2020 – Blender Big Projects

In 2019 2.80 was released world-wide, and well received by the community and the industry alike. But work doesn’t stop there. There are still big topics, old and new, waiting to be tackled.

To help guiding the Blender development for this year, here is a proposal for 10 big projects for 2020.

Library Overrides

“Replace animation proxy system, and support local changes from linked data.”

The initial support for library overrides was introduced in Blender 2.81. However the plan is to replace the old proxy system, and support:

  • Animators using overrides for character posing.
  • Multiple instances of the same character with different animations and variations (e.g., material tweaks, …).
  • Set dressing with local variations of instanced assets.
  • Linked in character with cached animation, using overrides.


“Support non-destructive sculpting in multiple levels of detail for skinned characters, preserving a base mesh for baking and animation.”

As part of polishing the sculpting pipeline, the multires modifier will be ironed out, and improved beyond its 2.7x iteration:

  • Propagation between subdivision levels will be tackled with support for neighbor vertices subdivision.
  • Sculptor working in multiple resolution levels.
  • Animation and playback of base meshes, in fully sculpted characters.

Particle Nodes

“Implement a stable, reliable and flexible node based particle system.”

Artwork by Roland Tiss

Inspired by the Animation Nodes project, the particle system will be entirely refactored in a node-based system to support:

  • Motion-graphics work that relies on time-based animations.
  • Set dressing of environment elements.
  • Generative weathering of surfaces.
  • Fireworks (particles that spawn new particles).
  • Dew drops on a leaf.

Volume Object Type

Volumes need their own native datablock type, so Blender can support other use cases like rendering OpenVDB files or procedurally generated volumes.

Artwork from Cosmos Laundromat – first Cycle

Supporting OpenVDB is no small task, but it will help integrating volumetric simulations in Blender with the upcoming nodes-based system:

  • Existing use cases (smoke modifier on mesh).
  • OpenVDB file rendering (import of files from other applications).
  • Procedurally generating volumes (part of everything nodes).

Hair Object Type

A node and modifier based hair object to replace the existing system. No new functionalities at first.

Victor hair simulation – artwork by the Cosmos Laundromat first cycle

Year after year artists produce great hair simulations with Blender. However this area has always been a big source of frustration due to it instability and convoluted code-base.

A first step towards a node-based hair system will be to port the system to its own object type. However no new features are planned for now.

Faster Animation Playback

“Animators should be able to work in an interactive and real-time environment.”

Animation scene from Daily Dweebs

The bread and butter of the animators is the viewport work and playback. And it should be as fast as the target frame rate of the animations:

  • Animation cache support.
  • Animation playback/render performance.
  • Motion curves should be update-able in realtime.
  • Playblast shouldn’t take much more than regular viewport playback.

Scene Editing in Object Mode

In heavy scenes, new features in Blender 2.8x make most operations over data-blocks more problematic than in the 2.7x era. The goal is to bring those operations back to a constant or logarithmic complexity.

Artwork by the Spring movie team.

One of the most noticeable drawbacks in 2.80 was undo in heavy scenes. This project starts by implementing optimized per-datablock global undo, followed by a more generic data-block management with many objects:

  • Undo changes in object mode in a heavy scene.
  • Undo changes in pose mode in a heavy scene.
  • Duplicating objects in a heavy scene.

Fast Highpoly Mesh Editing

Get back 2.7x performance level for mesh editing.

Artwork by Daniel Bystedt

A good performance for mesh editing is key. Although the deliverable for this project is still under scrutiny, in general lines it is:

  • Editing high poly objects without modifiers should be possible with good performance.

Alembic / USD

“More feature parity with other software.”

Spring asset visualized in the USD viewer.

While the industry moves towards open formats and standards, Blender plays it part by expanding its compatibility with these two very important file formats:

  • Alembic to export custom properties.
  • Alembic to import arbitrary custom data.
  • Wrapping up the Alembic importer with support for different kinds of data-blocks and render integration.
  • USD exporter with possible skeletal animation.
  • USD importer.

Asset Manager “Basics”

“An asset in Blender can be models/meshes, armatures, materials, textures, rigs, scripts, node groups, Collections and so on.”

The Asset Manager is about adding the ability to manage and browsing data, and a platform for anyone to extend and write their own managers:

  • A way to store, browse and load user-created content (such as pose libraries, primitives, materials, …) for use as a starting point for new projects and scenes.
  • A way to manage inter-linked projects, such as movies with linked characters, sets and props.

There are other projects of course, such as:

  • Vulkan
  • AR/VR
  • Greasepencil
  • Custom keymaps
  • Retopology
  • Compositor
  • Sculpting
  • Texture painting
  • Cycles
  • EEVEE improvements

These will happen in parallel, as well as module maintenance (i.e., bug fixes and patch reviews).

The projects listed here have been defined to be core development, for which the whole team takes responsibility together.

Dalai Felinto
Blender development coordinator

101 comments 52,436 Views
  1. Sounds like a good plan. I like the fact that near 30% of the list is about performance. Keep up the good work. I can’t wait to see the new Particles/Hair Nodes.

  2. Whats the status of Animation system revamp? Is that not expected for a few more years?

  3. This is already a great & very important list and would make blender even better than what it’s right now, but i do hope that you guys don’t forget the little details and general workflows they’re as important as the big features as the saying says “the devil is in the details”.

  4. I believe implementing a feature-complete USD pipeline would solve with it some of the projects above incl. animation playback, volumes and overrides. But it would need deep impl. of USD into Blender and perhaps assigning more people to it.

    A great reference would be the way Maya and Houdini are integrating USD; it would require a good amount of work, but it’s the industry standard, open source and the future.

    Keep up the great work!

    • I believe the devs have stated that they will definitely not replace the existing deps graph with USD. AFAIK, the actual structure of the program is way too different for that to be feasible, and USD in itself is not a silver bullet to solve all of blenders problems.

      • Yep and also Blender can’t integrate hydra delegate since it’ requires opengl 4.x, maybe this could change way in the future.

  5. Well with so many important projects I guess this makes as much as some other combinations would. I do wish asset manager didn’t seem like such minor priority..

  6. I am missing the video sequencer in this list. Any plans for this?

    With Eevee being a real-time engine, it might be possible to integrate videos directly to the 3D scene, perhaps by adding a special video object type that maps directly to the camera. Regular compositing could be used for videos and a sequencer could also be useful for other things like animations. This way the sequencer could be integrated better to blender and other areas of blender might also benefit from optimizations done for video editing – Blender could even become a powerful special effects and 3D video compositing tool. For example think about something like virtual film production in Blender, similar to what studios currently do for movies with LED screens and real actors – just in 3D in Blender.

    I also wrote a few words about this idea yesterday: https://blender.community/c/rightclickselect/kPdbbc/#5e2dfb259c122124d957d452

    • While the VSE isn’t mentioned in the post, it does have some things people are working on: https://developer.blender.org/T64682

    • They recently hired Richard Antalík for developing the VSE further. I heard they have plans of making it great for storyboarding. But for now he is working exclusively in the bug tracker for triaging.

  7. I really wish to see some level of support for AMD graphics cards on Mac OS. Mac OS users can’t use GPU in blender at all.

    • Well it’s your fault for buying a Mac.
      Macs aren’t for 3d or any serious work.

      • Well, this is a pretty lame answer. All Blender versions should be equivalent. Vulkan could be brought over Mac using MoltenVK: https://moltengl.com/moltenvk/.

        Come on Blender Team, you can do it! 🙂

        • …or maybe OpenGL support through Metal with MoltenGL: https://moltengl.com/moltengl/

        • Do you have any idea how huge of a task would be to switch from OpenGL to Vulkan? I bet a few years of full time work to bring it into an usable shape. Also MoltenVK is another layer on top of another layer which means additional issues.

          Just a as the guy said: Apple if for showing off not for serious work :))))))

          …OK. I am joking. It is good for some stuff (including showing off) but obviously not for some other stuff. Pick the correct tools for the required job. Don’t use a hammer to screw screws.

    • It’s not blender’s fault that apple doesn’t want to support openGL anymore.

    • sell your mac and get an actual PC lol

      apple is not gona help you much

    • Use Radeon ProRender! It might even be faster than Cycles because ProRender uses Metal 🙂


  8. Its a great proposal.

    There is something about Multires I dont understand. In 2.7x, a sculptor keeps in on resolution level. Unless he decides to switch to a lower or higher one.

    > Propagation between subdivision levels will be tackled with support for neighbor vertices subdivision.
    > Sculptor working in multiple resolution levels.

    If a Sculptor works in multiple resolution levels, does it mean that switching to higher or lower ones is enabled? Or does it mean subdivision resolutions adapt to sculpt detail density?

    • Sculpt subdivision levels will allow the artist to work on a low sub level to work on big shapes without seriously altering the little shapes stored in a lower sub level. This is the most common sculpt workflow in other softwares.

      • Thank you for your answer.
        I misunderstood how words are meant.

  9. Really excited to read this! Thanks for the article Dalai! If Blender is alredy GOOD I can’t wait to see what will happend in the next years! And also, thanks for going to the Blender Hoy & Blender Today livestreams! I like it more now!

  10. Blender used to be the best Modeling,Uving tool… but not anymore ,the focus has shifted to please the big donors.

    • It’s terrible for unwrapping. That crazy disconnect between what’s selected in the 3d view and uv view is a nightmare. I would never take the unwrapping tools seriously while this is the workflow. Also, the basic tools without addons are FAR behind the basic unwrap tools of Max/Maya.

  11. Great ! But why still forget the NURBS modeling, which is so much more efficient for everything that is not organic?

  12. Microdisplacement should be out of experimental preview.

  13. 1) In EDIT mode, right click should add selection tools like “Select Loops”.

    2) Could you add erosion textures for landscape add-on?

    3) Could you implement something like Octane Vectron ?

    /”Octane Vectron (Vector-Polygon) provides infinite procedurally generated scenes, volumes, and geometry which bypass meshes and volumes. With Vectron you can create procedurally generated scenes entirely on the GPU without using GPU VRAM and with zero memory footprint. You can now render triangle-free geometry using all of Octane’s built in OSL and texture shaders and more.”/

    • 1&2 are questions for Python developers. I’m pretty sure the answer to #3 is “No.” but you might try putting a feature request on RightClickSelect.

    • Why clutter the right click menu when you can just Alt+Click to select loops?

      • Some people absolutely refuse to use keyboard shortcuts. I know some older graphic designers who use several Adobe products 12 hours a day 7 days a week and they don’t know a single keyboard shortcut, not even ctrl+s.

  14. Very nice!

    Please don’t forget to fix the Sculpt Mode undo / redo.


    • It’s not just Sculpt mode, it’s other part of Blender too and aslo undo shouldn’t affect your tool settings, so annoying.

  15. The list looks well thought out, the year looks interesting! With this many tasks though, would it be correct to assume the push for revamping the animation tools/workflow in Blender (aka “Animation 2020”) will not be a thing?

  16. Vulkan is the most obvious to get optimized and installed to replace OpenGL. From there a Metal 3.0 API for OS X will be far less daunting.

  17. u guys are awesome…i am needing z brush nano mesh effects.

  18. My top Blender wishes, by priority:
    1 ) USD, ABC, and VDB I/O for wider industry acceptance.
    2 ) Library Overrides and Asset Manager.
    3 ) Undo performance
    4 ) Viewport performance (and performance profiling)
    5 ) Particle/hair system overhaul
    6 ) Viewport Compositing/compositor performance
    7 ) Render Improvements
    8 ) Retopology improvements
    9) Texture Paint performance
    10) UV workflow improvements

    You’ll notice that VSE and Physics are missing — this is because there is a FLOSS alternative to the VSE in KdenLive already. Physics have been improved quite a bit recently, and stand to gain from node based particles. But in the industry/studio world, better I/O will often mean importing the physics and simulations from Houdini, anyways.

    1) is essential for Blender’s future in the business world. Hopefully, studios that use Blender will be able to pitch in and help with this major feature in master, instead of building their own production tools because it isn’t being done!
    2) needs to be done for managing large projects efficiently. It should also help keep undo/redo performance reasonable.
    3) Undo performance is a bit tricky. Blender is slow, but reliable. I use other softwares that are faster, but sometimes have steps that can’t be undone, or worse, undo/redo cause bugs and mess up my objects. I then lose work because I couldn’t trust the software. So slow is better than unreliable.
    This other software, like I said, is fast. And I think the way it works is, the undo actually plays commands backward, like rewinding a tape. This is good for simple commands, since it has a fast internal API. But for complex commands and scripts, it can be just as slow as Blender. I think there may be a way of doing one or the other depending on the situation and improving speed while maintaining reliability.
    4) This is essential for animation, of course, but it will also improve the EEVEE workflow. I don’t think it’s necessary to aim for 60fps in huge scenes right away. I think if those kinds of improvements take some time to develop, it might be easier to offer better tools for optimizing scenes in the meantime. For example, LODs, off-screen culling (especially for particles), and proxy models — this could be handled in the Scene’s simplify settings and/or use a modifier. Tools for profiling performance would be great for tracking down the source of a problem– a modifier, constraint, polycount, etc. and the ability to enable and disable evaluation of certain objects would be awesome- freezing and caching etc.
    5) In its current state, Blender is far behind the industry in VFX and grooming tools. While better I/O will allow studios to use Blender for character animation and import effects, creation of effects within Blender would be better still– and will integrate with existing rigs and modelling workflow. Particles are used often enough for environment art that it really needs an overhaul soon.
    6) This is the one thing on my list that’s only there because I want it. Viewport compositing means better NPR shaders for us all!
    7) It seems one of the major sticking points for Architectural Visualization artists currently using Max is the lack of support for Corona render in particular, and speed in general in Blender. There’s a bigger market for this than for animation, probably, and making Blender the most attractive choice for such a big market will help a lot of people. Cycles is excellent, but since it is optimized for animation, it’s not particularly easy to optimize it for speed in still images. I wonder if the Blender Foundation could work with LuxCore to help serve the architectural visualization artists that are considering making the switch. At any rate, Cycles improvements are welcome!
    8), 9), and 10) are just nice-to-haves.

    My list lines up pretty well with the above proposal. I’m excited about 2020! last year, we shattered everyone’s expectations of what a “free” software could do! This year, let’s just try to shatter people’s expectations about what *software* can do.

  19. There is a mention of Particle Nodes but not everything nodes, is there a reason for that?

    I like that Multires, Scene editing & Fast Playback are on the list and i hope we can have a better and powerful Rigging/Animation systems that enable us to create blockbuster VFX Character.

    That’s the area Blender is missing right now, of course among other things but imagine if one of the big VFX houses like Framestore adopte Blender and succeed in using it on their big projects.
    This will completely change the Industry and will give other Studios the confidence to do the same , hopefully Ubisoft will try to achieve that for the gaming Industry then Blender becomes unstoppable force.

    • Because Particle nodes is part of everything nodes, but not everything can be made nodes immediately. Particles is just the first implementation, and is the primary focus for 2020.

  20. 100% agreement on the list.

    Every item on the list is definitely a ‘must have’ in terms of priority, including the current Tracker Curfew project, it’s only a question of getting them done, in what order, and how fast.

    In my opinion, stability, UX and performance are always more important than ‘features’. There’s no point being able to ‘do a lot of stuff’ if you do it in way that’s confusing, slow and buggy, so the UX, stability and performance projects should absolutely be the top priority. Being even more specific, I’d say the order of priority should be:

    1. Stability
    2. Performance
    3. UX
    4. Feature

    I define ‘UX’ as addressing issues with existing functionality, not new functionality, making existing functionality less confusing to use.

    So looking at the list of projects, this is how I see it:

    Tracker Curfew

    Faster Animation Playback
    Scene Editing in Object Mode
    Fast Highpoly Mesh Editing

    Library Overrides
    Hair Object Type
    Volume Object Type

    Feature/UX (a bit of both):

    Asset Manager “Basics”
    Particle Nodes
    Alembic / USD

  21. I’m surprised to see that there are no big projects for the Video Sequence Editor thought into 2020, since the VSE should become the hub for realtime story development and virtual storytelling inside Blender.

    If the design targets for the VSE are undecided, I hope it is okay I share, what I see as the top 3 of challenges for the VSE:

    The biggest all time challenge is INTEGRATION, simply because without it, the VSE go no raison d’etre. Or in other words if the VSE is directly integrated in the 3D View(In and Out), meaning you could use the 3D View for previewing the video VSE contents and control the 3D contents from the VSE(ex. camera and scene switching etc.) with the lightning preview speeds of eevee, no one would argue that during the storytelling process, you should render everything and edit it in an external NLE, at least not until you would have to assemble high quality renders, which could be argued, could be better done in ex. Davinci Resolve. With better VSE integration, comes also round-tripping of strips in the Compositor, but basically make sequences and strips available on a frame change basis from/to the other Blender Editors. In order to improve the integration a necessary step is to move sequences out of scenes, so changing the scene doesn’t change the contents of the sequencer.

    The second biggest challenge is PERFORMANCE which concerns speed of playback(in 3D View, preview of video, scene strips(playbasts), multithreaded, cache optimized, GPU usage, fast image scaling, fast and invisible proxy handling/cache to disk) and quality(unbroken color-manegement, 10+ bit color, high quality image scaling etc.). This being second, since the final export quality could be handled externally(but would need a proper export option like OTIO), but it is very close to the number one, simply because the slow playback of ex. video on a first run will scare off many first time users.

    The third biggest challenge is the TOOL SET & THE UI that comes with it. With the toolbar finally integrated it is now possible for proper icon based tools, but it needs serious planning to make it well designed, user friendly, UI integrated(on strip drawing/interaction) and (splitscreen)preview of ex. trim and slip/slide operations etc.

  22. Please more innovations for animators. For example, new tools for working with animation curves in the Graph Editor. Like animBot for Maya. More flexible motion path tracking. Motion path relative to the view from the camera. Ghosts for animated mesh. etc.

  23. Please add “extrude along curve” addon which was in 2,79.

  24. Architect here. I applaud the focus on performance, but was hoping for a bit more focus on precision modeling tools (https://developer.blender.org/T66337).

    We’re not yet contributing to the dev fund as most funds seem to go to vfx/animation/movie related features. I wish there was a fund that focused on adding features for the AEC field (architecture engineering construction).

  25. Two Question due to my lack of knowledge in the subject.
    So does changing to Vulkan mean we don’t need Open GL 4.0?
    Also Does the particle nodes mean a new physics system or just the old system with some new ways of interacting with it? for example does the performance and number of particles increase? Also can we have particles that hit each other instead of going through each other? (and what happens to boids?)

    • OpenGL will stay, although brings his own advantages, the ecosystem is still maturing.
      And more importantly, the support is very limited, only recent GPU’s are supported(you can see full list here).
      Since many users still have older hardware, Blender will have to keep supporting them with OpenGL, at least until enough time will past, and the Vulkan drivers will mature enough.

        • It’s nice to see Intel’s Ivy Bridge(3000 series) graphics getting a little Open Source MESA Vulkan love under Linux according to that Wikipedia link and all of AMD’s Terascale GPU Micro-architectures are getting totally ignored by every OS/Graphics API maker/maintainer.

          The bad thing about AMD’s last Tarescale(Terascale 3) micro-architecture is that even as late as 2013/2014 there where plenty of laptops produced/available sporting Radeon 7000M series discrete mobile GPUs that where not GCN 1/1.0 Micro-architecture based but where in fact just 7000 series rebrands of the Terascale 3 GPU micro-architecture found in AMD’s 6000 series GPU offerings.

          But I’m Fine with maybe waiting for System76 to start using AMD’s 4000 series APUs in their laptops offerings at some point before 2020-2021 ends and Linux Mint it will be for any of My older Intel Sandy Bridge and earlier Intel core i series based laptops along with Blender 2.79b.

          I’m having limited success under Windows 7 with using Blender 2.8(Edit Mode) on my Ivy Bridge core i7 mobile CPU based laptop except for one of Blender’s default editor workspaces. So I’m reading that under Linux there is better OpenGL support whereas Microsoft’s not been doing any OS/OpenGL maintaining for Windows 7/8.1 for more than 3 years now. So I’m becoming more interested in moving towards dual booting my windows 7/8.1 licensed laptops with Linux Mint 19.3 and maybe the Ivy Bridge core i7’s integrated 3000 series Intel Integrated graphics will benefit in the future from that MESA Vulkan support as well as better Linux OS OpenGL support for older CPU/Integrated Graphics based hardware,

          I’m more interested in any Laptop’s Blender 2.7/2.8 and above Blender edit mode performance as rendering can be done in the cloud if needed or on some Desktop/Workstation. But Blender 3D’s editor is where most of the actual 3D Model/Model asset creation is done and Laptops are powerful enough, for the most part, for that usage.

          I’m also looking at some Windows 10 based laptops at Microcenter with AMD’s Ryzen 3000H series APUs and Nvidia’s GTX 1660Ti based discrete mobile GPUs. But I’d rather not be be on windows 10 and would rather install Linux Mint 19.3. But I’m excited about AMD’s Zen-2 based 4000 series APUs and really I do not see any reason for System76(Pop OS) to not be able to begin offering some AMD APU based laptops as system76 is going to be designing their own in-house laptops instead of rebranding Clevo reference designs.

      • Vulkan isn’t new already. It was released in 2016. Vulkan also means rethinking and changing how Blender work internally. Blender was designed during a time when multiple cores and programmable GPUs didn’t even exist in the imagination of most developers. So I guess they have a lot of work which will span for many years. By that time drivers might start delivering OpenGL support implemented over Vulkan and most of the older hardware will be dead anyway. My laptops worked on average of about 5 years before usually their AMD gpu died.

  26. Please make UNDO work faster. Make it HIGH priority! Now it is even slower than in Blender 2.79.

  27. I love seeing Blender grow, now that Blender is being used in big animation projects, I like to see a HumanIK tool for motion capture retargeting

  28. Finally the animators recive a well deserved attention this year,
    Faster Animation Playback features is a need in big productions. Especially where you have more characters, with more complex riggings (yes the rigging eat a lot of frame rates)
    …I still don’t understand was what happening with the Open Subdiv, is still in Blender or was too hard to maintain?

    • Open Subdiv has replaced the older Blender Catmull-Clark subdivision, which is why the subsurf modifier is slower — It’s not GPU computed anymore. The current plan is to make a subdivision property for the mesh object, so that that it is always at the end of the modifier stack and can thus be handled on the GPU. This wouldn’t replace the modifier, which will still be available. This is a better solution going forward — everything nodes will eventually make it so that a subdivision surface may occur in any number of different situations, so a certain, final subdivision surface is necessary for the sake of creating a predictable enough behaviour for the GPU to work.

  29. Hello

    I did a proof of concept (within a perfectly working python addon) showing how a proper proxy system within blender can drastically improve viewport performance, for both huge scenery and slow animation.

    I really hope that my message can be recieved, and maybe one day talked during a meeting, i use this addon and the result are really nice already.

    Best Regards

  30. Hi anyone can help me, new in blender and I judge it interesting. But I have tow problems :
    1 import file dxf. ( I get just one part of plan)
    2 python, i get error for my first script.
    Can anyone help me koursiou.oumi@gmail.com

  31. Can’t see any type of integration with After Effects (at least camera\objects export-import animation data). It is very important in terms of motion design usage of Blender. Now I using old AE import-export addon that works with warnings.

  32. It’s a pity that 2020 seems to be another year when surface modeling, primarily nurbs, will receive zero love from the Blender Team. Nurbs modeling techniques are essential for designers, who would jump on Blender should this technique be there at a level comparable to Rhino.

    Dear Blender, can you create a project for improving the surface modeling techniques and donating specifically for this purpose, so you could do it in-house with your developers?

    Here’s a draft proposal for priority issues concerning the fondamental tools of surface modeling:

    1. Enabling the possibilty to draw a polyline, nurbs curve, bezier curve by simple clicks (like in Rhino) instead of a tedious extrusion of a primitive’s vertex.

    2. Enabling the possibility of having the control points right on a nurbs curve, instead of somewhere above it.

    3. Enabling the classic commands for the nurbs modeling such as: sweep on two rails, loft through multiple curves, surface from boundary curves.

    If it’s possible to officially open this project, I’d be the first to donate, and I’m sure that a lot of designers would quickly do so. That would grow your base for sure. Trust us Blender, this would be the single most important thing you could possibly do for your designer users in 2020.

    • Hello,

      I totally agree with you about the necessity to improve the tools for surface modeling. I have got in touch with Blender several months ago. I am used to work in Cinema and Houdini, but I was really curious about some interesting Blender features and the community. I have started to modeling a ship. I have step by step bump into couple of basic operations such as sweeping, copying/instancing over a spline etc. which seems to be a nightmare in Blender. Somebody else has mentioned here that the devil is in the details and I totally agree. It is nice to have a greasepencil, lot of plugins, fluid sims etc. but it doesn’t make sense to fly if you can’t land. Blender could be with all the plugins and so on the number one 3D modeling app, but there are these fundamental functions missing or not working like most of the users would expect. My biggest wish would be to polish the basic functions first so it won’t disappoint the users anymore and they will discover the whole great potential of BLENDER.

      I thank all the Blender team for their work and wish them really good luck

      • I recently tried to do a solid in Blender Nurbs and had to give up when I realized it wasn’t even close to capable.
        But should this be compared to Rhino? Really different design paradigms. Even Solidworks only does surfaces rather painful.

        • Hey Matthew, I meant to use Rhino as an example of a good nurbs modeling tool because Blender does have nurbs and Rhino is an industry standard for that. Only that Blender’s nurbs are clearly unusable. So my point is, as a designer who constantly does surface modeling (which is done through the nurbs and not the mesh), I’d be gladly donating to Blender for precise purpose of developing their nurbs into something that is actually useful.

          I’d like to pass a message to Blender to open this specific project with its specific donations line. There’s a tremendous request for efficient nurbs modeling workflow in Blender. Me personally, I have a lot of ideas on how I’d like to model but I’m not a coder or a developer. I’m a designer. I’d like to have good instruments to do my work in Blender and I’m ready to donate money for that.

    • You could always contribute 250k a year to hire an developer or two to work on these parts specifically 🙂

  33. I’m glad he’s fixing the multires for the sculpt but disappointed not to have the baking of displace and vector displace which are basic functions not prioritized.

  34. “Get back 2.7x performance level for mesh editing.“

    This is disappointing, we should be aiming for performance of other 3d packages not 2.7x which performance in that area was never acceptable to begin with.

  35. EEVEE + VULKAN = DEVASTATOR of Vray/corona/maxwell/arnold/lumion/twinmotion – this is the most important improvements of blender. We all know that most donations waa just because of EEVEE and making EEVEE a most powerfull alternative to all other VIZ software will make Blender an industry standart. Why? because with EEVEE creative teams will be able to render FAST, and to work with deadlines without renderfarms, without overpays for rendertime. All stuff like sculpting, VSE, etc – could be done well in other soft – but combining all and make final rendering – if the main thing. If Blender will have EEVEE with possibilites of vray but in almost realtime – Blender will be a next industry standart!

  36. Hi guys,
    Thanks for making Blender! 🙂 I’m just wondering – 1) when will Interactive Mode happen? , and 2) Will there be a version for the PS5, when it releases? 🙂


  37. For Faster Animation Playback i want to mention one thing related, the Armature modifier seems extremely slow, not sure if it’s doing it on the GPU or CPU, i have a rig typcial game model that’s less than 50k polys and a very basic rig steup but runs slow, so u can imagine if it was feature or vfx rig with more high poly count and complex setup of multipe modifiers..etc

    There must be a better way to do fast skinning and maybe move it away from the stack or something so it doesn’t bugged down, of course cashing the geo also help but i think the underlying problem is with the Armature modifer.

  38. Your proposal looks promising. I am happy with each one.

    I like to give two proposals. I hope you like some ideas.

    (Proposal 1) improve user experience and functionality of the NLA.

    Any animation has to pass through NLA. Making it a very important part of Blender. Yet it gets very little attention in the Blender Community and frequently vanishes behind feature requests.
    Improving NLA user experience or functionality might improve whole animation workflow.

    Some ideas
    * a kind of handle directly in NLA strips controls blending instead of tweaking properties.
    * support for different kinds of blendings. Enabling smooth non linear blendings. (Linear blending starts and ends abruptly.)
    * NLA generating actions, which could be-used in another NLA action. Main action could be selected for animation rendering.
    * a different kind of transition, which does not use action strip’s end poses, but instead it blends two strips which overlap. Just as if they where blended by stacking tracks.

    (Proposal 2) Non-Bit based bone layers

  39. This is a good list for big projects and i know people will keep asking for more big features but i do hope you guys don’t forget the other important areas that made Blender a good 3D Software overall.

    My wish is to see more improvements to uving editing modeling and rigging tools many of them require tons of external small addons and some don’t work in the recent versions unfortunately, hopefully Campbell will find sometime to work on them one day.

    We need something new like Maya’s Shape key editor/mode that has folders /groups to manage them and possibility to combine two or more shapes , a brush in sculpt mode to smooth or earse the blending effect..
    A New Pose and Layer editor/mode for rigging and Animation ,Component editor ,Set driven keys and many more stuff like this.

    My requests are not to be done immediatly but maybe squeeze one or two each year, hopefully there will be more volunteer developers who can help you with this burden of tackling many of these topics in one year.

    Best of Luck and keep Rocking 🤘.

  40. Amazing, a lot to look out for! Thank you 🙂

    Viewport performance which seems to have been split in three goals – undo/animation/and mesh editing is of prime importance. A proxy system of sorts would be great to implement natively as that will also significantly speed up complex scenes, especially for architectural design.

    Then, i am quite looking forward to:
    – particle nodes, especially that there is thought going in the initial design process for scattering objects
    – asset manager and more complete library overrides.
    – texture painting. For the casual texture painter, adding another program to the workflow is sometimes unecessary complication, so have a more robust toolset within blender for texture painting would be amazing.

    Thank you all 😉

    • Blender seems to be targeting the professional users with these projects.
      For some professional uses and for the casual users I think better snapping and precision modeling aids/tools and some improvements to the texture painting will be most beneficial.

  41. Hi,

    What about a real light linking like in Maya, 3dsmax, C4D and all others render softwares like Katana and co (include / exclude fonction). It really missing and the collection workaround is not good, 3 renders and compo to do light linking is really bad. It’s the only thing I hate in Blender Cycles, the light linking that not exist.

    Tell me if I’m wrong and already possible without do compositing (lightcollection is not a light linker…) but not found…

    All the other features are good, really good 😉

  42. Awesome!! As always, a great plan!! I’ve worked with Cinema 4D in the past. But now I’ve being a year with Blender since 2.80 alpha and I plan to stay! Great job guys!!
    My latest work with blender! 🇧🇷👊🏼😁

  43. Blender Asset Manager – I think it should also maintain Add-ons. Add-ons are an asset, have versions, and currently get installed in system specific file locations, which makes it hard to use Add-ons with Blender versions that are not installed (betas, custom builds, etc..)

  44. Above all, I wish that the developers are not under pressure with all these requests.
    Blender is not a company with shareholders like any other.
    No overwork and burn out in this fabulous team!

  45. I am glad Library Overrides and Volume Object Type are mentioned high on the list. The one very important thing Blender is poor at (despite being great), is scene assembly in broad sense. So what would be ideal is:
    a) auto instancing until override is needed
    b) everything overrides
    c) input data streaming instead of injecting (injection can be an override if edit needed)
    d) multiple methods of instance placement. Currently we have simple transformations, instancing based on topology or particle system. We need proper scattering solutions and not some hacky ways to populate scenes.
    e) up-to-date IO support – VDBs, USDs, Alembics (maybe it’s finally time to support curve rendering, eh)?
    f) data optimizations done on the fly. LODs, culling, you name it. GameDev is you design harvest field.

    I have mentioned it many times – take example from Clarisse. This is how this is done. Translate OO approach to scene editing – works great there.

  46. EEVEE PLEASE, EEVEE appears only once throughout, and less in 2.81-2.82, It makes me not interested. for users, EEVEE is the focus of the 2.8x series. please continue to EEVEE. LOVE Blender, LOVE EEVEE.

  47. Please also improve nurbs modeling. Now it is real pain. I want to have such abilities like:
    1) Simple mirroring of the curve
    2) Connecting two vertices
    3) Merging vertices on distance and auto welding
    4) Basic boolean operations
    5) Operation to make rounded corners

  48. Wireframes for technical modeling are still bad.
    No matcaps per object or Collection.
    Real Grouping of Objects.
    Random Colors per Collection.

    All these were promised before but never were delivered, at least keep your promised Blender Devs.

  49. Awesome that performance is one of the next big topics, if Blender can reach Max edit mode level and Maya’s Animation playback & rigging tools then It would be the best choice, I don’t expect Blender to reach Zbrush level anytime soon but any improvements from Pablo and other Developers are always welcomed.

  50. The ‘Particle Nodes’ I think is the most important of the ‘2020 big projects’. In any case, I believe that it is necessary, easy solutions with a minimum number of nodes (e.g. 10 nodes), because this is suitable for animations. Thanks to the developers for all this.

  51. I look forward to seeing the vulcan project happen.
    I wish you all luck.

  52. I wish for two things now:
    – Better tools/workflow for high precision modeling and precision snapping (with aids like in AutoCAD).
    – Improved and streamlined support, including layers and brush size independent of zoom level in texture painting mode – to make it easier for less experienced users.

  53. Guys, for real make some solid UV editing improvements what we have now it`s bad really bad. I do understand that you are chasing big sculpting dreams, but solid UV editing is must have. And one more thing macros creation like in MODO. I know there is possible via python to make it but not everyone is solid tech stuff

  54. For anyone hoping for Vulkan, here is the reason why this is not going to happen any time soon:

    /* Draw an overlay that shows what effect the brush’s texture will
    611 * have on brush strength */
    612 static bool paint_draw_tex_overlay(UnifiedPaintSettings *ups,
    613 Brush *brush,
    614 ViewContext *vc,
    615 int x,
    616 int y,
    617 float zoom,
    618 bool col,
    619 bool primary)
    620 {
    621 rctf quad;
    622 /* check for overlay mode */
    624 MTex *mtex = (primary) ? &brush->mtex : &brush->mask_mtex;
    625 bool valid = ((primary) ? (brush->overlay_flags & BRUSH_OVERLAY_PRIMARY) != 0 :
    626 (brush->overlay_flags & BRUSH_OVERLAY_SECONDARY) != 0);
    627 int overlay_alpha = (primary) ? brush->texture_overlay_alpha : brush->mask_overlay_alpha;
    629 if (!(mtex->tex) ||
    630 !((mtex->brush_map_mode == MTEX_MAP_MODE_STENCIL) ||
    631 (valid && ELEM(mtex->brush_map_mode, MTEX_MAP_MODE_VIEW, MTEX_MAP_MODE_TILED)))) {
    632 return false;
    633 }
    635 if (load_tex(brush, vc, zoom, col, primary)) {
    636 GPU_blend(true);
    638 glColorMask(GL_TRUE, GL_TRUE, GL_TRUE, GL_TRUE);
    639 glDepthMask(GL_FALSE);
    640 glDepthFunc(GL_ALWAYS);

    OpenGL is threaded into the code. They’d have to first create an abstraction layer and rewrite the whole engine.

  55. Mesh editing with subdivision surface modifier enabled is much more slower than before (2.79), when editing something more complex than the default cube.

    brick back cached BHV in cycles, it was much faster to render turn-table and fly through animations of hi-poly models.

    And finally the most IMPORTANT thing !!!!
    Please bring back 2.79 GUI and keep the rest 🙁 we are lost in space of user friendliness of alien users and alienated workflows. Things used to be straight forward before, even for newbies.

  56. Optimize Cycles like e-cycles
    Integrate a new physical render engine like appleseed o luxcore

  57. It’s the year 2020 and the blender team is just now figuring out that you need library overrides to manage animated scenes.

  58. For particles, it would be nice to have some kind of Interactive Creative Environment (ICE) like the one Autodesk Softimage used to have to enable you to select specific emitted particle(s) and then edit the behavior through the use of PARTICLES ID NUMBERS.

  59. How come everytime when the donations reach 100k per month the target for 20 developers jumps to 115k and when it is under 100k the target for 20 developers is exactly 100k?

    What sorcery is this?

  60. if you guys think that 2.7x edit mode performance is good then that’s not promising, 100k is nothing in today’s standard. it should be minimum 500k or even up to 1 million.
    We don’t expect Blender to reach Max in high poly editing but close enough.

  61. There was so much talk about ‘winning’ and ‘maturing’ during BCON 19: corporate sponsorship, AMD, Epic mega grant, and so on. Now, AMD’s own Radeon ProRender supports Metal 2 on Mac: has been for a while. Blender’s Cycles still does not, leaving Mac users, essentially, CPU-only, or forcing them on Linux or Windows. Yes, I’ve read the reasons for this: they were good in the past; not anymore. If Radeon ProRender does it, and AMD and BF are now cooperating, there is no reason whatsoever not to supports Metal 2 in Cycles anymore: not even the ideological ones (like, ‘good’ free open source vs. ‘evil’ corporations: they’re the BF’s sponsors and as such, are on the ‘good’ side now).
    I have to say I’m very, very, very grumpy with BF for continuing to sideline the Mac users. Apple themselves are bad enough: failed to resolve their kindergarten spat with nVidia, and when *this* comes on top of *that*, well… I’m very grumpy.

  62. Blender needs to introduce deformers like in Maya, they’re intutive and easy to use, just trying to make an object Bend on itself in Blender is really a pain in the butt and you have to repeat the steps each time.

  63. I still long for a modifier like the fracture modifier but on 2.8x…breaking and cracking things is a pain in blender

  64. Here is a good performance comparison video:
    But that is the old Blender, the new one is much worse. If you, for example, test the files available under this video:
    …you will find out that the performance is still getting worse. In my case, the animation bench is about five frames slower in 2.82 than it was in 2.80.

  65. it seems they want to cater only to big donors and animation studios. Just Like Maya. Nothing seriously regarding modeling, UV, motion graphics etc.

  1. Leave a Reply

    Your email address will not be published. Required fields are marked *


share this story on