Bringing real modelling support to Second Life

[Mircea Lobo]

This is a major suggestion, and like other things something I doubt will happen in years. I've been building stuff in Second Life for a long time, but more recently been working with actual 3D modelling and tools like Blender. I'm starting to realize that building things out of spheres, cylinders and cubes is starting to get old. But at the same time, I love SL's ability to build in-world together with other people, without needing any external tools. At this day, I believe SL should consider mixing the two together.

What this means is giving the viewer ability to create and edit meshes in real time, and the features a basic modelling program has. Currently, you can only spawn a primitive with a hard coded shape (cube, sphere, cylinder) and change several characteristics such as hole, shear, taper, and more. What I wish would happen is being able to select individual vertices of the cube and drag them around, subdivide the shape to earn more vertices, extrude faces, separate / merge meshes to / from multiple primitives, and anything else that would be useful.

I won't get into the advanced stuff at this point (such as in-world creation of skeletons / rigging / animations or UV mapping), even if I hope all this will happen over the years. For now, I just believe a replacement for the way you edit prims would be extremely welcome. Whenever I edit a cube in SL, I'm thinking "Why must I set a taper value instead of selecting the top vertices and scaling them on the X axis?" This is the next major change I  totally want to see in SL personally.

Now I'm not suggesting the removal of the current system either. This would be a horrible thing for current builders used to the existing way for years, even if I wouldn't mind personally. My suggestion is turning the current editor into templates, and maybe allowing users to create custom ones too out of meshes. For instance, the cube would be a template, and setting the hole, shear, taper, twist, etc. just properties and keys part of it. But if you wanna free edit your shape, you would click a button to finalize your primitive and transform it to a mesh, and from that point you would edit it like a model. This would be an entirely optional process, and of course existing primitives would stay as they are and be editable like until now. This is just a quick idea of how the feature could be organized to keep the same system for current builders... better ones can probably be found.

I would like to hear what people think of this, and maybe take the discussion higher. I personally believe it's time Linden considers this, and I'd like to hear their feedback a lot. Do other builders want the feature? What about Lindens... enough to make it happen? Are there any plans at this day, and have any attempts with custom viewers been made to implement free vertex editing of primitives?

[Nalates Urriah]

There are something you seem to overlook.

Adding the tools to edit vertics and UVMaps would require adding a program like Blender or 3DS into the viewer, no small undertaking. That would also spike up the learning curve for new users. The time to maintain two design systems and the support needed to teach people how to use a Lab designed modeling system like Blender seems prohibitive. Expect the vertex editing to stay outside the SL Viewers.

As to editing the vertices in existing prims like a cube... The array of vertices that make up a cube, deformed or not, is built into the viewer. The vertices array does not live in the asset server. So, if you change a cube's vertices in your viewer there is no way to transmit that change to others in SL.

The use of parameteric primatives was done to reduce bandwidth and asset bloat. The change to the system to allow one to edit verices was the addition of mesh. Trying to take that to editing basic primatives or as some now call them: legacy prims, is just counter productive and seems redundant.

 

[LuciaRiley]

We're talking about Linden Labs here. They'd never add support, let alone this. .__. I'd love the idea of it, though.

[Mircea Lobo]

From my understanding, they wouldn't have to add a whole program like Blender into the viewer (that would probably be bad). My idea was that, the same way you can select a prim and move it around, you could select a vertice on it and do so. You wouldn't do it from an integrated program, but directly click and drag it in-world.

But I don't know the in-depth details of how primitives are drawn. I was hoping that the cube for instance is just a set of 8 vertices between which surfaces are being drawn. If the client + server could store the offset and information of each vertex, getting the surface to draw in proper position would be easy. Even if it would require major changes, I support the idea.

As for the learning curve, I agree. But like I said I'm not suggesting a replacement of the current building stystem, which would harm many builders badly. Maybe vertex objects wouldn't be primitives at all, but an entirely new feature like mesh support. I don't think adding modern features is a bad thing... in my opinion 3D environments and virtual spaces evolved a lot since the days when Second Life first started, and this change should be part of that evolvement.

[Conora]

I see your point it would be awsome if thay did that it woul be way easyer to create awsome builds this way.

But i dont think its gonne happen:

1 thay would have to write a blender like program

2 there making on money from upload fees.

so thay basicly would shoot themselves in the foot, a lot of extra work on creating a huge new program and less payment

 

[Mircea Lobo]

True. But at the same time, I believe evolvement and modernization is going to require this. Being able to model in realtime will probably become a popular demand in the following years, as hardware and technologies will evolve even more. Eventually another 3D app will add this feature, and if SL won't it will get left behind.

Modelling serious meshes will always be much easier and more stable in a program like Blender. So for topmost quality, mesh creators will likely choose to import and go with the small fee. So I don't think this would practically ruin LL's business much... but on the contrary such a feature will likely make SL even more popular. Those who aren't into it yet will probably go like "Creating meshes in real time in a virtual space straight from in-world ? That's an incredible feature... I need to check this out".

[Masami Kuramoto]

http://en.wikipedia.org/wiki/Verse_protocol

http://www.letworyinteractive.com/b/2010/08/verse-integration-for-naali/

Adding mesh modeling tools to the Second Life viewer is not feasible. Remember that people can't even agree on what is the proper UI for a viewer, and Blender's UI is frequently considered cryptic. It would also make the viewer much bigger, introduce new bugs etc.

Verse integration may be a better idea because it would keep the tools separate from the viewer, but then again, the last Verse-enabled version of Blender is several years old and horribly outdated. There were some suggestions to revive Verse in Blender 2.5+, but I don't expect it to happen any time soon.

It seems that demand for collaborative mesh modeling is rather low.

[Mircea Lobo]

An integrated tool like that seems nice too. Maybe with mesh support, someone can already make an addon for Blender that can connect to the SL viewer while it's running, and hack the client to preview your mesh locally in-world without updating it on the server till you're ready to upload. But I'm not as interested in this personally. I don't believe it's fun if you can't do it straight from in-world, and have your edit be visible to all avatars in real-time (as well as collisions updating).

Under my imagined implementation, it wouldn't require a special UI either. Like I said, the same way you select a primitive by right-clicking it, you could select a vertice by right-clicking... then drag it around the same way you drag prims currently. Buttons for subdividing the mesh would be located somewhere in the edit window (where the Object settings are now, where you set hole / shear / etc) or something like that. So in total it would feel and work a lot like editing does now.

I still don't see it as something that impossible to make... but I have little experience with the technical part of SL, so I have no say over someone who knows the prim code even the slightest. Maybe someone who makes custom viewers will give this a try. If they need a custom server, OpenSim is always there.

[namssab1nad Piers]

Actually, the type of 3D mesh system that should be integrated into SL should be Google SketchUp or at least based on it.  It is simple, intuitive, UI is GREAT; and best of all you do not have to be a rocket scientist to work in it.  If none of you have tried it, do so.  There is a free version and a pro version.  Even if it was the just the free version that it was based on, it would add alot of functinality and build options to SL. 

[ColorAnyOne]

This would be the perfect place for a SAS(Software As Service) implementation.  How HUGE would it be if there was a place online that students could go to use Educational versions of software at a $/Hour rate?  Professors could charge $/answer.  This could be the begining of a school where students actually help each other, or get credit for helping each other.  There would be a "Destination" named Expression, where everybody was working with Expression.  Maybe even a way, for one student/Professor to view the screen, or a snapshot of the screen, of another student/Professor, inworld, and chat while mousing around the same shared view.  Software could be used on a rent-to-own basis, where SL would keep track of the amount of money spent $/Hour, or $/Unit, and if the student stuck with it they would own it.   The selling point would be dropping the cost of a course from hundreds of dollars per credit, to one of canned tutorials, and freestyling, leading to software purchase.  The Activation Key Please!

No monthly subscription fees, for software that doesn't get used.

Companies would be motivated to make "Lite" versions of their software available freely to the interested public.

Have you ever wanted to read only one chapter of a book?  Or maybe read about what another author's book says about a certain subject; all without paying the full price of a book? 

IE9 is sandboxed, and Chrome supports HTML5, it'll be a while.

[Coby Foden]

---

namssab1nad Piers wrote:

Actually, the type of 3D mesh system that should be integrated into SL should be Google SketchUp or at least based on it.  It is simple, intuitive, UI is GREAT; and best of all you do not have to be a rocket scientist to work in it. 

---

Implementing Sketchup (or any derivative of it) to SL viewer seem not to be so great idea. Chosen Few had something to say about Sketchup:

"The under-the-hood structures of models produced in Sketchup tend to be pretty amazingly disastrous."

Whole story here:

http://community.secondlife.com/t5/Building-and-Texturing-Forum/how-to-get-an-avatar-into-Sketchup/m-p/1525223#M6194

Sketchup is a nice simple tool to make for example a RL house plan and fill it with furniture.  But for making models to SL with it... not a good idea; for the many reasons found in the link above.

[Mircea Lobo]

Not experienced with Sketchup here so I can't say much about it. But like I said: Why integrate a modelling tool in the viewer? SL already has its existing builder (the edit menu) which just needs a few extras to allow this feature. They might be harder to implement... but if you can't model straight from in-world and by using the SL menu, I don't see a point (just use another modelling program and import your work when it's ready).

[ColorAnyOne]

ThreeDify Designer is an OpenGL based 3D solid modeling, CAD prototyping and view mark-up
application. It runs in either of two modes: as an ActiveX control in the Internet
Explorer and as a standalone end-user application. It features an intuitive, non-blocking
and customizable GUI, an import/export plug-in architecture, unlimited undo/redo,
up to 6 view clipping planes, associative 3D dimensions (angular, radial, point-to-point,
point-to-face, X-,Y- and Z-dimensions) for accurate model measurements, robust Boolean
operations on 3D meshes and 2D polygons, 2D mark-up tools, 3D slicing for arbitrary or
axis-aligned profile generation, polygon offset, one-click roof/wall/house creation,
local vertex editing on 3D meshes, polylines and polygons, and an array of other powerful
3D object creation tools (e.g., extrusion from complex and nested profiles, 3D polygon
and mesh creation from true type fonts, and optimal solid reconstruction from planar,
non-parallel contours).