Alisha Matova

I'm certainly not Falcon, but feel compelled to answer. Once the tools mentioned in this thread go live, we will be able to set linksets with magical pathfinding properties and bake them into the navmesh. There may be some limits I haven't found yet, but I currently have a test running on a (walkable)platform above 1000 meters. http://wiki.secondlife.com/wiki/Linden_Lab_Official:Alternate_Viewers#Pathfinding <---- looks like the slipped the tools into this beta and I think they are ready to sign up estates for Agni testing too!

I'm biased too! *waves! After 6 years using blender, the Max UI baffles me. Lol But yes, you're tottaly correct in that Blender has steep learning curve. This is because Blender can do sooo much stuff(animations, textures, modeling, heck whole movies and games). It doesn't have to be hard to learn, though. The trick is ignoring entire sections of the UI as ya learn the parts you do need. Only coming back to those sections once comfy with the basics. With a little hand holding and perhaps some screen sharing, I can get an attentive blender learnerer up and running in a couple hours. From setting up blender, to ready to export to sl. (this used to take 5 hours or more! Thank you Gaia and Jazz2 for taking the frustrating Python install hours out of the lesson!!) One of the concepts I push the hardest are Hotkeys. Once you memorize the common Hotkeys(there are even blender hotkey stickers you can stick on your keyboard) Blenders UI doesn't really matter. This is true for any software, really. No one can be productive mousing thru menus and dropdowns, no mater how shiny the UI. (Not intended to derail and cause some kind of my modeler is better thread explosion! I'm just doing my morning ramblieness.)

As you have experience with Max and maya, if you can justify the expense, I'd go with either or. I think you'll need some kind of FBX converter to export to SL correctly. (not an autodesk user myself) On the other hand, blender is free and has a gianormous web community of helpful people and tutorials. I believe the latest blender UI even has a setting to make it feel more Autodesk like, although I haven't tryed it yet. It's not so much which software, it's more about felling comfy using it. Once comfy, production of niftiness can begin.

I like to think about an object's shadows two ways. Shading, and Shadows. Shading would be more like an Ambient Occlusion bake. Where the modeling program shades the texture appropriately. Darker shaded areas between parts that are close together, and less shading in areas that are not interacting with other parts. Ambient Occlusion, or shading, does not (afaik) take any scene lighting into account. this is great for highlighting, and shading, detailie bits of geometry. Shadows, (again I'm making this up as I go) to me, means you are fully lighting the scene in your modeler and baking those effects onto your texture. This can look incredible, but as others have mentioned, it's easy to make something so awesome its lighting(and shadows) can clash once placed in an sl scene. I've found that subtlety is the key. Light your scene, but shoot for a very mild effect, and try to avoid directionality. Meaning, try to light the scene like its midday and the sun is high, not sunset where the sun is casting a Looong shadow. A good trick for this, use multiple light sources, but only set one lamp up to cast a shadow. =)

Like any mall in SL, things are going to load slow. Just due to sheer volume. Add a gaggle of avatars, things really slow down. It's not any one things fault really, it's the sum of all things. I would suggest, though, that if you're planning a big splash screen event, ya build the place sooo efficiently that it works well for the lowest common denominator. Ie: if it doesn't work well on yer old laptop, chances are, it won't for half your visitors. This does take a Lot of planning and optimizing. And realy isnt all that fun, but it's fully nessesary. Of course events like this with multiple vendors compound the problem and would take Herculean efforts to optomise, but it's still very possible. Once the sims overall load is in check, then it's time to think about layout. Landing people in the middle of everything is Not a good idea. As it forces everyone to load everything as they land. Landing people in a corner of the sim, with some walls blocking(occluding) their view, will help some too. Even if you get all that right, SL is probably gunna choke somehow. So someone needs to be there 24/7 to restart thing when they go turtle. One thing that would realy help. But I've only seen this done with big LL involved projects(CSI), would be to have copies of popular splash screened event sims sitting there ready to handle the overflow when one gets slow. I'm sure that service would be too spendy to apply to this particular event, though. /pre coffee ramble

Thanks! I hope they announce these tools when they go live. But I'm all sorts of excited about this and just couldn't wait!

Hi Chosen! I guess I would define pathfinding as AI that can navigate. It will help us bring life to SL! There are various script functions involved. llNavigateTo, llWanderWithin, llPursue, llFleeFrom(a very fun one where the critter tries to hide from you) and many more. And event updates like, near, stopping, goal unreachable, etc. Which can be used to trigger new states/behaviors. All of this comes right from Havok's AI code. For instance. You could infest your shack with pathfinding beatles. They could wander within the area, but when disturbed, race under the fridge. Or even better! Chase you around until you spray them with bug killer. Or a simple monster who protects a cave, jumping out and chasing anyone who dares enter. Or even something an AV can ride. Essentially, pathfinding will let us add behavioral personalities to anything we want too. From pets, to people movers, to monsters, Indian Jones like boulders, to stuff I can't even start to imagine.

I really don't know where to post this, or when these tools will go live. I'd guess they will work their way into a beta viewer in a couple weeks. They have put it on the wiki, so I feel safe talking about it. Here is a link to the wiki: http://wiki.secondlife.com/wiki/Pathfinding_Tools_in_the_Second_Life_Viewer There is a Lot going on on that page. We will have a few new floaters, a linksets tool, a character tool, and a navmesh tool. The later lets you view the navmesh, exclusion volumes, test paths, etc. The first two work like Top Scripts allowing debugging and optimizing of the navmesh and the critters. I think the hardest part for us to wrap our heads around turned out to be the linkset tools options. Firstly, this tool is Awesome. It lists all the linksets on the sim or parcel, allowing us to teleport right to it, search the list, filter the list, etc. it will be very useful for non pathfinding tricks too! Here is where it gets tricky. Each linkset now has a few settings as it relates to pathfinding and generating the navmesh(navmesh is a sim stored asset that the critters look at when they are first created). These are the linkset options: Walkable Static Obstacle Dynamic Obstacle Material Volume Exclusion Volume Dynamic Phantom These are explained in depth in the wiki, but I'll try to break them down. The top three are Non phantom, the bottom three are Phantom. Any option with Dynamic in its name is Not baked into the navmesh, and is an Active object. Meaning it can move within the scene and critters will react to it, no mater where it moves to. Example: a door, a gate, or a vehicle. Dynamic objects are the most resource intensive, as the critters must actively track them. Use as few dynamic objects as feasible. Walkable and Static Object will be the most common settings. Walkable adds to the navmesh, static object blows a hole in the navmesh, where critters will never go. Simple enough! Exclusion volume is phantom. AVs can walk thru it, critters can not. Useful to stop critters going places you don't want them too. Material volume is the most fun. It's phantom to critters and AVs, but uses the a b c d material setting. To keep this post shorter, I'll avoid details, but, for instance, an A type critter could be slowed down inside the material volume, while an AV (or B type critter) runs thru without slowing. Common game usage for this could be a race track, where the track surface is faster than the grass around it. Finally, dynamic phantom is phantom to all, AVs and critters. A good setting for small plants. Stuff we used to use legacy phantom for. (great for volume detect) I may be making this sound confusing. But a little experimenting with the linkset tools will quickly get all up to speed. One think that is quickly going to become important is going to be reducing the physics geometry that gets baked into the navmesh. This is easily accomplished with the newer 'none' and 'convex hull' settings in the features tab of Edit. Less physics geometry equals a simpler navmesh, and happier critters. I hope this helps, and does not add any confusion. And it might explain why I've been bouncing around these forums commenting on physics shape related posts. I've got to say though, I've never had as much fun coding things as I have with these critter behaviors. It is really fun and very rewarding to see your 'thing' be the creature! =D

Thanks! I'd never really thought about why the render goes bonkers in those situations. I guess i chalked it up to closeness of the faces. Wonky normals does make a Lot of sence. Thanks for explaining it!

UV smart projections(if i am remembering the button name correctly) is darn handy. Fiddle with its options to get the best results. It rarely produces a map that is easy to hand draw on, though. So some editing of the map will be needed. It's best used together with a second uv projection map. Making Blender generate the final texture, instead of doing it manually an image editor. Machinamatrix.com has a few tutes that will help. Another, longer, option is to unwrap each part manually, using project from view. Simple select the faces you want to unwrap, squarely point the screen at them, click unwrap from view. This can be tedious on complicated models, until you get the hang of it. This is how I unwrap my projection uv maps. My workflow. Select whole model or material. Smart unwrap and edit/clean up its output Make a new projection uv map Unwrap it so the area wears the texture correctly Set up the material and Uvs to bake to the first smart uv unwrap(kinda tricky, see Gaia's tutes) Then I do a full render, which creates the final texture Of course I skipped a lot of setup steps, lighting the scene, setting material properties, render settings, etc. but I think that workflow best abuses the smart projection feature, and it's wickedly Fast to do.

If you're really sneaky, you can add just one more vert. Make a plane> Select one vert > Shift D and move the vert down > select all the verts, except the one you first copied > Fill should auto fill in the faces. It might even get the normals correct, if you're really lucky. I think the poly count winds up being the same, but it does save 3 verts. Although.. If you plan on baking textures to your model in blender, something about the super pointyness of the closed end drives the renderer bonkers. But, so will baking textures to two interpolated planes with flipped normals. Pick yer poison. =P (keep planes with flipped normals a reasonable distance apart while baking tex to avoid bonkerness)

I'm not on 2.6 yet, so I can't get specific either. But to answer this question, both. If you wanted all 8 materials(faces) to all be on one big texture, you want to unwrap All the poly on to one uv map. If you want 8 seperate textures on your inworld model, make a uv map per material. Blender let's us have the best of both worlds. But this is where it gets confusing. So I'm going to paste a link to a Machinamatrix tute that shows how to use projection uv maps. I have a guess that this is where you are stuck too. Uv maps in blender have a few choices. Use them to render, or output. I figure you have one set wrong. Blind guess though. http://blog.machinimatrix.org/3d-creation/video-tutorials/sculpties-older-tutorials-page/texturingsculpties_multiple_images/ The tute is about sculpties, where we had no control over the final UV. But it uses the same tricks, and explains the UV settings I think you are having an issue with. Hope that helps!

Sorry Drongle, I don't mean to confuse. I'm stuck not being able to talk about new features until LL finalizes them. I can say that flexibility of physics stuff inworld is going to be important. It will help define where pathfinding critters can and can't go. So in the case of a building with one big physics shape, it will either be go or no go. Unless you add physics parts ontop of it.

Physics shape none and phantom are Not the same thing at all. Phantom still sends collisions back to the sim. Volume detect. None is completely invisible to Havok. Anyways you're completely missing my point. Have fun!

Lol, I fully understand LI/PE as I watched it evolve in the Beta. Don't worry about me running out of prims. The stage is still a good example. It's exactly how people will be building inworld with mesh builder kit. In its case, the planks physics shape should be turned off, set to 'none' and One prim should be placed over the whole stage.. There is no reason to walk across every plank(or for pathfinding critters to have to think about 50 planks when One prim will work...50 times better. The same works for a multi mesh house, windows, trim etc. can be 'none', walls and floor can be 'convex' or 'prim'. And with incoming Parhfinding it's going to be very important to both have simple physics shapes AND be able to adjust them inworld. Like I said before, when they nail down all the vocabulary and tools I'll start talking more about optimizing whole sim physics for Pathfinding. I think I can say that locking your physics shape to your build will limit things, or at the very least force you to add pathifinding prims in top of it. Flexibility inworld is going to be the key to awesomely pathfindable sims and builds.

Ah! Ok gotcha! To be honest I don't pay much attention to LI costs. For me it's all about being a fast and efficient worker bee. Far too much work and too little time. So anything that gets in my way, im forced to workaround to meet deadlines. I get what you mean about small meshies, though.

The plank stage example was to point out how using one prim for physics would be more efficient(Physically) than leaving all the mesh planks set to 'prim' or 'convex hull'. Of course uploading the whole thing as one mesh would be lower LI. =)

Ok. Well the first benifit would have been him ignoring some old firestorm bug. My suggestion to ignore the issue and keep being productive by removing his blocker was valid. I do agree with a couple of your disadvantages. A merchant can't really sell something with a linked physics shape. It be too confusing. And the scale axis stuff. But the others I don't really understand. A small object,unless really interactively complex, should just have a simple cube physic shape anyway. Try linking a small mesh to a prim cube. Set the mesh to 'none' and the prim to 'prim'. It's the same LI. (not that uploading with a cube physics shape is hard). The linkset mesh got me thinking! Do you mean you upload multiple mesh linkset with a unique physics shape? One that cover the whole linkset? That's an interesting idea! I assume though, that you mean linked meshes each with its own physics shape. Id actually suggest that one big,and simple, physics shape linked to the whole linkset would be way better that 50 seperate ones. I agree though, a multi mesh multi physics shape link object would be nitemare silly. Let's say I build a large stage out of mesh planks inworld(something that is already happening with all the new mesh builder kits). I could leave each plank set to 'convex hull', or I could link the whole stage to a single prim that approximates the shape. Set planks to 'none' leaving the prim set to 'prim'. Walking on One prim is much better than walking across 50. Sure I could build whole thing in blender and upload with same simple physics shape. That would be more efficient. But it's not the only option, sometimes isn't possible(somethings just need to be built in place), and you can't make changes on the fly inworld for the customer. Linking physics bits in is limited. But has its place. And is especially useful when optimizing the physics for a whole build or an entire sim. There are a couple other advantages I don't think I can mention yet. When they open up about how pathfinding objects navigate, I'll chime in again. But, objects physics shape flexibility inworld is going to be important, in some cases.

There is nothing wrong with linkIng a simpler physics mesh to a complicated visual mesh. It's no different than linking a prim to a mesh. And a good physics shape will be a fraction of a LI anyway and not add to the overall LI. The only 'loss' would be the object now acting as a linkset and not being able to scale on one axis. I'm with Drongle, what viewer are you on? As I can upload Blender created multi material meshes with no material physics shapes on latest beta too.

I think, only full permissions textures will show up in the texture window. You can still drag them directly from your inventory onto the prim. Iirc the sim shouts something at you though.

This isn't the answer you're after, as I'm not a maya user. But to keep things moving, I would upload the physics shape seperately and link it to your visual geometry. Setting it to physics shape 'prim' and the visual stuff to 'none'.

Gosh I love that one. And the story behind it is just too beautiful! I'm defiantly with you though, I wish LL would do some artistic mainstream advertizing. Anything related to sharing dreams will make me happy.

It may just be the song, but I get the 'this should be an sl commercial' vibe from this too. No idea if that link will work. Search YouTube for 'pure imagination commercial' if it doesn't. (no embedding options when iPhone postin.)

thank you All very much! I managed to get things working with Talia's example =) tyty! the "(key)message" part did the trick =D

I'm stuck! On something rather simple. I have a simple volume detect prim messaging the avatars uuid back to my main script. I need to take that uuid and insert it into one of the states functions. Could someone please post a quick example that parses a uuid from a message into a function for me? Thanks in advance!