries to tell me what I should do or not.
The thing I like about Polypainting is that you don’t have to worry about UV layout before coloring, and you never have to deal with Texture Seam artifacts. That’s not to say Texture painting isn’t needed for final texture touch-ups or when dealing with a finished, low-polygon Mesh.
Polypainting and Sculptris tech would actually go very well with one another. I’ve discussed it in the ZBC forums here: http://www.zbrushcentral.com/showthread.php?93393-Dear-Dr-Petter-and-Pixologic
—
Ya I’m defiantly no fan of people pushing their religious faiths upon others either. You can’t blame them for being outspoken however. Their “Our Voice” page states at the end that they can’t and won’t try and control you, but they request you think about the impact your art has on the community (lol). Though your story about them not selling to someone based on his portfolio is contrary to their statements there.
I’d like to see the preferences screen pop open in the middle of the screen.
I’m i being under ambitious, given the latest releases? :)
Actually, it opens up relative to where the Info Editor is located. For instance, in Default View, if you change the Properties Editor to an Info Editor, it will open in the bottom right instead. This is to minimize travel time to get to the Preferences window after you select it I suspect, and it’s actually well handled. Taking this into consideration, I would think it is pretty well thought out for a simple window. :)
Since developing new things can be hard, I’ll suggest a couple simple things (a few of which were in 2.49 and earlier)
I want:
* The “Link Scale” button that was in 2.49. Manually doing the math in your head when using the S key sucks. and doing the math in the 3 different edit boxes 3 different times is annoying.
* The counter that showed how many loop cuts (and other things) that you are going to make. Manually counting to 20+ sucks. the side panel options don’t always work.
* A built-in material database, which would have similar thumbnail previews like the Sculpt tools have.
* easy access button to clear all unused materials/textures/mesh data.
* Re-position groups of stuff. not just collapse.
* I also miss the horizontal layout! Looked less cluttered, IMO.
* I’ll stop here :)
I like the vertical, it seems less cluttered. I think it might be good to have a preset to make it simpler to change it to how 2.49 was, but I will stick with the vertical.
You can manually change the layout back, I believe it is a right click option. I am sure someone has made a layout just like the old one that might be importable. :)
yes, but it doesn’t work well, at all. try it. Everything is weird, things are not moveable
So better handling and undockable menu items is what you would like, am I correct?
Btw, if I am understanding you correctly, things are movable, the dashmarks the top right of each drop down menu can be used to drag and drop them.
I got bad eyes and can’t even see the dash marks, until you mentioned it :P had to get like.. 1 inch away from the screen x__x. they need to be darker (or whiter)! clicking anywhere else just folds it up, but now I know you can move stuff ! thanks, 1 thing down.
By the way, here’s a writeup on the rational for going to vertical menu items, it’s a good read in my opinion.
http://download.blender.org/documentation/bc2008/evolution_of_blenders_ui.pdf
Network rendering for Cycles. I know they are already working on that. Thank you.
I’d say dynamic tesselation first and foremost. Also, anything that can use nodes SHOULD use nodes. :D
The number one item has got to be better/faster vector blur. It is the one area blender could make massive improvements in render speed and quality with the native renderer and cycles into the future.
I say you should focus on the compositor and the video sequence editor. I would really like a project window for the VSE.
I know that there will be one in a year or two (when I become competent at programming), but I would like it soon.
P.S. Do any of you programmers know how hard this would be? Are there any problems that might happen because of the way Blender is written?
Also, I would like to know if/when cycles will be supported for network rendering/ renderfarm.fi.
FBX-import and export
Better modelling tools inspired in modo or softimage :] , not only the possibility to work with n-gons, we need more efficient modelling tools :]
the ability to create usable quality meshes/textures from a few pictures, with the necessary tools to tweak/stitch them
Freestyle in trunk – or is that coming sooner?
TRANSFORMERS SCRIPT OR ADD ON…. please create a script or add on to easily create transformers like animations !!! or at list make some video tutorials on how to model and animate a transformer like models and animations.
please post any links to learn how to do something like that ..thank you!!
You’re kidding right?! Please tell me that is just a troll post and not meant to be serious.
BRAIN SCRIPT OR ADD ON for those who wants everything by “click a button “make cool!” :)))
Yeah, it’s really trollish by me :)
merge all the AMAZING features of past GSoC… 8)
I don’t know if this should really be a GSoC project, as it is fairly small, but I would really really love to see the OOPS schematic make a come back in 2.6x. I used that thing all the time in big and multi scene files in the 2.4x days.
+1 for this.
I’d like to see the VSE get some attention. Easier and more intuitive way to load on effect strips (and maybe, if it wouldn’t be too far out of scope, a reworking of the way effects are applied); along with more and better effects. A non-hacky way to do transitions, picture-in-picture and so on would be nice. Better clip/resource management (maybe even something as simple as monitoring a directory and displaying the clips in Blender so that I can drag and drop them onto a track), better title management and text overlay and so on.
Being able to integrate the positional audio object in a scene as track would be awesome.
In a lot of ways, what I’m asking for is already present to some degree, so I guess what I’m really asking for is a UI refresh for the VSE!
I think it would be a good idea to implement Preselection-Hightlighting into the Blender Modelling Tools:
1) Preselection-Highlighting: Preselection-Highlighting is the ability to hover the mouse over a Mesh in Edit-Mode and continuously have the Closest Edge, Face or Vertex temporarily Highlight. The user can then click to Select Geometry with the knowledge of exactly what is being Selected.
2) The Ability to have Mesh-Select-Mode set to the type of the first piece of geometry selected after a Deselect-All: After Deselecting-All in Edit-Mode, the user can Select a Vertex, Edge or Face. The first piece of geometry that the user Selects becomes the Mesh-Select-Mode. This option is especially useful when combined with Preselection-Highlighting.
3) The Ability to Use Preselection-Highlighting as a Temporary-Selection for Mesh-Operations: After Deselecting-All with Preselection-Highlighting enabled, the Closest Edge, Face or Vertx is temporarily Highlighted. If the user clicks a Vertex, Edge or Face to select then that is the Selection but if the user instead presses a keyboard-shortcut for a Mesh-Operation (such as Extrude, Translate, Rotate, Scale, etc) then the Geometry that is Highlighted is taken as the selection for the Mesh-Operation. The user clicks to finish the Mesh-Operation and the Temporary-Selection is cleared resulting in the user being back in Preselection-Highlighting Mode.
4) Tweak-Mode: A mode where Preselection-Highlighting is enabled with the default mesh-operation being Translate. The Ability to Use Preselection-Highlighting as a Temporary-Selection is also enabled. Instead of using a keyboard-shortcut to trigger the Translate, the user simply click and drags to move the Preselection-Highlighted Vertex, Edge or Face and releases the mouse to Confirm. Tweak-Mode enables quick modifications to a mesh by manipulating one piece of geometry at a time. The user Right-Clicks to exit out of Tweak-Mode.
I feel that adding these features as options to Blender that the user could enable would make Modelling faster and more intuitive. In certain instances when manipulating a single Vertex, Edge or Face the user could use the Preselection-Highlighted Geometry as the Selection and save a mouse-click. Since the user is most likely to first make a Selection after entering into one of the Mesh-Select-Modes then time can be saved by instead using the Selection of either a Vertex, Edge or Face to enter into the corresponding Mesh-Select-Mode after Deselecting-All.
I am not sure how hard it would be to implement. Most of them are inspired by the open source program called Wings3D that has a Winged-Edge data structure that reminds me of BMehs. If these options already exist and I am just not aware of them, please let me know.
Hold Right Click and Drag to quick translate on the closest object, vertex, etc. :)
I really like some of these ideas.
Import/ Export:
The ability to import .step files would be huge. I get a lot of models to
work with form solid works. The best export is the .step. I normally import them
into Form-Z. But now that there is a Maxwell Render add-on for Blender. Also,
because blender handles large amounts of geometry better than Form-Z too.. Also import for .skp. So we can bring in objects from the google warehouse.
.step and .skp import please.
Aaron
Dynamic Tessellation, Adaptive Topology… what ever ya call it xD..
path-drawing directly in the compositor – huge possibilities there
blender for iPad 3 – rumours have it that quad-core and retina screen are possibilities – intriguing thought
+1 for material presets/online repository link, this could save so much time.
A better way to do quick masking like in After Effects which could be done directly in the compositor, which brings up the odd way that Blender draws curves and handles. It should really behave like every other vector drawing or composite masking tool.
And YES please give the VSE some love, I’m a video producer and I couldn’t even figure out how to do a simple split screen.
please some way to port games done with the the BGE into ios,android etc
mesh modelling – parametric modelling (node based probably), like Houdini but in Blender’s way ;) (faster and smarter)
physics – node based particles
physics – complete Hair (collisions at least)
physics – Phymec tools (I saw this wish already here)
physics – FluidSim and SmokeSim GPU boosted
physics – node based mesh destruction tools (recursion destruction physics and so on). Here I’ll explain what I mean:
it’s when you can set up a physic scene and point to all possible interactions (when you sure about it). And you will make decisions about what type of destruction you’ll get from concrete reactions (so it can be done without entering all proper phys. properties for objects but just choosing desirable effects – “sphere crash or directed cone crash” for example). Link this with node based particles and it will be the ideal CG FX machine (as I see it).
modifiers – node-able modifiers’ stack :)
unlimmited clay
and
dx11 tesselation
and particle meshing for sph
Unlimited clay and Nick Bishop’s SKIN MODIFIER!!!! I want BSpheres and I know I can’t be the only one!!!
Auto-retopo like in 3d coat or topogun is really needed!!!