Yes, i use same work around, but it is painful especially when you work on UVs…
When i go over 150 000 tris, everything start to be slow and laggy…
I hope they improve performance in future, because today is already standard have 100 000 tris models in a game or whatever real-time graphics… And during the production is polycount even higher…
But otherwise, my user experience with 2.8 is great so far!
The problem is they promised to fix the edit mode performance before the final release and as it’s look like they skipped completely
B2.8 it’s awfully slow in edit mode , you can’t release that kind of software as stable and the fun is back in 3d it doesn’t make any sense to me
7 Likes
Regnas
(• I don't speak English "by default", so... )
419
Unfortunately, if editmode performance was something that could’ve been done in a few days, the devs. would’ve done it. This is compounded by how expansive the targets were for the initial 2.8 release (ie. tons of technical debt). In hindsight, the list of targets could’ve been narrowed a bit more so as to have more time for such issues.
The glimmer of hope in this is that the BF has an army of devs. and massive funding now (in part due to the focus on making things feel more standard). They have to address these lingering issues if they want to sustain that large team and prevent brain-drain to commercial vendors.
Viewport Performance was never good in blender.
Folks allways told that it would be tackled at the end before release. I feared that a long time ago, seeing so much focus on fancy stuff like eevee and imo needless ui changes. One of blenders best advantages was modeling compared to other 3D packages. Now editmode is slower than before🤦♂️
Bad Joke
At this point the BF has to release something, they can either launch 2.8 now with the warts (and bring about the return of the 3 month release cycle) or postpone it to some undefined date in the future. The latter would’ve generated a major wave of pessimism regarding the BF’s release and development planning (not to mention Ton’s competence) and could’ve driven away a lot of much needed volunteer development.
They even put Brecht in management mode so they can even do this next week (so part of the issue was the BF being resource-constrained like usual). I still remember the wonderfully pessimistic discussions here when 2.5x seemed to be stuck in a never-ending beta pattern.
But before 2.8 many people claimed that the main problem that Blender had was the UI… Now that they have a UI and LMB select, they are discovering the performance and feature limitations of always. Which I think is good, apparently UI discussions are starting to disappear, and people are starting to worry about things that really matter, like performance and features.
So now just being patient, performance improvements will surely come after 2.80.
Thing is, blender performance in editing was okayish. Now it’s worse than it was. Depensgraph was supposed to add performance gains and it did but not across the board (edit mode right?). Personaly I really hope that we will get superb performance compared to now (btw there was edit mode perf improvement in 2 waves, stil suc*** tho). Also UV relax… whyy is such a problem to add relax algoritham when shi*** load of people are asking for it?
A bit biased, isn’t it
From my experience with Maya and 3dsMax, rigging and animation playback performances are more than decent in Blender 2.79, plus it allows CPU and GPU open subdiv acceleration. I rarely use/edit high poly, sculpted meshes so I can’t talk for this part though. One thing is for sure anyway, Blender 2.8 viewport is slower than Blender 2.79. Let’s hope it will be fixed. There have been many core changes related to this such as Eevee, the new depsgraph and the new open subdiv implementation, it will probably take a bit of time to get the most out of it.
Welcome to FOSS, it’s normal in this world for developer decisions and priorities to sometimes not make sense. Even Ton has said the massive influx in dev. fund money is a sign they need to start improving the way they do development and project planning.
I wouldn’t be surprised if the money does force the BF to start working harder to avoid the tropes associated with FOSS development, now that there’s the potential for real consequences if they don’t (ie. greater than someone starting a fork of Blender).
As much as I would have liked an improved performance (or one that is at least up to par with 2.79) it probably wouldn’t have helped as much as the new features and UI in terms of getting more exposure and with that, more funding.
Somewhat of a different topic, but it doesn’t seem like the primary use of the recently acquired additional funds would be ironing out these issues, but to ensure that the foundation in terms of coordination, documentation and code standards makes it easier for new developers to join, be it volunteers from all over the world or coders they hire.
Lets hope that with growing funds there will be a dedicated team focusing on these issues. We all know how huge tasks like these can be, it takes other softwares many years to make such changes, if they ever do it.
Don’t get me wrong, I’d like to see a snappier undo system, high poly modeling or hair grooming - that surpasses the ones seen by the competition - as much as the next guy, but apparently that wont happen for quite a while.
The logic would like a developer to be charged with dedicating himself exclusively to the performances …
but maybe it’s not even so simple, since everything is connected … I begin to suspect that an entire codequest would be needed exclusively dedicated to this … something like Mozilla did with Firefox with his Servo and Quantum projects …
Try to select an object if you have hundred of objects in the scene. Blender took several seconds in b27 ->hard to work with. Try the same in 3ds ->almost no lag.
Yes, Objectmode Performance in b28 improved but edit the opposite
Maybe 3dsmax is a bit better, but try to select an object in heavy scenes in Maya… very laggy too.
Not to mention the frequent crashes, plus long loading times when opening a scene in Max/Maya… but grass is always greener on the other side
The whole viewport project was started because people were clamoring for a faster viewport. When they started it and decided to make it pretty, there were jokes about lens flares, but we were assured performance was the priority.
Well guess what. We got bloom, and depth of field and what not, but the performance still sucks like a Dyson, now more than ever.
Well, at least Object mode got a little better.
Ugh, it sours my mood every time I think about it.