Chat log from the meeting on 2017-07-18

From OpenSimulator

Jump to: navigation, search

[11:03] Andrew.Hellershanks hm... Small gathering today
[11:04] Sheera Khan: we can't always be such a great crowd ...
[11:13] Andrew.Hellershanks I finally got back to working on a personal project over that past few days. I haven't been doing much with OS recently.
[11:13] Andrew.Hellershanks Some of the recent changes made to the code by Ubit address some issues regarding keyframe motion.
[11:16] Andrew.Hellershanks Some other changes are related to object rezzing via scripts.
[11:16] Sheera Khan: hihihi, larger but quiet crowd ^^
[11:17] Andrew.Hellershanks Sheera, it is slowly growing. :)
[11:17] Sheera Khan: is there any news about getting an official release maybe?
[11:18] Andrew.Hellershanks I haven't heard anything more about that since last week.
[11:19] Sheera Khan: It may not mean much to the Devs.. but it certainly matters to grid operators ..
[11:19] Leighton.Marjoram Would be good to have an 0.9 release and I agree with Sheera
[11:20] Andrew.Hellershanks It does to some. It is really just a tag of the code. The code as of a given commit is still the same whether tagged as release or not.
[11:20] Andrew.Hellershanks One of the big difference to grid owners will be in the physics engines.
[11:21] Sheera Khan: the physics engine will matter to region owners most
[11:22] Sheera Khan: but Bullet being the default a lot of people won't even notice ^^
[11:22] Andrew.Hellershanks Yes. It can affect a lot of scripted objects being used in their grid.
[11:22] Selby.Evans Release is Dev certification that the code is ready for production use. Commercial grids are reluctant to push code onto customers with out that certification.
[11:22] Sheera Khan: not only commercial ones...
[11:22] Bill.Blight Yes that is the other issue is changes to scripting, most are minor, but there will be a difference ...
[11:22] Andrew.Hellershanks True, they may not. Even BulletSim has had some changes. I don't know how they might affect objects that move.
[11:23] Sheera Khan: MetroPolis is an open grid like OSG... and we are hesitating too
[11:23] Bill.Blight I'm tired of hearing about the 'physics" and "Mesh" issues, you can still use bullet, no one is twisting arms ... If you want the new features, prepare to take some lumps ..
[11:24] Andrew.Hellershanks Bill, based on some recent IRC discussion, some of the mesh issues were caused by how people have been (or did) upload mesh to the grid.
[11:25] Bill.Blight Yeah I know .. I was there
[11:25] Bill.Blight LOL
[11:25] Bill.Blight for the one in dev anyway
[11:25] Andrew.Hellershanks I know you were. I was commenting on it for the others who weren't.
[11:25] Bill.Blight HEHE
[11:26] Bill.Blight the mesh "issues" are entirely based on how the mesh was originally uploaded ... Opensim is pretty much the only 3d environment that has let people upload shoddy mesh ...
[11:26] Selby.Evans I hope to publish an article about that mesh problem when Mike lorry gets it to me
[11:27] Andrew.Hellershanks Selby, Ok, good. That should help a lot of people know how to properly deal with mesh objects.
[11:28] Selby.Evans I want to get the word out on what the problem is and is not -- and how to fix it.
[11:28] James.atLLOUD Last week I recall that someone (Diva?) needs to do something to proceed to release.
[11:28] Andrew.Hellershanks nods
[11:28] Bill.Blight The problem is pseudo mesh experts that don't even really know how to model for 3d environments being leaned on as the "go to" people ... A lot of "creators" only became creators in OS because they could just upload any old crap and call it a creation.
[11:29] Selby.Evans I prefer to describe the a problem that can be fixed
[11:30] Andrew.Hellershanks Most of the modelling I do is for the creation of real world objects. It means I have a different criteria about what is a good model. Converting some of the things I can make to something that can be uploaded to a virtual world becomes a minor problem.
[11:30] Bill.Blight I prefer to not call a fix a problem ... and this Mesh "problem" is actually a fix not a problem, the problem existed before, and is now fixed ..
[11:31] Andrew.Hellershanks I don't deal in triangle counts and faces until the model is pulled in to Blender for conversion to Collada format.
[11:31] Bill.Blight This is part of the issue, is the mesh fixes in .9x are being called an problem or an issue, when in reality it is a fix of a long standing actual problem ..
[11:33] Andrew.Hellershanks hehe... you can fix the mesh problem easily by not allowing mesh. No mesh, no problem.  ;)
[11:33] Leighton.Marjoram thats very true Andrew the simpler of two paths. :)
[11:33] Andrew.Hellershanks Having good documentation on how to create and/or prepare models for upload and how to upload them will be good. It will mean less bad mesh in the future.
[11:34] Andrew.Hellershanks Dealing with existing mesh is another part of the mesh issue/problem.
[11:34] Bill.Blight Pretty much you can follow the standards that SL uses, other than step 2 in the uploader, and these standards are not much different from any Low-Poly 3d game environment ..
[11:35] Andrew.Hellershanks Physics or no physics when uploading mesh is one question, and how physics should be set is another.
[11:35] Selby.Evans Of course, Bill, but people who get unexpected things with existing mesh will see that as a problem. And I will speak their language when I try to communicate with them.
[11:36] Andrew.Hellershanks I've seen reference to "step 2" and I am not sure to what that refers.
[11:36] Andrew.Hellershanks Is it step 2 in the SL docs about how to upload mesh or a step called step 2 in the viewer?
[11:37] Bill.Blight Selby, yes true, but re-enforcing that belief that it is a problem and not an actual fix, just leads people to want to have it still "fixed" back to the broken way.
[11:37] Bill.Blight yes the step 2 on the physics upload tab
[11:37] Bill.Blight we don't use step 2
[11:37] Bill.Blight that is made for the Havok physics engine
[11:37] Andrew.Hellershanks Bill, ok. i'll have to start the process of uploading a mesh to see what is called step 2 in the viewer.
[11:37] Selby.Evans Bill -- when I publish, you can comment telling me how wrong I am.
[11:38] Bill.Blight will do
[11:38] Bill.Blight ;)
[11:38] Sheera Khan: but Step 3: Simplify would be ok?
[11:38] Bill.Blight umm I don't use anything past step 1 , don't think step 3 works without step 2 actually
[11:39] Leighton.Marjoram I dont think Bill is calling you wrong at all its the framing that is at issue a "problem" versus "a fix" those frames lead to different results in peoples thinking.
[11:39] Andrew.Hellershanks I just started the process. Step 2 is the "Analyze" step under the Physics section.
[11:39] Bill.Blight yes
[11:39] Bill.Blight Correct Leighton ...
[11:39] Andrew.Hellershanks ok. Don't analyze models.
[11:40] Bill.Blight People want to call a fix a problem, and then complain that it is broken, because it is different .. It is irresponsible for ANYBODY to to re-enforce the propagation of a bug or incorrect method of doing a particular thing .. But if that is what people want to do, I guess we can agree to disagree.
[11:41] Andrew.Hellershanks In any document about uploading mesh it should point out that the "step 2" people should skip is on the physics tab in the viewer. Some people may not realize that.
[11:41] Bill.Blight Not sure why this is a new thing Andrew, avoiding step 2 has ALWAYS been a thing in opensim
[11:42] Andrew.Hellershanks I don't know either. Even with good docs some people won't read them and will just do what someone else says they should do.
[11:42] Sheera Khan: It sure would help if there would be a definitive guide on how to correctly upload mesh-objects to Opensimulator-based worlds on the OpenSimulator-wiki
[11:43] Andrew.Hellershanks If it "works" they won't know they aren't doing something wrong.
[11:43] Leighton.Marjoram totally agree @Sheera
[11:43] Andrew.Hellershanks Sheera, there is a wiki page about mesh. It needs to be updated and could have the other information added to it.
[11:44] James.atLLOUD so, does the code need to prevent less than ideal behaviors?
[11:44] Sheera Khan: if it is on any place it would lack the official approval of the devs and be just one tutorial among a lot of ohers - and wrong ones^^
[11:44] Andrew.Hellershanks True.
[11:44] Sheera Khan: I even would write it and publish it there (I have write access) if I would be expert enough to do so
[11:45] Bill.Blight the new code does prevent oversized meshes , and missing physics blocks a lot more actively now, but the problem is the physics part is set by the viewer .. not the physics engine
[11:45] James.atLLOUD Same for me Sheera
[11:45] Bill.Blight if you upload without physics you don't have any
[11:45] Selby.Evans Sheera -- If you find any such documentation, please let me know.
[11:45] Leighton.Marjoram that would be good Sheera and then post links to that information for the wider Opensim community to find it.
[11:46] Sheera Khan: And I tried... but failed miserably on a bridge i tried to model...
[11:47] Andrew.Hellershanks I think I previously pointed out the page that is titled Collada Mesh at
[11:47] Bill.Blight If you try to upload mesh without physics and it is totally missing physics it will deny the upload .. If you set the physics too low you end up not being able to walk in or on mesh ... Been this way in other 3d environments for ages ...
[11:48] Bill.Blight this just needs to be defined as Steps 2 and 3 ..
[11:48] Bill.Blight "You will need to ignore any Havok physics engine specific parts as this is not used on OpenSimulator. Unfortunately, there is not yet any equivalent advice for the OpenSimulator Bullet and ODE physics engine with regards to mesh performance."
[11:48] Bill.Blight on the wiki
[11:48] Andrew.Hellershanks How much of an impact is there to OS if you have a physics mesh with a high poly count vs. low poly count?
[11:48] Bill.Blight to bullet a LOT
[11:48] Bill.Blight to ubODE not as much
[11:48] Andrew.Hellershanks ok.
[11:48] Bill.Blight but high poly counts cause viewer lag more than server lag ..
[11:49] Andrew.Hellershanks A model may have high poly count but a lower poly count mesh can be used for the object physics.
[11:50] Bill.Blight Unless it is a physical object, meaning the "Physical checkbox" then a high poly model can blow up the sim .. just like on bullet ..
[11:50] Bill.Blight yes andrew
[11:50] Bill.Blight that would be the preferred way to do it, but that would take someone actually learning 3d modeling , or at least how to upload separately and then re-combine
[11:51] Sheera Khan: so the physics mesh is taxing the simulator whereas the visual mesh is more important to the viewer
[11:51] Andrew.Hellershanks Using physics from file where the file used is the same as the main mesh could just contribute to some viewer lag.
[11:51] Bill.Blight yes
[11:52] Bill.Blight When your viewer has to send that visual information over the wire, your viewer lags, the sim does not really care about the visual data, as MOST bandwidth can handle that, it is the viewer rendering that is the issue
[11:52] Andrew.Hellershanks Using the same mesh file for physics as the object itself is a sort of lazy way for people to provide a physics model without creating a simpler mesh for the physics.
[11:52] Bill.Blight Since our viewers don't use GPUs efectively
[11:52] Bill.Blight EXACTLY andrew
[11:52] Andrew.Hellershanks nods
[11:53] Bill.Blight and laziness is what created this mesh issue in the first place, not the any sort of physics engine
[11:53] Andrew.Hellershanks I'd like to get a better video card. The one I'm using now doesn't work as well as the previous one I had. It makes a noticable difference in the viewer performance.
[11:56] James.atLLOUD me too - those headsets are so very tempting
[11:56] Andrew.Hellershanks Any other topics for today?
[11:56] JayR.Cela the so called better video cards are not all what they are cracked up to be unless you are playing a modern video game
[11:56] Andrew.Hellershanks James, if you mean VR headsets I don't have any interest in them myself.
[11:56] JayR.Cela waste of money $$$
[11:57] Andrew.Hellershanks JayR, I wouldn't spend the several hundred dollars that is the price tag for some of the "higher end" cards on the market.
[11:57] Sheera Khan: the link Andrew posted helps me in that discussion. The section about not using step 2 was added back in 2013... Any creator should have had the opportunity to read it. If he did not, well, his fault
[11:58] Andrew.Hellershanks Sheera, We can provide documentation but we can't force people to read it.
[11:58] James.atLLOUD lol - seems the common sentiment
[11:58] Sheera Khan: right Andrew... but they can't blame the devs for that
[11:59] Sheera Khan: it was them not to read the instruction manual ^^
[11:59] Andrew Hellershanks nods
[11:59] Leighton.Marjoram nods Sheera
[11:59] JayR.Cela @Andrew / I agree and the lower price one use less power and do not require the cooling necessary in order to run the higher priced versions
[12:02] Andrew Hellershanks: The 8600GT video card I used to have worked better than the nVidia 9500GT I have now in terms of the viewer.
[12:03] Andrew Hellershanks: I once read a comment that said the second digit was more important in terms of performance than the first number in the case of those four digit models.
[12:03] Andrew Hellershanks: My experience seemed to have backed that up.
[12:03] Bill.Blight I run dual 960's and as far as the viewer goes not much better than the single 780 I have in the other system
[12:03] Bill.Blight but as far as 3d rendering and video editing, the dual cards run circles around it
[12:04] Andrew Hellershanks: I don't have a lot of room in my machine for those video cards that take two slots.
[12:04] JayR Cela: If you are into high end stuff / it's a fine hobby
[12:05] Bill.Blight Our viewer use such an old openGL standard that they can't take advantage of newer cards
[12:05] JayR Cela: not very practical and of no need here in OS
[12:05] Bill.Blight some of us do things OTHER than os ... LOL
[12:05] Bill.Blight :P
[12:06] Andrew Hellershanks: I am interested in some higher-end stuff but viewer performance is more an every day need.

Personal tools
About This Wiki