BMesh Preview #1

Hey all, did the first vid of BMesh development branch:

wow that looks cool!
What is a Bmesh?
It looks like you can skin a surface without worrying too much where the vertices’s are placed ?

man… i cant wait for this to be in the Blender core :slight_smile:
the useability seems a bit strange to what i’m used to, but the functionality is all there.
This is sure to help me speed up my modeling in Blender.
(if people dont like this… the edit mesh will remain in blender i’m sure)

That’s pretty sweet. Looks like we finally get true N-Gons.

Do we also get a legitimate beveling operator?

I know it’s a demo video and it has to show off more than just one tool, but there’s a bunch of things that could have been done a bit more efficiently :o You can just connect those edges to create a new edge in between them for example, instead of doing cuts, selecting verts, creating faces, etc.

Show people the true power of the connect tool! :yes: :slight_smile:

As far as I’ve seen in the SVN version, it’s a replacement for the current mesh editing system.

Personally I think it’s (going to be) a hell of a lot better, too. BMesh will give you pure modeling-zen :RocknRoll:

This is really neat! (On the other hand I am quite happy that I have old quad experience and habits that are still best for organic (subd) modelling.)

Blender 2.50 will be really really amazing. Actual development in Blender is nearly insane. In one year there are upcoming all long awaited features.

Big thank to all coders, you are doing excellent job!!!

oogsnoepje: he he you are right, my mistake, I just forgot about that tool XD I’ll keep doing better videos when more tools are added or so, actually I didnt use all the tools available in this one :slight_smile:

Oh and yes, edit mesh will disapear, and you will not miss any of the old tools dont worry :slight_smile:

The new bevel will be available soon, briggs just needs to port it over to bmesh branch

Dang. That’s beautiful.

Thanks for the video! Its nice to finally see some of the new mesh developments. The video was fairly confusing for me because you were using a transparent view mode and a lot of the time it was hard to tell what I was looking at. I would recommend using a solid shaded view next time you do a video like that.


sadly Ocluded mode is not implemented yet in bmesh, you cant select anything in that mode

Also please turntable view and perspective not orthographic for the videos.

This looks neat for doing certain types of modelling. I’m usually against n-gons because I feel they encourage laziness but in those situations, it would improve modelling speed.

Do you have an image of the result of the topology after applying a subdiv modifier onto the house?

theres no modifier support yet, bmesh is pretty much what you see in my vids, I’ll keep aking new vids as things advance :slight_smile:

I can already start thinking of how much this might simplify stuff…but also unfortunately how it can be abused :wink:

Just out of curiousity, I never found out if we can do N-Gons now…is it possible?

Those YouTube vids really help when explaining a concept!

Well, now that I think of it, it’s not exactly N-Gons, but it helps speed up a lot of stuff, instead of having to think how many times you want to cut your object up at the beginning.

I was wondering while reading this thread: what software do you people use for making those instruction videos. Is it freeware/open source. Does it support audio commentary?

Bmesh stands for ‘Blender Mesh’. See the (rather old and outdated) design doc on the wiki here (warning, highly techincal boring coder stuff!)

Edit mesh is long past its prime. I designed bmesh to be capable of everything editmesh is and more. So far it has proven to have fufilled these design requirments. Anyone who doesnt want to use new functionality won’t have to. At least thats the plan… we will see how it goes.

Everyone is entitled to their opinions, although I can’t quite understand this one to be honest. There are plenty of problem domains where N-Gons are more appropriate to use than quads/tris (yes,even in organic modelling as well). Further its never the job of application developers to hold back their users based upon some nebulous concept of not making them ‘lazy’.

Regardless you will notice from the bmesh wiki that n-gons were not even part of the requirments I gathered for the structure… it just happened to be that a N-Gon based structure was the best way to accomplish the design goals.

Probably wouldnt look good for the house unless you were using creased subsurfs. However CC subsurf on N-Gons can be a good thing. I whipped up this 5-10 minute example in silo a long while back to illustrate:

Notice how triangles, 5 sided and even 6 sided patches create nice loop patterns on the subdivided surface. Thats a lot of work that you save from having to place it by hand.


Briggs,can you tell us something about the bmesh speed,compared to edit mesh?
With time Blender modelling is going to slow down,I hope bmesh can be at least competitive in terms of speed.

Looking forward to seeing this work in the next release!
There is always the worry that important projects like this one will never see an official release - but I am hopeful.


In the best case, Bmesh is orders of magnitude faster than editmesh. The ‘worst case’ scenarios right now have to do with undo pushes and pops and editmode enter/exit. Just like editmesh, the bmesh structure is recreated completley in these cases. The only problem is that with the complexity and ‘rules’ of the bmesh structure, its potentially slower to do this. Actual performance I havn’t been able to measure yet since there are other issues to iron out first like faster memory allocation.

All of these are technical problems that can be overcome if needed though, I wouldn’t lose too much sleep over it.


T-Joints are something similar than N-Gons
and to be honest they don’t promote laziness
they are very good.

If a modeler does not understand it, than it is his
own fault and not of N-Gons. For texturing it can be
very helpful because it makes the mesh less dense.

Question is how good N-Gons will work with SDS.

Ngons surely look like they’ll help in archetectual modeling.

You go Briggs, I know that Bmesh has a long way to go yet but it’s getting there and will be a bit better then the old mesh structure.

Will this new mesh structure work in the Game Engine too, or will it be converted to Editmesh for that?