Nice, yeah I remember the demo for that. Its a great concept. No idea how that would fit into the pipeline on linked meshes. But it is just cool to see it.
Blender 3.0 is out now and there’s still no support for mesh edit library overrides, or as far as I can find any plans to add such a feature.
I absolutely agree with plyRo and Hadriscus. The need for animators to directly sculpt into character mesh they’re animating with in a production scene is a normal and important part of any professional workflow. So is linking/referencing those assets (not just for performance, but also for maintaing parity as assets often get modified mid-production). We need to have both, and maya proves that both is possible.
At the moment we have to settle for workarounds. my go-to hack is to append the character mesh, and then replace its object data with a local copy of the library overriden version of the same asset - this ensures that not only is the mesh editable, but also all the drivers are preserved with the correct targets, and I can add new shape keys. Since the linked version is still present in the scene any other linked assts inheriting deformation from the character mesh can continue to do so, and if I do have to propogate a change from the source, there are often ways to kludge it through with join as shapes or something like that. It’s an imperfect hack but it’s not too labour intensive and it does work 95% of the time.
These are all symptoms of a deeper problem with the shapekey system in blender. It’s a very old part of the program and as far as I know hasn’t seen any major change in a long time. IMO it needs a complete rework to make it more flexible and pipeline-friendly, especially considering how vital shapekeys are for production-level animation tasks. The current absence of shape keys from the 3.X roadmap troubles me.
1 Like
This is a complicated issue on the Blender side.
Blender allows you to use a layout/scene as the background scene, or just link the layout as a collection in the scene with the actual character asset and do the work that way.
Basically if you have such needs you need to rethink your animation pipeline, the other solutions/workarounds are not that bad. Also the final layout is not where the asset work should be done anyway, obviously this is my personal opinion and how I would do it myself.
1 Like
I think I can see what you’re saying, but there’s a good reason linking characters into production scenes is standard practice - if you’ve got five characters in 50+ shots each across 7-9 environments and then you have to update one or more of them, what are you going to do? If the character assets are linked into each scene you only have to update the master file and all the linked copies inherit the change. Same goes for layout as, you say.
standard workflows are, for the most part, not just arbitrary. The reason doing asset work during final layout is, as you say quite rightly, a bad idea is that animation, techanim, FX and lighting all depend on decisions made by rigging and character design - no use grooming hair on an asset if you don’t know how many arms the thing has yet, to use an extreme example. The problem arises when you take that line of thought too far and say “no asset modification should ever be done after layout” - which is effectively what library overriding linked assets locks you into in blender. There are many very common scenarios where minor, local asset modification is very important to make good animation, many of which people have already raised in this thread.
As you said there are useable workarounds, the reason I bumped the thread is to post my own workaround, hopefully someone finds it useful. But all the hacks we can use to circumvent this limitation are counter-intuitive, cumbersome and often behave in unexpected ways. Which is why I say we need a better shape key system.
2 Likes
I’m not even that into animation and I thought that this feature was strangely not available. Considering that 3.0’s milestone was to add an Asset Browser to be more workflow friendly as a step for becoming industry standard, the override features are still lacking. Considering how Blender has a LOT of other features that needs improving that most users aren’t even aware of, I’m just hoping that whatever they prioritize for now would be the most productive in improving Blender.
I’m not sure when the big animation revamp is happening, but it’s long overdue now (not pointing fingers) and refreshing the shapekey system would be a perfect target to add to the list.
1 Like
Is there any updates on this?
Not that I know. The animation module has meetings every other week where they discuss future plans and try to best outline the future of aniimation in Blender. You can read the latest report here : https://devtalk.blender.org/t/2022-09-15-animation-rigging-module-meeting/25812 You can attend the next one and suggest this for consideration.
Spent hours looking for a solution for this and found this clever add-on by Demeter Dzadik 
PS I had to use the “Link” option in the Add-on preferences to get it working.
https://github.com/Warnakala/blender-studio-tools/tree/master/geonode_shapekeys
1 Like
Just adding my voice to this old thread, and hoping there’s a built in solution at some point soon.
What might be interesting to some is that there can be a stage after animation and before lighting. I’m working at a decent sized studio now, Maya based, as a technical animator/QC. Doing what is often called ‘finaling’. What we do much of the time is check for and correct small mesh intersections that occur in the day-to-day of animation. Especially with character clothing it’s a pretty essential step. It’s not really feasible for rigs to foresee every possible pose and situation and cater for that in the rig. And we need to address these issues quickly to get shots to lighting. Sometimes we’ll send back to rigs for fixes, but mostly we just need to fix things and get it done.
So yes, mesh-level editing is definitely used every day to address these issues. Animpolish and StickyMods by friggingawesome are good examples of useful Maya tools to aid this process.
I can’t wait for Blender to have this level of access to mesh data in linked files, somehow. Pretty please 
Looking forward to trying this geonode_shapekeys tool above though. That is a step in the right direction by the sounds of it.
Here’s the latest link for it I believe:
2 Likes
Hello Hammers !
I suspect being able to add shape keys to an overridden mesh is going to be difficult.
That would probably means making mesh data local since shape keys are very close to mesh data in the chain. Maybe if at some point we can add attributes to mesh data, then it’s possible to store modified points position and do shape keys… This is probably going to append at some point but not in the near future I fear.
Eventually geometry nodes might help to solve this issue since there are some plan to be able to bake geometry and be able to edit that as long as the mesh stays baked.
This add another issue since it’s probably going to be difficult to update the mesh after that, since the reference stays the baked data. So in some ways it’s very similar to the addon you posted about, but maybe in a more organised way.
I’m wondering, how helpful it is to keep data linked and overridden at this stage. I don’t mean for the whole scene but for these fixes, what kind of change you generally want to propagate from the original data to each shots at this stage, after “finaling” animation ?
Would that really be an issue if some of the cloth data or else is made local to the shot and therefore easier to edit ? On the average, how many object do you tweak per shot ?
In the end there is very little you might want to change, isn’t it ? since topology change, skin, rig or else would break animation.
1 Like
Reading this thread (after the other one complaining about the state of overrides and assets), I am really starting to get the impression that what the BF has done was only implement the bare minimum needed for their cartoons and art house films (for instance, you do not really need to add much for a cartoon dog with 2D shading). The team is really good at getting a minimum viable product out the door on a decent timeframe, but the chance is still there that old-school FOSS decision making (the type which causes professionals to flee) manifests itself once again and the ball is dropped, then lost, followed by the team forgetting it was ever there.
As of now, the best bet for further work will be when the funding and the manpower is in place for the big animation overhaul to finally come to fruition (as it has been delayed a few years and is one of the biggest things still separating Blender from Maya in terms of viability for professional work). Once again, the Institute needs to put the cartoons away and look at projects that will require film-quality assets and animation (like Heist, but even more ambitious).
Blender already has solutions for the transfer of attributes and data amid topology changes (as anything who worked with Geometry Nodes can attest). For a lot of common cases where overrides help (ie. where changes in things like topology are not completely different) it could work.
1 Like
Hum, I don’t have the same feeling as you… To me they must follow a step by step approach, since it would be super risky to take blender 2.79 and do all that it takes to make the most rock solid software that can do the most complex projects for blender 2.8.
Actually I suspect that the issue here has more to do with mesh data structure than overrides.
But for sure, overrides looks like a slow and painful way toward a better blender. It’s been years and it doesn’t feel finished yet :S
But to me it looks like they make the best of their budget, developers pushing code the weekends should also tells us that they really invested and it’s more than a regular job to them !
Sure ! what I don’t think is possible yet is to add some new attributes to a overridden object, without using geometry nodes. Which would basically solve the shape key issue.
Anyway, it’s always possible to work around, you can always create some scripts to circumvent many limitations of the software, and geo nodes is also an amazing tool for that.
1 Like