dxf import?

While there isn't a way to directly import model files into SL, I was hoping to find that someone has done some work just cutting and pasting the contents of DXF into the world. It is the uncompressed textfile equivalent of DWG for AutoCAD. Has anyone done anything like this, is it even possible?

I'm having fits trying to imagine some of the models I've seen being done wholy with the given available prims at a sandbox.

Thanks in advance,
Eternus

Closest so far is my OBJ importer. Most tools I'm familiar with can convert DXF to OBJ. At least, I think.

Importer available here.

Awesome, my company actually makes 3D modeling apps and can export to OBJ as well. I was just finding a way to get from SketchUp to Second Life. I'll give it a try, but I imagine that will suit my needs perfectly.

You will probably not find this an acceptable solution in the long run. Those object importers use 1 primitive PER VERTEX. A single cube in OBJ format would become 8 prims.

At least I think that's how it works, if I remember right. Maybe it was per face.

As I understand, the .obj-import script imports each poly as a prim - basically, every triangle. So, a straight-up cube would be imported as 6-prims. But, a skewed cube would require the 8-prims.

Since each sim has a prim-limit, and each parcel an even smaller limit, prim usage is critical. Basically, each 512sqm. of land is limited to 117 prims... and 15,000 per entire sim. Import your .obj's with care.

Ouch. I'm still getting down the basics of makin stuff in SL, so I hadn't given much thought to it. So there isn't really an easy way to build your objects other than prim by prim inside of SL? Is LSL robust enough to create a modeler (has anyone tried?)

Thanks,
Eternus

This issue has been brought up many times. The most realistic answer would be, there will be no model import from any 3d format (save for quirky methods that read text files to reconstruct highly primmy objects) anytime soon.

While the SL method of building things out of parametric objects may seem clunky at first, I can't imagine anything that can't be build with the tools we currently have. They may be slightly rough approximations, but with a little ingenuity and good texturing, they serve their purpose well. Just keep at it!

Building your builds from an outside app and uploading it somehow takes away the fun of building things together and seeing things appear interactively. I would say, its the same like having a baby via artificial insemination.

I just got into SL a couple days ago. At first I was excited to play around in an open, modifyable 3D environment. The building process, to me, may be a deal-breaker though. I totally understand the reason behind it and all. It just seems completely counter-intuitive for me to use. Part of the problem, I think, is that I've been doing 3D modelling for over 10 years in various CAD programs. The SL build format seems terribly tedious by contrast. If there was even a standalone app to build models that was more user-friendly to people with CAD backgrounds, it would be great.

This comes up whenever anyone with any modeling experience joins SL. At first, the in-world modeling tools seem childish, even silly, and you spend 90% of your time saying "If only I could import my Maya models, or my 3DS models, or my CAD models..." As you spend some time in SL though, you'll begin to discover that this "limited" tool set, which you find so silly at first, actually forces you to employ creative problem-sovling skills and techniques that you would never normally consider. As a result, you end up becoming a better modeler both inside and outside of SL.

To use myself as an example, outside of SL I'm a Maya person. I can atest that my poly counts per average model have gone down tremendously since I started using SL while the quality has gone up, and my texturing skills have improved a hundred fold. The same is true for just about everyone I know who is in the 3D field who uses SL. Quite simply, it forces you to pay attention to detail in a way you otherwise woudn't, and that process provides tremendous growth potential for you as a modeler. Give it a few weeks and you'll see exactly what I mean.

Seems like kind of a form-letter reply.

So, if this comes up all the time with experienced modellers, why not make a better modelling interface? At LEAST a standard four viewport arrangement would be helpful. I'm used to working in a perspective viewport in Rhino3D, but you only use it half the time or so.

Here's a typical project I've been working on in Rhino3D:
http://www.deviantart.com/deviation/22254271/

I guess if this is what you have to work with to use SL, I'll be back when things advance a bit. It's not worth my time to go through a pain in the ass for what the outcome is worth.

Some would say that it's not the toolset, but the community that keep them here.

While the tricycle was nice, we do face real limits (as in other games) as to the number of polys and textures we can use.

Unfortunately, SL is a little different from other games/environments. While SL doesn't teach many bad habits, it doesn't really teach any good ones either. No builder in SL is going to know about eliminating execess polys and vertices out of their build. They can't. Unfortunately, that is a very critical skill outside of SL. Ask the average builder how many polys are in their build and they can't tell you. SL is the only environment that I know of where you measure all your builds in prims. (Which is the barrier to importing/exporting.) Ask a builder in SL about vertex shadowing, lighting or just shadow/lighting in general and you'll get a "Huh?" And so on and so on.

There is no denying that the SL 3D toolset has a lot of room for improvement. By itself, it is quite far behind the times. This is why I think considerable progress can be made by simply giving us the ability to import and export files from and to other 3D applications. Who doesn't want SL to look better? We are supposed to be the showcase of a virtual 3D world. We need to look that way. We shouldn't have to resort to a bunch of hacks to make something look passable.

There is some concern about high prim builds. What stops you from importing a lagtastic 2000 prim house? Nothing. It is the same nothing that keeps you from building it by hand (or script) right now if you have the land to support it. At least the imported one might look nice.

Here is my theorietical build process (using 3DSM):

1. Create simple build in 3D application using the exact number of objects as the number of prims, you will use in SL.
2. Select all the object files and export them in a *.obj file.
3. Increase the complexity of the build with additional objects (for all the neat shadow effects).
4. Apply textures.
5. Apply bump maps, etc.
6. Create light sources.
7. Render to file.
8. Slice the texture file from 3D application into multiple files for upload into SL.
9. Have script create the prims from the *.obj file.
10. Load the textures onto prims.
11. Adjust textures and prims location and size (rounding errors, etc.) for final build.

I think the problem we are running into with 3D applications is that we are trying to import objects that would never work even if we weren't creating a prim for each poly. SL objects need to be kept simple as possible and additional detail can be added in through textures. People who really want to build those 100m diameter spheres can ignore that last sentence.

I have been kicking around a script that would create box from an object file as single prim and not six. However... However, there are some caveats. Information such as size, rotation and location for SL all have to calculated from the vertices information for the center of the new prim. (Face information in the *.obj files an be discarded as there is no way to make a prim in SL without its full allotment of faces.) This is simple for a sphere, easy for the box, a little harder for a rotated box and will make your brain explode for a "cut", rotated, "hollow", squashed torus. Just the sheer number of vertices involved with tori. Sheesh. And the fact that "cut" and "hollow" only really have meaning in SL.

This has me really frustrated as I was working on some "organic" builds that use a lot of curves and those look like they would be next to impossible to import into SL. I don't any progress can be made there until we get the ability to import *.obj files directly into SL.

The other caveat I expect is that you have to use the SL's limitations when it comes to building in whatever 3D application you are using. No prim should be larger than 10m, no radii can be larger than 5m, etc. The size limitation might be ignored if your export script in your 3D applications was good enough to break up your objects in sensible way - a very big "if." Other limits such as the 5m radii limit cannot really be circumvented. A 20m diameter sphere is just going to look irregular.

I would like to be able to create solely on one platform and then import the final results into SL. It would be so much faster. The switching back and forth are real time killers.

Thanks Noel. I totally agree with you. I originally learned 3D using AutoCAD 12, building models poly by poly. In the last 5 or 6 years, all I've used is solid modelling in Rhino. 'Pure' soild modelling is a lot like it is in SL. You start with primitives and add/subract them to get the model you want. There's no mention of polygons involved.

Whoah, you posted while I was clicking 'reply' =)

It sure seems possible to have a clean, simplified external modelling program. You'd almost need LL's direct involvement though, and that seems unlikely.

Some issues come to mind:

Do prims have a fixed number of polys, regardless of size, stretch, etc.?

Does SL care how many polys are in a given prim?

Is the coordinate system absolute or relative?

Sorting out texture mapping could be a nightmare by itself.

In terms of OpenGL drawing, afraid they don't. Prims have different levels of detail, varying the closer your camera is to a given object. Compounding this is the fact certain things like twist also raise the level of detail. So in terms of pure polygons, it's a bit tricky.


Besides LOD versus distance, it doesn't seem to.


The coordinate system is based on the sim's location, where <0,0,0> is the origin of the simulator you're in. This number maxes out at 256 on X and Y; 4048 on the Z.

To find an absolute coordinate with regards to all sims, simply add llGetRegionCorner to the value of llGetPos.


As for things like a prim's "center," it's fixed close to the center of an unmodified version of that prim.

Clear as mud?

To avoid the obvious troll trap, I'll ignore the insulting "form letter" comment and just concentrate on the substance of the subject at hand.

First, the model in your link looks well constructed, and the foreground is nicely rendered. It's everything I'd expect to see for a work in progress from an experienced modeler using a good 3D package. I hope you'll continue to post your progress on it because it's off to a great start.

It pretty much goes without saying though that something like that is not appropriate for any real time application, let alone SL. What you've got there is the beginnings of a film-quality model, not something you'd put into a live video game and certainly not into a streaming environment like SL.

It is the streaming, by the way, that is the reason for SL's prim-only building palette. Every machine running SL can draw anything inworld almost instantly because every model has the exact same building blocks, and every client has fore-knowledge of how to create those building blocks (prims). Therefore the only information that needs to be streamed is where to put them, what size to make them, and how to rotate them, which is a very tiny amount of data. Were "good" models like yours to be in place, then every single instruction for how to create every single polygon would have to be streamed for every single model. Multiply that by the millions of models inworld, or even by the hundreds that are onscreen at any given time, and you can see where the problem is. It's just too much data to handle in any semblance of real time.

For that to work, it's not just SL that needs to improve. It's computers in general, and more vitally, internet speeds. The day we're able to fetch massive amounts of graphical data from remote servers as fast as we can from our own machines is the day models like that will be practical for an environment like SL. Until then, like it or not, this is the state of the art. Heck, we can't even get the viewers in professional 3D modeling applications to produce output-quality imagery in real time like what's in your rendered picture. There's just no way something like SL could do that at this point in history, as much as we all wish it could.

All that having been said, there's no reason you can't build an excellent approximation of that model in SL. Of course it won't have all the features of the one in the picture, but it will also have many features that the pictured one doesn't. For example, you can't exactly get on the one in the picture and ride it around town, but on its little SL brother, you certainly could. Also, and perhaps most importantly, who gets to experience that model in the picture as a 3D object besides you, its creator? Really, no one. Sure, we can see the nic rendering after the fact, but we can't interact with it. If you can find a way to get 80,000 people into Rhino all at once to view that model instantly, god bless you, but obviously the reality is you can't.

My point is that while, sure, in SL we're sacrificing 99% of all the great tools we're used to in our "real" modeling programs, what we're getting in return is a level of communal interactivity with our creations that those programs could never even begin to offer. Outside SL, I build film-quality models myself, and I'd love nothing better than to bring them into SL, but at the same time when I spend a hundred hours at a time building them in Maya it gets awfully lonely. The real question is is SL lacking in good modeling tools, or are good 3D apps lacking in communication tools? The answer, obviously, is both.

Anyway, as I eluded to in what you chose to call my "form letter" I have found over the past 2 years that I very much enjoy the challenge of figuring out how to best approximate my "real" models in SL, despite its limitations. I was not kidding when I said it forces you to expand your mind. It really does. The problem-solving intelligence you develop as you get better and better at building great looking models in SL is a skill that will only help you in all your other modeling endeavors.

I do agree with you about the interface though. It would be nice to have a 4-pane viewer. That would make a lot of headaches go away. However, since SL is a multi-funtion platform, and modeling is only one of those functions, it's unlikely we'll be seeing anything that specialized for a long, long time if ever. All we can do right now is make due with what we have, and embrace the challenge.

It may interest you to know, by the way, that Linden Lab has stated on more than one occasion that they do have every intention of allowing full mesh model import from programs like Maya and Max in the distant future. However, when they say distant, they mean DISTANT. The internet has to get several hundred times faster or else data management has to get several thousand times more efficient before that can happen. Like you, I look forward to that future with longing eyes, but in the mean time I'm not willing to wait. Building in SL is fun, socially rewarding, and even financially lucritive. All that more than makes up for the lack of "professional" tools in my opinion. I hope you'll reach the same conclusion and join in the fun.

Keep on practising!

all I can say, Cotton, is WOW!! awesome builds - are they yours?

Yes they are Ben! I'm also posting a small bar I did for Trinity Cole. This is in Weber, underneath her club if you'd like to see it. (My apologies Trin, if this generates more traffic for your club, LOL) Some parts aren't totally finished yet tho but its fun to look at.

The reason I'm posting these is because, partly, I believe we must always strive to push the limits of SL and not sit in the corner and wait for it to 'grow up'. Keep building your high prim objects everyone! If not the Lindens would think you don't need that much prim allocation and you'd be stuck with cardboard cutout santa clauses and christmas trees.

A clear example of this would be the gaming industry and how it directly pushes the PC hardware market to strive for faster and better graphics cards which are always getting cheaper. The sheer demand from the masses actually benefit the smaller niche markets indirectly too.

I remember the days when everyone was using 56k dial-up connections, and most would say, the copper wires will never ever be able to push any more data through it. This wasn't that very long ago. But now look! We're pushing SL through plain ol copper!

Beautiful work, Cottonteil. I'm impressed. Great stuff.

Here Marcello, I finished your bike today. LOL. I'm sure I could do better, but the linking limit got me.

Now THAT'S the reply I was looking for. I really wasn't trying to be a troll, or even to seem offensive. I simply saw the same reply on a similar thread.

Centainly, I didn't mean to imply that I wanted to build a model a detailed as the Lunacycle in SL. Just wanted to make sure it was clear that I knew something about 3d modelling.

The reasons for the prim-based modelling are completely clear to me. And I think it's fine. Great even. It obviously works very well. I think one of my big problems is in the interface difference between a CAD package and working in a window on the 'world' in SL. I'm used to rapdidly switching between front, side, top and perspective views. This lets you sort out the axis controls really easily and also to visually check the alignment of parts. All it would really take in SL to emulate this are some additional camera controls. Once you're focus is centered on the object you're dealing with, the orbital camera control works fine, but what if you could hit a button and snap to the side, top or front view?

I also understand that the experience is meant to be an interactive one, with other people able to see what's going on and communicate in real-time. You're definitely right that too long in Maya, 3DStudio, Rhino or even Photoshop forces you into radio-silence mode. Enough of that and you'll go off your nut. After my last post there the other day, I went into SL and was greeted by Noel Marlowe. It was great. He(she?) showed me some really cool stuff and a few more pieces of the 'building in SL' puzzle clicked into place. That's exactly the sort of thing you lose in a dedicated design app.

You hit on another thing you lose: being able to 'use' the things you're designing. Sure, you could design a house in 3DStudio and render some killer, photo-realistic walk-throughs. But in SL, you can LIVE in the house. That's cool. It'd be nice to have the prim limits bumped up a bit for that though.

Cottonteil, you rock =) The Lunacycle and the other models you posted look great. The only thing I'd modify would be the rake angle on the front. Should be around 25-30 degrees. Any chance that the Lunacycle be made ridable? I'd love to see it cruising around.

Oh, thanks for the angle. It was quite hard to judge looking at the single pic. I'll try and change it. Some other parts, like the fuel tank, I just simplified it. Hee hee.

Yes, it can be made rideable. But I simply don't have the scripting skills to get it to. I need it to automatically attach to an avatar when you sit on a 'proxy' vehicle, since its above the (arbitrarily imposed) prim limit for physics objects.

I'll send you a modifyable copy, since it's your design that I ripped off off.

I started writing this post with my "rain on their parade" cap on. Let's try this again.


In Second Life, I'm VERY impressed by the work some people manage to pull off (that stuff is really nice, Cottonteil). That said, I feel the brilliance is lost on the system you're building in.

I spent roughly four months toiling on a fully featured game with my teammate, Alan Beckett. While we won the "contest" the thing was for, the recent script changes throttled our work to the point I had to finally take it down. The thing simply refuses to work now, and no degree of workaround or fix will make it come back to life. I've tried.



Apply this to prim building. Suppose tomorrow they changed the way textures display on all objects. You would have no say in just about anything, and all of your creations could potentially get wasted.



The need for import/export holds water because it means your creations are yours, as opposed to being limited by what the Lindens decide to change, add, or break.

Second Life is in some ways a very shaky foundation. Make sure you keep a backup of your work somewhere if you plan to invest large sums of time on building exclusively in it.

I spent one day on Marcello's bike. I'm not actually concerned if I don't get to keep it, if suddenly SL goes kablooie! Because I actually learnt something from the one day exercise and earned two new friends! (I think)

Sorry to hear of your troubles, Jeffrey. That really sucks. I can imagine how disappointed and frustrated you must be.

I've been hit by things breaking due changes as well, although by the sound of it, not as severely as you. The change to the way cuts are calculated in 1.7, for example, is particularly annoying. Objects I had taken time meticulously to fit together seamlessly with exacting cuts are no longer exact, and therefore no longer seamless.

I don't know what to say about this kind of thing except that second life's not fair sometimes, just like real life. Sometimes bad things happen for no good reason. I hate it, but since there's nothing I can do about it, I try not to let it get to me too much. I know that's easier to say than to hear, so I wish I had something better to offer, but that's all I've got. All we can really do is try to accept these things and hope for a future where SL is more stable.

For what it's worth, I'm sure the Lindens are just as concerned about this as we are, even if it's hard for use to see it. They want SL to evlolve to become the next generation of the internet. That's never going to happen until it's a million times more dependable than it is now. They have no choice but to stablize SL's behavior if they ever hope to accomplish their goal. I try to remind myself of that whenever these problems crop up.