This infographic by XRG does a great job at explaining some of the differences between quad modeling and the upcoming new Bmesh module.
Here it is in full size:
What are YOU most excited about?
Oops! I don't want to take credit on this one. It's actually XRG from this thread: http://blenderartists.org/forum/showthread.php?243978-BMesh-vs-non-BMesh-where-is-the-benefit that made the infographic - I just reposted it to twitter because it was so informative!
Is there going to be provisions to make it backwards compatable to regular blend files?
Back when I first jumped from Wings to Blender I was confused as to why I couldn't use N-gons. I clearly remember reading somewhere that Blender's forcing you to work with tris and quads was superior because it made you think about your topology and doing it "correctly." Now N-gons are okay?
So you read somewhere from someone without competence enough :D
Look, it could have been on the Blender forums from ages ago. This was around 2.34 when I found Blender far far too frustrating to continue with. (I came back at 2.48 and was blown away by the progress.) WhereI read it and who said it doesn't matter much now. I was sucking up as many tutorials and hints as were available at the time and that one stuck in my mind.
As long as nothing strange happens with lighting or deforming for animations when there are N-gons present (or they convert easily to quads and tris without adding a lot of work...) then I've got no problem with them. Mostly I wish the Game Engine would see some love since real time applications are where my interests lie.
Yeah, it’s funny that. All the mesh-modelling tutorials seem to concentrate exclusively on quads, and tell you to avoid tris and anything else, and pay attention to your poles. So what’s the big deal about adding N-gons?
Well it's important to mainly use quads when it comes to animation because when you start animating, quads deform more nicely because they have a better edge flor thwn other shapes do. They also smooth out better.
While modeling though, N-gons are easier to use because instead of figuring out how to make quads fit your shape, you can just draw and manipulate the shape itself. Things such as Google Sketchup and 3DVia Shape work kinda like this, but if you look at the wireframe, you'll see loads of triangles you may have to fix later on unless you don't plan on animating, and even then you can still get some odd lighting and UV map issues.
I think even Autodesk's 123D program (their version of SketchUp) does the same thing and auto triangulates everything.
The only time it's really good to use triangles is when making things for games, and even then, it's good to start modeling in quads and let the engine convert it to tries if that's possible.
Joshua, there have always been the n-gon-hating, blender-shortcomings-justifying group of users. They have never represented the development efforts, nor the Blender userbase in general. But they are very vocal.
I've been using Blender since 1998, and I've always been supportive of adding n-gon support to Blender.
It's important to note, of course, that n-gons is not the only benefit of bmesh. Bmesh also provides a more flexible, easier-to-code-for base for further modeling tools development. So even the n-gon haters should support it. ;-)
i wish we have a button that says "select all ngons" or "find all ngons" just so we could keep track of where our ngons are. Ngons are great, but id rather take them away afterwards specially if i were modelling organic meshes.
Already there, you just have to select by number. set the number of vertices to 4 and set your option to greater than 4 than it will select all the ngons in your mesh and you can get to work cleaning up.
Everything has its uses. People who are into solid modeling and who don't care so much about how a mesh would deform and such, probably wouldn't remove the Ngons afterwards. If people would want to use Blender for more engineering type work, this would be very much ok. If you want intimate control over how the subdivision modifier works and what edges you want to crease edge, etc. a good edge flow is crucial and probably only quads will do. It would be interesting to hear want a self-proclaimed topology snob like Jonathan Williamson thinks about this.
" It would be interesting to hear want a self-proclaimed topology snob like Jonathan Williamson thinks about this."
Sad to read that
I don't mean this in a bad way, I mean that I'm genuinely interested in hearing how an important and respected blender instructor would incorporate these fascinating new options in his work flow. New developments can trigger people to rethink their methods. This is a good thing, right!
Perhaps "Topology Aficionado" would have been the best phrase. Or Connoissuer!
To the original question... all of the methods are simply means to an end. Using the right method at the right time is how you get superior results.
(I can't believe I just mistyped "coonnoisseur"... I had the right letters in my head but they got reversed as I typed them.)
Argh! "Connoisseur." I can't get my typing hands to wake up today!
Wanted to correct you and say "connaisseur"... but i just found out, strangely enough, that the English spelling is indeed with an "o" instead of an "a", like in the actual french word.
"connoisseur" was the exact spelling in french back when a Normand became King of England, period at wich many french words transited to the english idioms (bœuf (boeuf) -> beef etc). Many have since evolved to their modern counterpart in french while they remained unaltered in english.
He himself uses the word snob though.
Aha! Then "Self-described snob" is excellent!
Sorry guys, I misinterpreted :)
Jonathan Williamson was the guy who did the face modeling a little while ago, you can read what he thinks if you really want to...
Do you have a link to where he talks about bmesh? I would be very interested.
Thanks for the link. The timelapse looks very bizar, somehow unnatural. Blender is the only modeling tool I have ever used and doing blendercookie tutorials and the vehicle training series are the only training I ever had. I feel like I I just got used to the idea of quads over tris and now I have to start over again *grin*. I've said it before: Blender develops so fast I don't have time to learn to use the new feature before yet newer features appear.
I can't think of another software project that develops as fast and consistently as Blender does.
we have had an addon "fake_knife tool" since 2.53.
I keep bug reporting bmesh bevel.
again the benchmark is "mesh_bevel addon" since 2.53.
BMesh is great, a very needed polygon system and one of the shrinking list of things Autodesk's tools do better than Blender. But to answer the question above, what I'm most excited about BMesh, is that Unlimited Clay is sure to follow in it's wake. The digital sculpter in me can't wait to use Sculptris-like technology on my Linux machine :-)
what I'm most excited about BMesh, is that Unlimited Clay is sure to
follow in it's wake. The digital sculpter in me can't wait to use
That's what i'm excited about too :)
@facebook-1369107853:disqus +1 on that, especially with complex meshes....
And couldn't something like Remesh be adapted to clean-up to tris or quads?
Come to think of it... what would happen if you tried to Remesh a model with ngons now... ?!
I'm not sure but even if it doesn't work, since only triangles are handles by you graphic card, it means that the ngons are converted to tris at a lower level. So I don't think it would be hard to adapt the modifier.
Remesh always outputs quads, doesn't matter whether the input is triangles, quads, pentagons, etc. Of course, you'll have the standard "convert to triangles" button too.
Dunno why everyone is so freaked out about converting NGons to other polygons. Really not rocket science here :)
We get it's not rocket science, but could become a load of molar-gnashing clean-up work on complex models. Actually, hearing "clean-up" and "new feature" in the same sentence about any software application makes me cringe. But if it's clean-up with Remesh... then thumbs up for Bmesh
I think these guys are like me. we don't know much about the mystical language of C programmers lol. I for instance understand how it is important to convert polygons to quads and why but at the same time I don't know the math behind the scenes and if the variables called by B-mesh are compatible with Remesh. I think that this must be what the above comment is about so It's really nice to have a respected developer just say, "nope, don't worry." sometimes lol.
By the way, thanks for all your open source work Nick, been using your software since Sharp Construct and still fall back to it when I need to put a mask on my model as long as I can use just those tools for what I'm doing.
I'm looking forward to a car modelling tutorial with NGons!
Question... It will work with Blender GE?
At a guess, no. BGE supports trigs and quads only, and actually converts any quad faces to triangles when running the GE anyway.
Of course theres nothing to stop you using NGons for your modelling, then converting the model to quads/tris, I expect this will be an addition to many users BGE workflow.
in theory, since the game engine will convert everything to tris anyway, then you could assume that it would do the same for ngons automatically. The problem with that is that the results of that would most likely be less than ideal, and if the area in question is anything that would require any sort of deformation, then you'd probably want to convert it to quads or tris beforehand anyway.
I don't think anyone could say that bmesh is a downgrade.
I think bmesh will give us more than just ngons.Finally we will be able to use a modern modelling system with freedom and cool/useful accompanying/future toolset.
Makes me think of the intuitivity of Sketchup. Can't wait for 2.62!
According to the developer meeting notes of jan. 22 you'll have to wait till 2.64 or maybe 2.63 at best for BMesh
Sounds great, I look forward to using it.
It's odd to me that there's seems to have been resistance to NGons in blender. The commercial programs have supported them for decades. The lack of them has frankly kept me away from really trying Blender. I've wanted to create a character from concept to animation in Blender to assess teaching it in a program I was a part of, but lacking this basic function meant wasting extra hours working around the basic poly functioning of the program.
I learned Subdivision modeling with N-Gons using Silo. however, this certain "freedom" always gave me headaches later, in trying to fix all the ngons i've created to solve "difficult" areas on characters. Next thing you have to learn is how to dissolve Ngons into tris and quads using addtional edgeloops f.ex. rotating edges etc. and leave some cases without a solution.
So for me modeling in Blender with only quads was extremely refreshing. Only thinking in quads may take some extra steps in the beginning, but saves you time in the long term imo. And helps you to think ahead about how to model things. This is just my experience...
Also, I don't see the advantage of using ngons on the models above...
so you never "really" used blender?
I've been tinkering with functions in Blender for the past couple years. I've been following the 2.5 changeover pretty closely. The program is just not handy yet. I have Maya and Max for the contract work that I do, so when I run into a function like this quad modeling limitation, it doesn't make much sense to invest my time when I have better tools. For hobbyists, this program is great. For professionals, it doesn't do enough right to justify the time to retrain.
To answer your question, I was assessing this tool for how useful it would be to train students. I started a humanoid. I Got to the point of starting finer features, and very quickly said that this program lacks a key function that most major packages can handle. Given that students would need to learn something more similar to the packages most studios would use, it didn't make sense to pursue. I've been patiently waiting for this function to get implemented before trying the program again.
your opinion is very interesting, then you are a true 3d professional? can we see some of your work? it looks that you have a lot of experience
I'm a web developer, not a professional 3D artist, but I've done lots of 3D modeling both as a hobby and recently for a few smart-phone games. I own Maya, but I haven't launched Maya in almost a year because Blender is much better in many ways: Texture painting/baking, performance, built-in camera tracking, multi-res sculpting, constraint and modifier organization, even some basic modeling features like the cut-loop tool are more intuitive than Mayas. That said, Maya still has some better features as well, most notably Ngons and Normal editing, but that list seems to be shrinking every few months.
Games industry mainly, Max.
Philip, that's why I've been watching Blender intently. I see a lot of things done right. There's just some oddities that stop me from delving very deep. Multiwindow support was a great welcome, since I have dual display. I'm looking forward to mouse over menus, and I hope they consider a shelf function for common tools (Maya like functioning). I think it's very liberating when a program makes several ways to go about an action. Each artists has their own workflow, so it's quite gratifying when a program allows for it.
File > User Preferences > Interface > Menus: > "Open on Mouse Over"
May be that i misunderstand you, but AFAIK there is a toolbox with the most common tools.
While hovering over the 3d view [T] key toggles the (mode dependent) toolbox on the left hand side of the viewport, [N] key toggles the properties panel on the right hand side of the view. Even works with the classic quad viewport. (ctrl + alt + q)
Very nice! I want this.
Is it going to be integrated into 2.62?
More like 2.65, perhaps.
It's projected release is 2.64 or possibly 2.63!
will this cooperate with UVs, Modifiers and the rest of the things?
That is exactly what I was thinking .... I would only hope that BMesh will work on all levels in modelling tools , that goes alone with :) booleans and other modifiers , O I also hope that they would :) write an optimizer too .
Yes, it works with all modifiers.
Dont like Ngons, im not excited.
It's a common misconception that Bmesh is all about Ngons. They are just one of the many new additions and in my opinion one of the least significant.
You're right that Bmesh is more than that.
But " least significant", don't say that to any sketchup user at least ;)
Like Greg said, NGons is just a small part of BMesh, from what I read some time ago, the actuall big change and motivation behing this project was modernizing the dated and lacking mesh engine in blender. It really changes a lot of things under the hood for the better, like performance, better and different tools for modeling not possible before (not because of NGons but because the way information is handled), and the way data is accessed allowing for much better, faster and efficient modifiers for example. Ngons are a great addition and probably one of the most visible to the end user, that's why it is talked a lot.
For the quad puritans they are not forced to use NGons, in fact like it was said it is encouraged to to "quadify" in the end, though not strictly necessary, since all quads/Ngons are in reality triangles simply "hidden" from the user whether in the game engine, viewport or watever, because all digital 3D geometry is always represented in terms of triangles, even NURBS. Anyway NGons should help modeling even for quad meshes.
As for the modifiers, UVs, Game engine, Sculping, etc. I believe it is being worked on, that is one of the reasons we will probably see this only in 2.65. But expect everything to work even if at start not all features are supported. Addons probably need to be fixed, but the game engine should probably work fine with this, since like it was said all meshes are actually triangles anyway.
Clean Boolean operations ... what else is sexier ?
I am ready to use N-Gons, however, I gonna continue with my "quadian" vision and modeling and use N-Gons when I really have no other way to make a certain shape to stick to my mesh. If that prevents pinching due to the surreptitious presence of poles, please by all means, add it!
I am glad this is coming to blender. Not having N-gons has been a major problem in modeling efficiency, especially when compared to to packages like Max and Maya. I am excited, maybe I will get a lot less frustrated when modeling in blender now, Its other features are awesome, B-mesh brings the modeling functionality and workflow current.
If you get frustrated modeling in blender, maybe you're doing something wrong
Or perhaps Joe is just not used to it. I've been using blender for about 2 years now, and on Monday I started learning Maya for work. "Frustrated" isn't a strong enough word for how I felt. But after just three days I'm more comfortable modeling in Maya than in Blender for the simple fact that I've used it more these last few days, regardless of the fact that Blender is better for modeling.
In my opinion, Ngons are no big deal. You don't really want them in your final model, but they are useful to have while modeling to control edge-flow and make sure you're not sticking looks in places you don't want to.
Hmm... The first and only serious 3D app with me has been blender with it's max 4 sided polygons so I've never missed N-gons. N-gons seem to be faster to do stuff depending on the situation. And rather alienating thing for me. I won't say nay for them, that's for sure. But I'm kinda hoping this gets over so new stuff can be made on top of the Bmesh system rather than N-gons themselves.
I am interested about all this as it does open up more possibilities in modelling, but I've never used N-gons before so, this is kinda like, the yard going to bet bigger when the current one suits me just fine. Still, I'm quite curious. :)
Also seems like the people coming from max and maya are really missing N-gons in the current blender.
I do my modeling in Wings, which handles n-gons by way of braking them into triangles internally. This lets you do an operation on the n-gon as if it were not broken, and creates new triangles internally during the operation. This seems a great improvement to blender. It's not that anyone really wants n-gons in their final model, but being able to perform bevel/extrude/inset operations on n-gons means you don't have to think as many steps ahead while modeling. I can clearly see where the model shown would be more difficult to accomplish without ngons and how options for further modification might be restricted.
Anyone wanting to try it out (Only tested the Win64 version), here are some test builds from Graphicall.org
Linux(compiled on Fedora): http://graphicall.org/807
@Article: I'm very excited about this getting intergrated! Also, hows that topology sculpting thing-a-ma-giggy going? Haven't heard of that for a while...
I think this article should focus on BMesh's goal, NGons is a small part of why BMesh is an upgrade. I think people click on this link thinking... oh, it's just going to tell me about ngons... and then it does. I think it's a more powerful article if it shows something most people don't know. NGons to me is the icing on the cake, but it's not the cake.
Something to note: Adding 1 subdivision level to any manifold mesh, including those with N-gons, results in a mesh that is ALL quads. Think about it- take a pentagon, and draw a line from the midpoint of each edge to the center of the pentagon- all quads. It works for polygons of any number of sides. The only exception is an polygon with an internal hole (for example, cutting a square hole into a quad without any edges connecting the inside boundary to the outside. I don't know if B-mesh supports this, and this might be "non-manifold" topology, which Catmull-Clark sub-d's can't use.) Highly concave N-gons would look really ugly and possibly self-intersect when subdivided, but the same is true of concave quads- they are to be avoided.
Making a few assumptions here, but even if export of models with N-gons doesn't work at the beginning, it should be possible to subdivide once, convert to the old mesh type, and export that. Further subdivision levels can then be added in whatever package you're sending it to. If you didn't want the mesh smoothed, you could just turn off smoothing in the subdivision modifier. This would result in a denser mesh, but at least it would get your model out.
I've been 3D modelling for many years. Making the jump to all open source software this year. I can't imagine not being able to use N-GON's at the very least during the construction process. This would seem to me to be priority #1.
This is the main feature I'm looking toward to in Blender.. When I switched to blender after working in max and maya, n-gons is the feature I missed the most.. This will seriously speedup modeling work-flow!
When n-gons will be implemented, a lot of people will switch to blender from other 3D apps. Lack of n-gons held a lot of people back from using blender.
Like Sketchup. Yay!
To add a profile picture to your message, register your email address with Gravatar.com. To protect your email address, create an account on BlenderNation and log in when posting a message.
If you are a human, do not fill in this field.
Notify me of followup comments via e-mail. You can also subscribe without commenting.
Notify me of new posts by email.