Feynt Mistral

Well, first, this forum would be for full blown meshes rather than sculpties, so you might not get the kind of answers you want here. I'll say that Render/Bake Sculptie Meshes doesn't work here as you need to use Blender's packaged File->Export->COLLADA 1.4 option to make your object into something you can upload to the mesh grid as a mesh object. This could also be why the Mixamo Collada 1.4.1 importer fails, because it's not meant for use with sculpties. But perhaps I'm just not understanding what you're asking in this case. To enter edit mode, you either use the appropriate dropdown menu (which reads Object mode for you) or press Tab in most cases.

Well I can only tell you that in Blender you can create a mesh with more than two textures, one of my creations (a character which I intended for a Torchlight mod) has one UV map for its body and one more for its hair, and this imports properly into SL. Now, a scene with separate meshes (completely disconnected) will import each separately and attempt to rescale everything accordingly and position them like they should be in the original scene, but as I mentioned this did fail early on, so it's possible it's failing again (related to mesh import issues specifically or just the scene reconstruction as a whole I don't know). It's also possible that the version of the mesh sim you're trying to work in has compatability issues with your work. I know I was trying to upload my stuff in several sims the other night and it kept coming in as plates approximately 0.01m tall and perfectly flat, but when uploaded in a final sim where someone else was working and they successfully uploaded, my creation worked too. So, try a dozen sims at least, or at least ensure you're not working in one of the experimental sims (listed... somewhere around here).

Which leads me to two questions: Are you sure your video drivers are up to date Are you setting your Macbook to use your GeForce card exclusively?

Well, only 4-8 weeks left and a long way to go still for stability and size reduction (see: new mesh format), so they have their work cut out for them. But I'm sure it's less working over time and simply shift work for constant production.

Or two other failsafe (for the sim) solutions: Auto-return the new objects which were created (disasterous to say the least for a builder's build) or make them temp objects so they'll be cleaned out after a short while unless they're relinked or taken into someone's inventory.

In a way I think this is intentional because the "prim cost" largely reflects the physical impact of the prim on the simulator's state (both world presence and physics impact, even when phantom). Setting to physics type "none" removes the selected objects from the physics simulation entirely where as "phantom" merely removes it from collision checks (as is my understanding from previous discussions). So having lots of prims that don't affect sim performance is okay. I imagine before long though a finger will be put in that hole so we don't get giant 400k poly houses over several meshes that cost only as much as the physics representation of the house as a whole.

Definitely doubtful that it would be made flexi, because determining where the root of the flexing should be is likely to produce odd results, but I think I can come up with a system that would work using greyscale maps.

Welcome to it. In the past (and emphasis that last part) we noticed Maya's export to COLLADA wasn't always "true" and would produce wacky results (if at all). There's supposedly a third party exporter which works well, but I know little about Maya myself, just what I recall hearing from others. Another thing is it seems you're trying to upload an entire scene of different objects rather than one single object? That might be the cause of your issue as well, as scene parsing by the importer was at times unstable. While I'm sure the Lindens do want to know about all consistent crash cases, it can be disheartening to know your uploads don't work for some reason. If you can take a particular COLLADA file and constantly crash the client (and make sure you're using one of the newest builds) you should probably post a JIRA entry with a copy of that file and the revision of the client you're using (most recent as of this post is 232004). The method of exporting might be helpful as well (which plugin did the exporting, what options did you have selected, etc.). This way the Lindens have something to muck around with the file and maybe find a flaw in the code, or tell you what you're doing wrong "because <option X> is incompatable with SL's importer."

I'd like to see procedurally generated textures myself. It's a lot more complex though than "hey, there's an engine, let's use that for our texturing!" Windlight for example took a long time to get implemented, and they even brought in members of the team that made it to work with the people who "knew" the SL client (could anyone really know the old SL client's code base though?). I feel an important step toward supporting procedural textures is the planned material system. Not only will it allow conventional texturing which we're all comfortable with, it lays a solid framework to put the aforementioned Substance engine into (if they choose to go with that). Not only that, procedural textures are less useful than the textures you would draw yourself. They're designed to fill an area with interesting stuff (like grass, tree bark, or craggy rock. Maybe a dungeon wall hewn from blocks of stone), but they don't get more specific than that. User defined textures can do that (with outside procedural textures, maybe) and more.

As far as I know Macbooks come with an Intel Graphics chip to keep down the heat when not doing graphically intense stuff. Look into adjusting the video chipset as mentioned earlier and I'm sure it'll start working. The GeForce 9600 should definitely be able to do shadows, mobile version or not.

How similar something is and whether it's possible to kick out someone with a similar design is for the lawyers to determine. I admit at this point, the questions far out strip my legal knowledge. Son of a legal secretary that I am, I don't know all that much about copyright enforcement.

Drongle McMahon wrote: Several people have expressed the fear that the introduction of mesh will alienate them from building because it will bring benefits only to those who have the skills to make meshes, and those using the existing tools will not be able to keep up. Their skills will be devalued. One of the answers I have given to allay that fear was that people would produce mesh components, that these would be cheap or free because of competition, so that nmesh would enhance, not devalue their building. But is this true? We are talking here about making copyright protection more effective for meshes. How far will this go? I am concerned that a result could be to constrain the provision of simple building components and thus compromise the benefit that would otherwise be gained by builders who can't make their own mesh. This is because I am unclear about how simple a mesh might be the object of protection. It's true, meshes can enhance their building. Just as sculpties do now, and in some places sculpties exclusively take hold (avatars for example). Meshes will see similar use, where some applications will see tremendous mesh use (again, avatars), but they will also enhance conventional builds employing a lot of prims simply because prims are easier to work with and faster to load. What good is a 60 room mansion if it takes 5 minutes to stream and costs L$3-6k to upload? This system doesn't just apply to meshes though, and I never intended that. I fully see this idea working for every object, or sub portion of an object (flagging a texture on an object because you know it's yours). This works to protect every creation on SL. Drongle McMahon wrote: For example, if A makes a set of columns, or bevelled building blocks, can A then stop B from offering a similar set? What about window frames, doors, fences, stairs? Can copyright apply to these? How similar do they have to be to be infringing? How different do they need to be to avoid competitor's takedown? I am afraid there will be a mad rush to monopolise all kinds of component to maintain a high price them by claiming copyright for what are essentially trivial designs. Will the first mesh brickmaker have a monopoly of mesh bricks, or the first mesh fence maker have a monopoly of mesh fences? What if I included the stairs in a single mesh house? Could the stairs mesh monopolist have them ripped out of my houses? Would this confirm the fears of the meshless builders? Is it a paranoid nightmare? Are we going to see a nasty DMCA war start the day after mesh release? How many of you have your components ready for the competitive upload binge? Copyright law has been overturned on creators when trying to enforce it on simple objects. The idea of a door frame for example is so common that you couldn't enforce a copyright even with every lawyer in the world on your payroll. Even the lowest judges in court would laugh you out. If there was some cause for that door frame to be unique though, like elaborate molding, it could then be enforced that someone with that unique look is infringing on your property. Likewise a brick could not be enforced unless it was special in some way (a set piece for a dungeon scene, all chipped and pitted), and even then it's possible it could be overturned. But there's no way someone making a mesh pillar or mesh cube could claim against a prim builder. In fact, copyright law allows for this. Even though two people can make nearly identical devices that perform the same task, if it can be proved that they came about because of entirely different methods of construction or from different sources of inspiration (the harder of the two) then both copyrights are valid and neither is in the wrong. It's part of the fair competition law. Hell, reverse engineering is perfectly legal as well, as long as you don't use any of the original subject in the creation of the derivative work.

Drongle McMahon wrote: Executive summary: A more explicit claim of right to upload might help a bit, but then again, it might not. I'm afraid arguments based on the claim that copying is not theft are irrelevant. Copyright law exists. The reason is that illegitimate copying harms the genuine creator, by loss of sales and/or reputation*. The fact that he is left with his original copy is immaterial.  I'm not disagreeing, to a very fine point. I'm not my friend, he would argue that copyright does great harm and should be abolished. But I do argue that the loss of sales is an illusion provided someone else doesn't copy the object and sell it to someone instead, gaining a profit off of another's work (plagiarism). The person wouldn't have bought it in the first place most likely, so the "loss of sales" doesn't exist for them. This is why I also laugh at the idea that software piracy hurts the various software industries. I won't buy Maya. I can't, it's thousands of dollars I don't have. But I'll pirate it to dink around in it for a while and get bored, then uninstall it. If someone is copying stuff on SL for their own personal use, it's annoying and if I discovered it I'd probably report them, but at the same time I almost don't care because they aren't profiting. As soon as they give away even one copy, I care, because they are affecting sales at that point. The analogy though that you made is in error and I stand by my assertion. The damaging or theft of property is completely different to the theft of an idea. At the end of the day, you still have your idea in your own form, untarnished by outside influence (unless it isn't created yet, in which case your potential thief is really good). And again, unless someone turns around and uses your idea for their own purposes without due credit (and permission) you likely would have lost nothing because they probably wouldn't have bought your item anyways. Drongle McMahon wrote: Looking for exact matches in comparing images for similarity would be naive indeed. The fact that it is possible to recast the same shape with different meshes is actually the basis that should allow algorithms to reduce meshes to a consistent minimal set of essential geometric determinants. I suspect it is hard, and certainly not feasible in this case unless a solution already exists. There are certainly some papers published on the subject, but I haven't read them and probably wouldn't follow them if I tried. So I will refrain from judging that issue. However, I see no reason at all to preclude automatic detection of exact copying simply because it will not detect all copying.  The point is it will work for entry level copiers only, and I'm talking about the people who know absolutely nothing except "if I use this I can copy someone's stuff!" The trouble is it'll only work for a short time before the community around those tools realizes that they only have to perform X change to make it dissimilar enough and the process is brought to its knees. To take sounds as an example, slowing it 3%, trimming the peaks and then normalizing can produce largely the same sound but it's completely different to look at in a wave graph. For 3D someone just needs to spend a minute removing or adding some edge loops, resizing the odd loop slightly, or removing parts of the mesh itself (like the inbetweens of toes, which aren't necessary. For textures, a light gaussian blur, rotation or paralellogram by 1 degree, and rescale by 2% will defeat most similar picture analyzers (it's a subject that incurs a great deal of cost to maintain and develop, look at facial recognition software). Drongle McMahon wrote: The automated community solution sounds to me like an institutionalised lynch mob. The idea that reputation (or conspiracy) should be capable of determining propriety is contrary to basic principles of justice. Reputation is a very poor guide. Have you forgotten Madhof? Then there may legal issues. Whose RL details will be provided to the complainant when the counter-DMCA is filed? Who would be liable for the charge of perjury when the automatic DMCA is found to rely on insubstantial grounds? I am not a lawyer, but I believe a valid DMCA can only be filed by (one who claims to be) the copyright holder or his representative. How can that be arranged with an automated system based on unsubstantiated gossip? The problem is that any such system is unaccountable and manipulable, while it's effects can have serious consequences, including legal consequences in RL.  The submission can differ in that it enters a queue for review. An actual person would then review the object (flagged by the community) and contact the alledged owner about the item and if it's found that it is a copy a DMCA claim can be filed and processed as normal. In the case of "the mob flagged the item, the creator wasn't even involved" then it can be beneficial because the mob potentially has the owner's best interests in mind and is alerting them to the (possibly elusive) forgery through the agent contacting about the flagged item. On the flip side, a group maliciously intending to flag an item with no grounds to fight the true creator (as in, the item flagged is indeed the true item, and the copier is trying to get their item banned) would be subject to investigation themselves, possibly forfeiting the item in question in their ownership or even facing banning for attempting libel. Drongle McMahon wrote: Unfortunately, both reporting and detection break down for things that come from outside sources. There are at least four cases of two "creators", both unlicensed, both licensed, first unlicensed, second unlicensed. I wrote a analysis of each, but it is too long. Suffice it to say that the community system is quite likely to make the wrong decision in any of the cases, and is always subject to nefarious manipulation. ...  This I entirely agree with. Outside copyright holders and licenses are difficult to deal with. But such a system could not be automated, and likely never would be, as external DMCA filings are where the legal pursuits come in. Internally, licensees would have to provide proof of their license and have it verified by the creator, but again that's where actual people rather than automation comes in. There will always be a human element to this process.

This sort of thing has been available before using multiple web cams if I recall. There's a project or two on Sourceforge about stereoscopic scanning if I remember the name correctly. If I knew more about OpenCV I might try it myself with a webcam. I know for a fact that there are people exploiting the Kinect to do this sort of thing, at least on a real time level grabbing the user's body shape and making a 3d shape out of it. It isn't exported, but there's no reason you couldn't take it a step further and make statues out of whoever you scan with the Kinect.

Depends on the hair. It will most certainly not be flexible in the beginning, but they may allow us to apply "flex maps" at some point which could be greyscale textures allowing avatar physics like motion (i.e. not affecting the sim, completely server side). Rendered hair and fur like what Maya or the new Blender hair/fur system will not be importable. Low poly hair like this however: http://www.cgarena.com/freestuff/tutorials/maya/lowpolyHair/index.html Will certainly be allowed (provided you spend the extra polys on the other side of the hair, since we don't get double sided polys for free right now. *HINT* *HINT* *nudges LL with a clue by four* )

The Macbook Pro has an automatic graphics switcher for when you start doing higher end graphics. Its primary adaptor is the Intel Graphics chipset (for displaying desktops, browsing the interwebs, and doing emails). Until you start doing something really graphically intensive it won't automatically switch over to the Radeon chipset. If you follow my posts, you'll know that I am a very strong opponent of the Intel graphics chipset because it doesn't support most shader instructions (like shadows) and will crash aaaaaaaaaaaalllllll the time. Luckily there is a way to manually force the Macbook to use the Radeon chipset all the time, and that will likely resolve your issue, however I haven't the foggiest on how you'd make that change (since I don't own a Mac at all).

Ok, so being a programmer, modeler, and uber leet haxor (on the grounds that I know how to write a program now that can read vertex data from video memory) I think I can answer this exhaustively for everyone concerned. First things first, it is impossible in the current system to prevent someone from obtaining data about textures, sound, and mesh. Any work to make it more obscure to obtain this data through hard drive scans is pointless. It's akin to putting a 4 foot thick bank vault door in a simple 1/2" thick plywood wall. There's no special tools that aren't readily available necessary, even a bit of effort on your own part will get you through to the treasures you seek (in the analogy, taking a running start and shouldering through. In reality, my path of learning to read the video memory is a good example). The analogy of locking your car doors is also a poor example in this situation, as the end result is a devaluing of your property and a removal of that property (damage, plus probable car theft). You are losing the car. IP theft will always leave you with what you made, someone else just gets a copy of it for little or no cost. The correct analogy would be if you were on an magic island with two other people and infinite resources, and you used those resources to make something to sell to the other two. Imagine, then, when you've made this wonderous item, one of the people sees your creation and uses the magic of the island to conjure up an exact duplicate. You still have your original, he just has an exact copy. And now he sells his to the other guy for a lower price than you were offering him just a few moments ago. The magic in this case is hardware level inspection of the data in use. We also can't do a shape comparison because, as mentioned, 3 seconds (or more, if you're more involved about the process) once you have someone's data can change its form entirely in the appropriate software. In the GIMP you rescale, recolour, and/or rotate ever so slightly and you've made a brand new texture that doesn't match the original by any automated test. Likewise in most modeling software packages there are automated retopology functions which preserve the shape but redo the lines to be neater, or can make shapes more complex (multi-res or subsurf mods) in a moment. The weighting portion of a mesh is harder to copy, and that you can't do from video memory, but a dedicated enough copier could repaint it all in a couple of days (or hours if he knows what he's doing) and be selling it himself. This means that automated methods of preventing copying, or further encryption is pointless. The counter argument that "locks on doors are easily defeated, we should stop making better locks" is moot because they cover different things entirely (theft of property versus theft of an idea). Instead of focusing on prevention, there needs to be a system in place to report and enforce in a quick and effective manner. As previously said, the community is the best tool for finding and pointing out copied materials. This item needs to be easily flagged by the community on a global basis (the base item, rather than an individual item being worn by someone). With enough flags the DMCA process should be filed automatically and actions taken. Likewise a community certification process would allow people to register in the database that not only did they upload this at such a time, but the community (friends and fans) recognized the file is authentic to the creator. This is essentially a digg process for items on SL, you rate up or down items and at certain thresholds they get reported or certified, only they don't detract from each other. You get X reports against the item and you get a DMCA claim filed based on the flags stating the original creator's name. Once certified, the report process will take longer to occur. After a certain time period (like, a half a year to a year), no further ratings can be done and DMCA processes would have to be done the current way. There shouldn't be a notice to the public about what ratings the items have either. Now to those of you who will continue to contest that an automated solution to prevent copying can be obtained, yes, there are two solutions to automate the process: The first will negate everything but screen capture copying. This is the oft rumoured server side rendering. We would never get actual object data, we would only ever get a picture of it. This stops almost all copybot activity and mesh appropriation from the memory buffer simply because there's nothing there to scan. This is ridiculously expensive for LL to maintain with over 30k people online concurrently, not only on a render farm level but a bandwidth level as well (rather than a burst of object, animation, sound, and texture data there's a constant streaming cost). Onlive, if any of you have used it, is not an example solution to this situation. It's a good idea, but it's poorly implemented. Second Life streamed in this manner though will be subject to lower frame rates (for some, drastic improvements for others) and a delay in responsiveness relative to the rendering time and network speeds. This is a step backward however, as it will make anything requiring quick interaction an impossibility until we get standard gigabit connections throughout the world (or quantum networking where stuff is there the moment it's created, in theory). The second is hardware level DRM. Microsoft already tried this with Vista and its hardware enforced DRM hardware/software bit combo. But that's been defeated now, such that you can be using any version of windows and still see and hear things at maximum clarity even though you downloaded it from some guy you know across the river who got a copy that fell off a truck. Not only is it ineffective (long term) it's ridiculous to think that hardware manufacturers would listen to Linden Lab about this change and implement it in their hardware. Likewise, everyone who uses SL would have to get this hardware or the copy protection would be useless. As a strong proponent against DRM (my media, my methods *gives RIAA/MPAA the finger*) I would reject such hardware outright, and would leave SL if it came down to a decision about using such hardware or not being able to log in anymore. I'm certain a good many people would share this sentiment.

Unless you does want, like a house build. Or a car with turnable wheels and openable doors.

At present it seems as though the default Blender 2.5x COLLADA exporter rotates all bones in place 90 degrees. Obviously this is not ideal, though I'm sure it seemed like a good idea to someone (maybe someone who mistakenly rigged an avatar with their bones sideways, somehow). The Google Summer of Code has someone working on the COLLADA exporter (amongst other things, like weight painting) so those of us without a double major in spacial mathematics and image processing will be able to export our models again the right way around.

While it seems like common knowledge to me, a part of all rigging is testing your avatar in wacky poses. Make them do the splits, the wave, touch their toes, etc. Multiple poses is best (saved animation frames in different poses) and then just cycle through each pose to determine if there are any loose vertices (or any weighting that could be fixed up). If everything looks fine at every LoD level with each pose, you're looking at some kind of mutant vertex or a bug that should probably be reported.

Oh something else to mention, Blender also uses _L and _R postfixes in the same way as .L and .R. And I've recently learned case doesn't matter, but most people prefer capitals anyways. I'm also wondering if this joint info will make it into a nightly build soon, I'd like to upload my avatar after making some adjustments and I've recently done this renaming. >)

Gaia Clary wrote: From all formats that could be choosen by LL, in my understanding Collada is the best choice they ever could do. Tell me any other format, which is open and as widely supported as Collada and allows the same level of complexity Heh, without that distinction of "open" I could think of a couple formats. In all honesty though the .blend file is an appropriate choice. It has massive support (an entire open source community, plus a thriving business centered around maintaining Blender), is openly available, and supports quick loading at runtime (see Blender Games). COLLADA is something exportable by most modeling programs though, while Blender's the only one that deigns to use the .blend format (lest the Autodesk master race dirty its genes with open source). Anyhow, to answer the initial questions is somewhat hard to explain unless you already know a thing or two about modeling, but... Have you ever been working on a build or piecing together an avatar and thought one of these things to yourself: "hmm... that doesn't quite line up right." "if only I could put a hole here and here on the same side of this cube I could make a door and window" "Wow, I spent 20 prims making a shoe... why?" "Why is my hair made of 200 prims?" Mesh solves all of these issues. It allows you to make one seamless object of a complexity that simple prims may never allow. It also allows you to decide how your textures are applied. If you want, you can make only a small portion of a larger texture fill a vast portion of your object in any shape you want (and if you look into UV unwrapping on Youtube, you'll see some pretty interesting shapes to be sure). For some truly prodigious examples of meshes, look no further than your favourite 3D enhanced games (Assassin's Creed, Knights of the Old Republic, Crysis, Gears of War, etc.) or online sales sites like Turbo Squid. Now, why do you need it? You don't honestly. If you're not a hard core builder, or even if you are but you can bend prims to your will and never see a limitation in them, it's not something you really have to get into. Building will always evolve and change, but the use of prims will always be a solid and reliable way to prototype quickly. But if you've ever wanted a useful skill to bring to the employment side of life ("Hey, I know how to model in 3D, check out these characters/buildings/objects of interest in my portfolio!"), or just want to improve your building quality "to the next level" then this is the way. There are a vast array of 3d modeling tools to choose from (Blender being one of the best free ones and featuring perhaps the most complete collection of tutorials freely available online) and (almost) all of them are as capable as the others when it comes to what can be done in SL at the moment. There's also a thriving Blender community on SL, though the majority of them are currently focused on sculpties (a trend I see falling off fairly soon). It should also be pointed out that making sculpties is actually a great deal harder than using the entirety of mesh making knowledge due to its strict limitations. Just look up some tutorials on making trees in Blender and you'll have all you need to make a forest in half an hour.

Can I suggest another group for the Blenderites? mJointMap["mPelvis"] = "mPelvis"; mJointMap["mTorso"] = "mTorso"; mJointMap["mChest"] = "mChest"; mJointMap["mNeck"] = "mNeck"; mJointMap["mHead"] = "mHead"; mJointMap["mSkull"] = "mSkull"; mJointMap["mEye.R"] = "mEyeRight"; mJointMap["mEye.L"] = "mEyeLeft"; mJointMap["mCollar.L"] = "mCollarLeft"; mJointMap["mShoulder.L"] = "mShoulderLeft"; mJointMap["mElbow.L"] = "mElbowLeft"; mJointMap["mWrist.L"] = "mWristLeft"; mJointMap["mCollar.R"] = "mCollarRight"; mJointMap["mShoulder.R"] = "mShoulderRight"; mJointMap["mElbow.R"] = "mElbowRight"; mJointMap["mWrist.R"] = "mWristRight"; mJointMap["mHip.R"] = "mHipRight"; mJointMap["mKnee.R"] = "mKneeRight"; mJointMap["mAnkle.R"] = "mAnkleRight"; mJointMap["mFoot.R"] = "mFootRight"; mJointMap["mToe.R"] = "mToeRight"; mJointMap["mHip.L"] = "mHipLeft"; mJointMap["mKnee.L"] = "mKneeLeft"; mJointMap["mAnkle.L"] = "mAnkleLeft"; mJointMap["mFoot.L"] = "mFootLeft"; mJointMap["mToe.L"] = "mToeLeft";  The .L and .R endings are automatic endings when making mirrored skeletons in Blender, and also are used to mirror weights from one side of a character to another.