Simulator User Group/Transcripts/2013.07.02
| Prev 2013.06.25 | Next 2013.07.09 |
List of Speakers
| Adamburp Adamczyk | Andrew Linden | ANSI Soderstrom |
| Baker Linden | Duckie Dickins | Faust Vollmar |
| Ima Mechanique | Kelly Linden | Kennylex Luckless |
| Kitto Flora | Levio Serenity | Lexbot Sinister |
| Lucia Nightfire | Mona Eberhardt | Nalates Urriah |
| Qie Niangao | Rex Cronon | Simon Linden |
| TankMaster Finesmith | Techwolf Lupindo | Vincent Nacon |
| Whirly Fizzle | Yuzuru Jewell |
Transcript
[12:01] Simon Linden: I can get going with the release news ...
[12:02] Kennylex Luckless: Sorry for all 100 prime I has on floor here just now.
[12:02] TankMaster Finesmith: heya simon, andrew!
[12:02] Yuzuru Jewell: Hello, Qie.
[12:02] Rex Cronon: hi qie
[12:02] Simon Linden: I was, Friday and yesterday, but am back in the code mines today
[12:02] TankMaster Finesmith: fun!
[12:02] Simon Linden: Anyway, no releases this week due to the US holiday
[12:02] Simon Linden: The info from last week is here : http://community.secondlife.com/t5/Second-Life-Server/Deploys-for-the-week-of-2013-07-01/td-p/2063041
[12:03] Simon Linden: We should, at minimum, have new RC code next week on Wednesday
[12:03] Ima Mechanique: new as in update or replacement
[12:03] TankMaster Finesmith: yes
[12:04] Andrew Linden: I was just in a meeting with the sunshine guys. They are trying to get a simulator ready for next week's RC that has some new server code (disabled) in preparation for that project's viewer.
[12:04] Simon Linden: Sorry, wrong link there. This is what's currently deployed: http://community.secondlife.com/t5/Second-Life-Server/Deploys-for-the-week-of-2013-06-24/td-p/2058449
[12:05] TankMaster Finesmith: yeah, i heard there was some merge issues with the new region SSA cap handshake on the main grid
[12:06] TankMaster Finesmith: baker!! :D
[12:06] Baker Linden: hello
[12:06] Mona Eberhardt: Hello Baker!
[12:06] Rex Cronon: hi baker
[12:06] Adamburp Adamczyk: yo baker, happy birthday btw dude
[12:06] Andrew Linden: Similarly, "project interesting", which is some viewer-side work to complement some improvements in the server-side interestlist code, is still trying to fix all of their blocker bugs before they make the code public
[12:07] Baker Linden: hah, thanks
[12:07] Whirly Fizzle: Heya Baker, Nice 'fro ;)
[12:07] ANSI Soderstrom: hi chicken :)
[12:07] Mona Eberhardt: It's your birthday, Baker? Happy Birthday!
[12:07] Andrew Linden: they still have some crashes they're working on
[12:07] TankMaster Finesmith: nice project name!
[12:07] Baker Linden: Well, it's not my real birthday, my SL birthday was 2 months ago though
[12:07] Kennylex Luckless: Baker is a developer, not a chiken.
[12:08] Andrew Linden: meanwhile, what I'm working on right now is to get some data for comparison between pre-interesting and post-interesting performance for scene load
[12:08] Simon Linden has another item when you're done
[12:08] Andrew Linden: and also some scene load comparisons for the improvements in the HTTP texture download system that Monty Linden has been working on.
[12:08] Andrew Linden: ok I'm done
[12:09] Simon Linden: Speaking of viewer work, we just threw the switch on a repo that is now publicly available: https://bitbucket.org/lindenlab/viewer-bear
[12:09] Whirly Fizzle: o.O
[12:09] Simon Linden: This has bug fixes that we released as a project viewer a couple of weeks ago
[12:09] Mona Eberhardt: Andrew, I want to ask about ALM. What do you guys define as "reasonable performance" for someone who's running with it enabled?
[12:09] Whirly Fizzle: Snapshot fixes! \o/
[12:09] Simon Linden: We'll be updating that download soon ... it just was merged with the latest viewer-release code
[12:10] Qie Niangao: ALM? Advanced Lighting Model? That seems pretty viewer-centric, no?
[12:10] Andrew Linden: ALM = ?
[12:10] Mona Eberhardt: Advanced Lighting Model.
[12:10] TankMaster Finesmith: ALM is what used to be called lighting and shadows before materials
[12:10] Mona Eberhardt: I read on Nal's blog that LL believe 75% of the users can enable ALM and have reasonable performance.
[12:10] Andrew Linden: Hrm... I'm not really the right person to ask.
[12:11] Mona Eberhardt: Here: http://blog.nalates.net/2013/06/30/second-life-news-2013-27/
[12:11] Andrew Linden: I'd guess that performance with that will depend very much on what video card someone is using.
[12:11] TankMaster Finesmith: Oz has access to those stats, ask at his UG tomorrow
[12:11] Whirly Fizzle: I think the "reasonable performance" is 10FPS or over with ALM enabled.
[12:11] Lexbot Sinister: on an empty sim? or?
[12:11] Whirly Fizzle: ...which is noy really that reasonable.... ;)
[12:11] TankMaster Finesmith: Oz mentioned he thought it was higher than 10
[12:11] Mona Eberhardt: 10FPS... But where? On my land, which is relatively empty, i get 15-20.
[12:12] Mona Eberhardt: On other areas... It depends.
[12:12] Mona Eberhardt: In*
[12:12] Whirly Fizzle: ALM was enabled by default for a lot more cards with the release of materials though
[12:12] Lexbot Sinister: Right. Would be interesting to see the testing area for a "reasonable expirience"
[12:12] Duckie Dickins: I got it turned on and I'm getting 15fps
[12:13] Andrew Linden: Duckie, you're getting 15fps now... what were you getting prior?
[12:13] Mona Eberhardt: 7fps here, but I can't be bothered to disablee it.
[12:13] Kitto Flora sets Ultra, 257 range, shadows on, and gets 22 FPS
[12:13] Duckie Dickins: the same....12 to 15...depending on which direction I'm looking
[12:13] Whirly Fizzle: Big list of the cards ALM is enabled for now: http://community.secondlife.com/t5/English-Knowledge-Base/Materials-FAQ/ta-p/2051467 (bottom of that page)
[12:13] Qie Niangao: I'd sooner just not login at all than turn off shadows at this point.
[12:13] TankMaster Finesmith: having a mass of avatars in your view will kill performance no matter what system you have
[12:13] Nalates Urriah: I have ALM enabled in 4.4.2 and I get 19+/- here
[12:13] Mona Eberhardt: True.
[12:13] TankMaster Finesmith: processing avatar data is highly CPU intensive
[12:14] Lexbot Sinister: it's not an e-penish measuring contest. The point is that we kinda want people little behind the curve to be able to see materials, too.
[12:14] Kitto Flora wants 3-D
[12:14] Andrew Linden: Heh, yeah I usually run with my settings low but I enabled shadows recently and really liked it.
[12:14] Whirly Fizzle: Oz said the last stats they got showed about 30% of sessions had ALM enabled
[12:14] Mona Eberhardt: I think it'd be good if we could have a yardstick to define this reasonable performance of 10+fps, i.e. how many avatars in the region, how loaded the region should be, etc.
[12:14] Simon Linden: fwiw I go from 24 down to 14 enabling it
[12:15] Andrew Linden: So, what would you all consider to be "reasonable performance"? That is, where would you draw the line between "reasonable" and "unreasonable"?
[12:15] Ima Mechanique: 20 FPS
[12:15] TankMaster Finesmith: usually around 30 fps is considered reasonable, 24 is movie speed though
[12:15] Mona Eberhardt: I'd say 20fps too, that'd be a good number.
[12:15] ANSI Soderstrom: 5FPS is already just a dream for me
[12:15] Nalates Urriah: When key and mouse instructions go laggy and slugish.
[12:16] Whirly Fizzle: Yeah, 20-30 FPS
[12:16] Ima Mechanique: any lower and SL gets unusable. preferably should be 25+
[12:16] Faust Vollmar: Thats a very subjective question. You'll get higher answers from combat folks than you will from others
[12:16] Rex Cronon: i think 30fps sounds resonable, or at least 25. if u want 100pf that is kind of high:)
[12:16] Mona Eberhardt: At 20 or so, it's really smooth and precise (at least to my eyes).
[12:16] Mona Eberhardt: Especially compared to 10 or less.
[12:16] Kitto Flora: If your client FPS is too low - time to buy new hardware, and its getting quite cheap now. Computer parts dealers eager to make sales :}
[12:17] TankMaster Finesmith: i can get 170 fps in a simple seen with no avatars around and alm off
[12:17] Lexbot Sinister: well, doesn't the eye see only 26 frames per second or so anyways?
[12:17] Lexbot Sinister: and the rest is for movement...
[12:17] Mona Eberhardt: Heh. Laptops aren't exactly upgradable.
[12:17] ANSI Soderstrom: ^^
[12:17] Duckie Dickins: I guess it also depends on what stuff you got turned on..I don't use the preset graphic settings.....I got the sliders all pushed to the right except sky......and a draw distance of 128 meters..one thing I don't have on is anisotropic filtering
[12:17] Rex Cronon: i think we can see at around 30 fps
[12:17] Ima Mechanique: slightly lower Lex, which is why cinematic rate is 24FPS, a little over average human perception
[12:17] ANSI Soderstrom: 45.1FPS
[12:18] Faust Vollmar: 24 is where you prevent the flicker effect, but smoothness of motion improves all the way up to ~100
[12:18] Nalates Urriah: Duckie, does anisotropic make much difference on your system?
[12:18] Andrew Linden: myself... I think 30 fps is reasonable and 10fps is not, so my line would probably be about 15 fps
[12:18] Kennylex Luckless: My Laptop can give me 35 40 FPS, but just now do m fan works bad so I run it only on low power and has 10 FPS wirh ALM and DD 256
[12:18] Whirly Fizzle: Iys noy just a performance issue with ALM though. A lot of places look ugly with it enabled because there are too many local lights around. Plus you see half the grids feet broken because they still use invisiprim shoes. Also FXAA isn't as sharp as AA (my pet peeve lol)
[12:18] Duckie Dickins: yeah it drops the FPS by about 4 or 5
[12:18] Whirly Fizzle: *not
[12:19] Mona Eberhardt: For me, anisotropic makes no difference. Neither does changing the antialiasing from 2x to higher values.
[12:19] Lexbot Sinister: so for avarage walking around and not doing anything specific, aroud 24 gives a reasonable expirience, i'd say.
[12:19] Faust Vollmar: The combat crowd would probably say that 30 is the bare minimum
[12:20] Whirly Fizzle: If ALM is enabled, your AA is fixed. Changing the setting doesn't work.
[12:20] Mona Eberhardt: I see.
[12:20] Duckie Dickins: I was talking to lex the other day though. I think user experience is down because SL really doesn't have gamers with high end gaming rigs as their target audience......most of us have rigs a few years old and it seems the target audience is sorta getting forgotten....it's why people like Rex is stil using a non mesh viewer. lol
[12:20] Lexbot Sinister: The combat crowd usually don't mind spending on hardware. The avarage home mom who like photography and shopping, does mind ...
[12:20] Whirly Fizzle: Then if you disable ALM, AA is broken till relog heh
[12:20] ANSI Soderstrom: you see it wrong, if wurope would have a better ping to the US, then LL would get much more gamers
[12:20] Rex Cronon: duckie. it has more to do with money:)
[12:21] ANSI Soderstrom: wurope = europe
[12:21] Techwolf Lupindo: I can work with 10FPS update with 60FPS monitor rate. What bugs me is jitter, I get driven nuts with a fps update change of 1fps update every few seconds. Prevents smooth walkning and so on.
[12:21] ANSI Soderstrom: the ping is always about 250-300
[12:21] Lucia Nightfire: unless you have AA disabled, which I do, heh
[12:21] Duckie Dickins: well sure money also is a factor otherwise everyone would have gaming rigs
[12:21] Lexbot Sinister: It's not a coincidence that fashion and clothes shopping are so huge in SL, you know?
[12:22] Faust Vollmar: Yeah theres too many variables for a concrete answer. Framerate for what kind of scene on what kind of hardware. I'd imagine most of SL is on Intel HD which is notorious for poor OpenGL support in general
[12:22] Kitto Flora: Intel HD because of laptop chipset?
[12:23] Duckie Dickins: Yeah faust..in general a large portion of SL's target audience are on rigs purchased at Best Buy that were on sale for $599 with a monitor...
[12:23] Whirly Fizzle: Intel HD driver crashes make up 25% of Firestorms total crash rate. Id imaging v3 is similar.
[12:23] TankMaster Finesmith: intel HD is getting much better OGL support lately
[12:23] Faust Vollmar: Lot of consumer prebuilts use it too iirc
[12:23] Lexbot Sinister: Doesn't LL have statistics for all hardware used?
[12:23] TankMaster Finesmith: the haswell CPU doesnt have any rendering issues in SL and supports OGL 4.0
[12:23] Kitto Flora: MMM... SL dont work well on a C64 either. Or a Vic20
[12:23] TankMaster Finesmith: yes, lex, and yes, intel is quite popular
[12:23] Faust Vollmar: Ah so they are getting better about it? Last I checked you couldnt use OGL newer than 2.1 because Intel refused to support it
[12:24] Mona Eberhardt: Heh. Pretty soon we'll see people bitching about how SL can't run on a ZX Spectrum 48.
[12:24] Kennylex Luckless: http://www.youtube.com/watch?v=vAZHJa91JHk
[12:24] Mona Eberhardt: Or on a ZX81.
[12:24] Kennylex Luckless: SL on Apple 2
[12:24] TankMaster Finesmith: the sandybridge supports up to 3.2, both ivebridge and haswell supports 4.0
[12:25] Kitto Flora: Thats using a Crapple 2 as a monitor. SL is running on something else.
[12:25] Lexbot Sinister: The way i see SL, is like a pair of shoes... the shoes shall fit me, for me to enjoy it. And you don't tell the customer to get different feet, because then the customer gets a new store instead. So telling people "update your computer or GTFO" is throwing out customers..
[12:25] Adamburp Adamczyk: hey nowt wrong with a spectrum
[12:25] Adamburp Adamczyk: i learned to rpogram on one of those
[12:25] Adamburp Adamczyk: even got eh emulator for my pc
[12:25] Kitto Flora: Spectrum owns all your bytes.
[12:25] Kennylex Luckless: Well, I did run SL in a IBM a day ago, 1 CPY no 3D.
[12:26] ANSI Soderstrom: how about a texas instruments calculator.... any chance for SL ?
[12:26] Kennylex Luckless: 1.9 FPS
[12:26] Lexbot Sinister: The home mom will shrug, get a tablet and play angry birds instead. Since i bet nobody has missed the computer sales going way down, no?
[12:26] Adamburp Adamczyk: and the Spectrum is hte only computer in th eworld it;s 100% legal to emualate, thank Sir alan sugar
[12:26] Rex Cronon: we need a slviewer with plugins for graphics. so today i can use a 2d rederer and in few hours a 3d :)
[12:26] Duckie Dickins: mm angry birds.
[12:26] Vincent Nacon: sorry I was late
[12:26] Lucia Nightfire: lol, these meetings get snowballed so easily
[12:26] ANSI Soderstrom: we need a WebGL interface for SL too
[12:27] Mona Eberhardt: No worries, Vincent. I'm sorry my country's politicians are alive.
[12:27] Simon Linden: It's a tough problem, since there really is a huge spectrum of hardware that it would be nice to draw on ... from phones and tablets all the way to high-end machines
[12:27] Duckie Dickins: well there wasn't any server updates and the only real news is tanks team accidently killed the stats server. :P
[12:27] Whirly Fizzle: @.@
[12:27] Vincent Nacon: good going!
[12:27] Kitto Flora: To get SL on your calculator, enter 75, tap the bottom of the case 3 times, and turn it upside down.
[12:28] Andrew Linden: well, the reason this topic was brought up, I think, was that maybe LL's list of supported cards for ALM was too... liberal? such that some people with "supported" cards would end up with "unreasonable" performance.
[12:28] Kennylex Luckless: Now when we ralk about bad and strange things, is LI calculation a srever thing?
[12:28] Lexbot Sinister: Sigh, Kitto. You aren't really getting it...
[12:28] Andrew Linden: The line between reasonable and unreasonable might be as high as 30, and as low as 10... it depends on who and what they're doing
[12:28] TankMaster Finesmith: yeah, it is, i think its based on OGL level supported by the card
[12:28] TankMaster Finesmith: and not performance
[12:29] ANSI Soderstrom: shh kenny, first the DEVELOPERS news
[12:29] TankMaster Finesmith: plus theres a bit of errors in the GPU table that are scewing the results
[12:29] Whirly Fizzle: Well, as long as the user realises they can disable ALM whe needed, its not really a problem.
[12:29] Whirly Fizzle: *when
[12:29] Duckie Dickins: yeah..usually when you say supported you have a reasonable expectation of performance..however in most games a card is support if it will work on low settings.......so maybe ALM was a bit too liberal
[12:29] Andrew Linden: however, the graphics settings allow you to turn things way down, so I guess people will just have to disable things as they see fit
[12:29] Andrew Linden: but if we're supporting a card that most people think should not be in that list, we should update the list
[12:29] ANSI Soderstrom whispers: but kenny is right, developers see : there is a Agenda on this OH
[12:29] Levio Serenity: whats acceptable is too subjective to really quantify anyways
[12:30] Whirly Fizzle: The fact disabling ALM breaks AA until a relog is a real pain though
[12:30] ANSI Soderstrom whispers: https://wiki.secondlife.com/wiki/Simulator_User_Group
[12:30] Lucia Nightfire: evolution or die, lol
[12:30] Andrew Linden: Ah I see Whirly, so that would be a bug.
[12:30] Meeter: Timecheck : User Group is half over
[12:30] Rex Cronon: sorry. i have to draw u all as cubes so I can get "resonable" fps:)
[12:30] Whirly Fizzle: Yeah, its filed
[12:30] Lexbot Sinister: Evolution sometimes means you get adapted all too well to a narrow envoirment, and then YOU die. Meaning, Sl evolves out of their target audiences reach, and then it dies.
[12:31] Andrew Linden: ok, anybody have a new topic for discussion (did I miss a question that went by?)
[12:31] Andrew Linden: ?
[12:31] Rex Cronon: i have
[12:31] ANSI Soderstrom: See the Agenda : https://wiki.secondlife.com/wiki/Simulator_User_Group
[12:31] Rex Cronon: the new llxorbase64
[12:31] Lucia Nightfire: oh boys
[12:31] Rex Cronon: doesn't seem to play nice with php:(
[12:32] ANSI Soderstrom: (and girls)
[12:32] Adamburp Adamczyk: someone pick a goos xmas song
[12:32] Adamburp Adamczyk: good*
[12:32] Kennylex Luckless: I want to talk about what I think is bad LI calculation, so bad I see it as a bug.
[12:32] Lucia Nightfire: rex did you use the first php example when *Correct() got axed?
[12:32] Lucia Nightfire: in the wiki
[12:33] Andrew Linden: Uh... llXorBase64()... this is the first I've heard of it. Kelly?
[12:33] Kennylex Luckless: Look at torus here, one is 90 LI the other is same torus cutted up in bits, linked and 9 LI
[12:33] Rex Cronon: i am not sure lucia
[12:33] Kelly Linden: Rex: llXorBase64 doesn't innately have anything to do with php
[12:33] Kelly Linden: so you will have to ellaborate a bit.
[12:34] Rex Cronon: kelly if u want to communicate with an outside server that use php. it does
[12:34] Kennylex Luckless: Both are material enabled
[12:34] Mona Eberhardt: Simon, have there ever been thoughts of supporting FIPA-ACL and KQML for bots? Perhaps bridging LSL with these two languages?
[12:34] Duckie Dickins: I think the point that lex and I are trying to make is that SL evolving...first with sculpties....then mesh...ALM..however the target audience is getting left behind....if everyone had to go out and get an Nvidia GTX 760 to enjoy material and lighting....I'd rather go play Diabo 3 or some other game now that I have a card that supports those games.....it's a hard sell to get someone to upgrade their gear and then not get drawn away by another game that looks quite a bit more pretty without the lag effects of not being graphic optimized since anyone can upload a texture here.
[12:35] Simon Linden: Mona - no, I've never heard of those ... do you have any links to check out?
[12:35] Andrew Linden: Kennylex... that is interesting that the LI is smaller for the linked piece
[12:35] Mona Eberhardt: KQML: https://en.wikipedia.org/wiki/Knowledge_Query_and_Manipulation_Language
[12:35] Vincent Nacon: that's the price for using OpenGL render....
[12:35] ANSI Soderstrom whispers: oh no, "they" increase now the LI for linked Prims :(
[12:35] Mona Eberhardt: FIPA-ACL: https://en.wikipedia.org/wiki/FIPA-ACL
[12:35] Whirly Fizzle: re what Kenny is talking about MATBUG-217
[12:35] JIRA-helper: http://jira.secondlife.com/browse/MATBUG-217
[#MATBUG-217] Adding Normal/Specular breaks PE calculation causing objects to be deleted or returned due to SIM beeing full. - Second Life Bug Tracker
[12:35] Faust Vollmar: The optimization paradox is a problem for anything with user-created content.
[12:35] TankMaster Finesmith: actually, duck, a geforce 760 wont work with LL's viewer
[12:35] Faust Vollmar: OpenGL doesnt have much to do with it
[12:36] Kelly Linden: Rex: I can understand that you would pass data back and forth between outside systems and maybe pass some of that on to llXorBase64 - but you are going to have to be more specific about what it means that 'it doesn't play well' - what is the bug? What is it doing wrong, is it doing things wrong in LSL or PHP?
[12:36] Andrew Linden: I wonder if the pieces are small enough that they get turned into boxes for the physics engine, or maybe just convex hulls
[12:36] Mona Eberhardt: SL would be an excellent tool for merging human agents and scripted autonomous agents for group decision making etc.
[12:36] Kennylex Luckless: Thatt is 4 prims with 2 LI even they has PRIM as shape
[12:36] Vincent Nacon: it does do much since it doesn't even use DX
[12:36] Lucia Nightfire: no, I think he's just talkign about the change in LI calc with complex primitive "Prim" and "non-prim" physics settings
[12:36] Andrew Linden: That is something we noticed with the Havok physics engine... it was sometimes cheaper to collide things as a list of convex pieces rather than a single concave shape.
[12:36] ANSI Soderstrom: the LI calculation System will fail on Torus's
[12:36] ANSI Soderstrom: other rpims go well
[12:36] Vincent Nacon: but either way... too many polygons in the system
[12:37] ANSI Soderstrom: but Torus is a bit "random"
[12:37] Lucia Nightfire: *combinations
[12:37] Rex Cronon: that is the thing. i don't know if the lsl fucntion is borker or the php is borked. but i don't get same result
[12:37] Rex Cronon: is borked*
[12:37] Lexbot Sinister: i thought all other prim shapes except the box got 1 LI in the revised LI calculations? Same with sculpts in a linkset?
[12:37] Rex Cronon: i used the php example provide on the lslxorbase64 page
[12:38] ANSI Soderstrom: kenny made a little torus with over 15.000LI, this will fill a whole sim and can block Sandboxes
[12:38] Lucia Nightfire: I'd say the first example should be compatible with the current function and the second was compatible with *Correct
[12:38] Simon Linden: Mona - I can't foresee any work along those lines in the near-term. Perhaps the best way to work on bots or agents would be external systems running the interesting code, communicating back into SL via the LSL web interfaces
[12:38] Duckie Dickins: someone kiill kenny....problem solved. :P
[12:38] ANSI Soderstrom: no
[12:38] Kennylex Luckless: Yes, my "it os over 9000 LI prim went over 15 000 LI
[12:38] Mona Eberhardt nods. Thanks Simon.
[12:38] Andrew Linden: Lexbot, all legacy "extruded prim" shapes get an LI of 1... unless you use one of the new features that are tied to new LI calculations.
[12:39] Lucia Nightfire: theres a jira on the overflow bug
[12:39] Simon Linden: Mona - you could also, of course, connect into the viewer directly
[12:39] ANSI Soderstrom: the jira is closed before the "good" comments are postet
[12:39] Lexbot Sinister: Oh, so materials would break the legacy calculation then?
[12:39] Kelly Linden: Rex: Those code snippets were not done by me - and they look to be just using llXorBase64StringsCorrect not llXorBase64. I would imagine they would need to be fixed or modified to work with the new function's behavior
[12:39] Andrew Linden: Yes, I think maybe materials are tied to the new LI.
[12:39] Simon Linden: anything with materials doesn't use the old calculation
[12:39] Qie Niangao: yes, they are
[12:39] Whirly Fizzle: They are
[12:40] Rex Cronon: i modified them:)
[12:40] Faust Vollmar: Its best to assume any new feature forces LI calc
[12:40] Whirly Fizzle: Thats what Oz said on MATBUG-217 yeah
[12:40] Andrew Linden: The idea was that the legacy system of LI = 1 / prim was just incorrect. A torus really is more expensive than a box for some things.
[12:40] ANSI Soderstrom: but why doeas material increase a torus up to 15.000LI and other prims get "as expected" ? there is some bork in the Torus
[12:40] Lexbot Sinister: it's best? I don't think it's "best", it means sculpts can't be used for whatever reason.
[12:40] Kelly Linden: Rex: Without more details about what you are getting, what you expect, and why you expect I don't know how to help.
[12:40] Lucia Nightfire: more verts
[12:40] Kennylex Luckless: I can accept new LI, but when you need to build a torus out of 10 torus to lower the LI, then something is bad.
[12:40] ANSI Soderstrom: a ring is even more complex than a torus
[12:40] Kelly Linden: Rex: It is possible that PHP
[12:41] Lucia Nightfire: I have doubts that a ring is more complex, lol
[12:41] Kelly Linden: Rex: It is possible that PHP's base64 encoding also doesn't support nulls in the key and so behaves like llXorBase64StringsCorrect and you should use that.
[12:41] ANSI Soderstrom: then cut him and make it hollow
[12:41] Simon Linden: I don't have an answer on that LI calculation, but it does sound funky
[12:41] Kelly Linden: But I am not a PHP expert.
[12:42] Vincent Nacon: I think it's safe to say people don't understand the problem with increasing number of polygons instead of generalizing by the number of prims instead.
[12:42] Andrew Linden: Kennylex, yeah. Really we should be using a list of convex shapes for collision under the hood... however that is a bit difficult to implement because of how list shapes interract with other features, such as scripted collision callbacks.
[12:42] Lucia Nightfire: if it was I'd had used it for the base of my 12 million LI experiment
[12:42] Andrew Linden: legacy content... scripts gotta know which pieces were hit
[12:42] Faust Vollmar: Toruses / Rings are little monsters, especially on the physics side
[12:42] ANSI Soderstrom: we had some tests with torusses with a high number of polygons and low numbers.... just by cutting the torus (and CTRL-Z) we got very different LI
[12:43] Object: Hello, Avatar!
[12:43] ANSI Soderstrom: Hello Object with the UUID of c44e9ecd-856d-777a-3ae6-26d1e7c07885
[12:43] Rex Cronon: str00[34](df2389-0$%(SD9a)*%#P9#$Dert43L;Ddg), pass00[15](DB2XvJD9CAX4Ob2), result in sl: xor00[48](ICQAa05zaQlnZHBnC1tTbWgXeyZzZx0HJCpAe1F+OwAmVQ==), result in php: xor00[48](ICQAa05zaQlnZHBnC1tTbWgXeyZzZx0HJCpAe1F+fwZWPw==)
[12:43] Qie Niangao: Well, but Vincent, poly count has surprisingly little to do with LI, unfortunately. As witness Kenny's diced-up tori scoring lower than a full torus, despite having many more tris.
[12:43] Kelly Linden: so very close rex.
[12:43] ANSI Soderstrom: if a torus will get less Polygons, sometimes increases the LI about 200Prims, sometimes decrease the Primcount... on just the SAME cut every time
[12:43] Kennylex Luckless: My 4 prim hat become 270 LI for I did want use normal map to remove a torun simulare a seam.
[12:44] Andrew Linden: Indeed. I suppose if I were to try to make an extruded shape object creation UI again, I'd limit it to convex shapes, and make it automatically line up multiple pieces for more interesting shapes.
[12:44] Lexbot Sinister: Will it mean that you can no longer build with the SL prims, if you intend to use materials? I mean, if you make a torus, expecting it to be 1 LI and it's suddenly something else, that is pretty content breaking... and the content being broken is the entire inworld building system (which is old, but not redundant)
[12:44] Lucia Nightfire: you're SOL when it comes to wanting materials to not use the new accounting
[12:44] Lexbot Sinister: Then thats simply wrong.
[12:44] Vincent Nacon: that's not even.... correct
[12:44] Qie Niangao: Right. Some prims are fine with Materials, some: forget it. They're just obsolete.
[12:45] Lucia Nightfire: but in regards to the penalty for just "Prim" and "non-Prim" combo's on legacy prims, that may need looking at
[12:45] Vincent Nacon: and you actually have less polygon when you cut up the torus
[12:45] Levio Serenity: LL wants you to buy more land to use the new fancy content. shocker.
[12:45] ANSI Soderstrom: if you vut a torus, you'll get different LI every time
[12:45] ANSI Soderstrom: *cut
[12:45] ANSI Soderstrom: https://my.secondlife.com/de/ansi.soderstrom/snapshots/51cdd971c20f7549d0000001
[12:45] Kennylex Luckless: The blue one, diferent LI even openind is ames size but diferent ways
[12:45] Kennylex Luckless: sides
[12:45] ANSI Soderstrom: https://my.secondlife.com/de/ansi.soderstrom/snapshots/51cdd8e8b52dc71565000001
[12:45] ANSI Soderstrom: https://my.secondlife.com/de/ansi.soderstrom/snapshots/51cdd892464be5310a000001
[12:45] Vincent Nacon: unless you're gonna count the cap ends
[12:46] Andrew Linden: Lexbot, basically yes. It is a push to use mesh... the creation pipeline isn't as nice (unless you like developing in 3rd party app and uploading) but the final result can be more optimized.
[12:46] ANSI Soderstrom: and the poligons are DEcreasing, but the primcount is increasing about at least 100LI
[12:47] Lexbot Sinister: Materials or not, for the sake of enjoyment of SL, and keeping some building options in SL, prims must count as 1 LI. Or it's just becomming a way of getting rid of the prim building toold altogether, alienating people from SL.
[12:47] ANSI Soderstrom: ^^
[12:47] ANSI Soderstrom: yay
[12:47] ANSI Soderstrom: save the Torus !
[12:47] Lucia Nightfire: don't forget spending money on uploads, winks
[12:47] Duckie Dickins: hahah
[12:47] Lucia Nightfire: no temp uploads allowed eithers, lol
[12:48] Whirly Fizzle: Wel, not in another 8 days heh
[12:48] Mona Eberhardt: Part of what made SL so inviting to people was that they could create content within it. And now we're basically without any reasonably easy to use tools for in-world creation of up-to-date content.
[12:48] Lexbot Sinister: A push to use mesh? No, essentially a push to log out and stay logged out. One MAIN key aspect of SL is that you can build inworld, like with no other game/program. You want to get rid of that?? Really????
[12:48] Duckie Dickins: well there's always cloudparty. :P
[12:48] Kennylex Luckless: I could not resist, sorry :D
[12:48] Lexbot Sinister: Someone must have fallen on their head, really hard.
[12:48] Faust Vollmar: See what I mean about the UCC Optimization paradox?
[12:49] Qie Niangao: Bear in mind, however, that the wackiness is all about Havok, not so much about rendering... so if you change physics type to Convex Hull (or None), most of the dopiness goes away.
[12:49] Levio Serenity: my pet-peeve in all this new accounting, you can optimize your mesh down to 1 prim download, and add a script and it goes up to 2 or 3 LI from server costs : /
[12:49] Vincent Nacon: Lex... it's kinda a moot point because if people don't care about render weight, it's going to get laggy for their videocard
[12:49] Andrew Linden: Actually, I agree with Lexbot. I like the prim-based building model and would like to see it maintained as an option.
[12:49] Vincent Nacon: so people would come less because of that
[12:49] Lexbot Sinister: Thank you Andrew.
[12:49] Simon Linden: It sounds like you're conflating adding materials to SL with a grand scheme to re-design and shift the whole product. It's not that big ... it was a project that came out of the open source world, done with cooperation with Linden, and we tried to get it into the system as best as possible
[12:49] Kennylex Luckless: Problem with convec hull, it remove function of prims.
[12:50] Lexbot Sinister: There's no point in optimizing SL while getting rid of the things that people find entertaining here, and things that make SL unique
[12:50] Kennylex Luckless: Then you need 4 prims to make a roon and not only one as all do today
[12:50] ANSI Soderstrom: we dont want to say there is something bad, we just want to say there is something with the Torus calculations and we should try to do it better
[12:50] Rex Cronon: i don't think adding 1 script to a prim affects havok
[12:50] Faust Vollmar: Yeah, script penalty is discouraging mesh in furniture... Or more correctly, its encouraging use of an un-linked normal prim to do the script container work.
[12:51] Whirly Fizzle: Adding scripts can push the LI up
[12:51] ANSI Soderstrom: adding material too
[12:51] ANSI Soderstrom: thants not the point
[12:51] Lucia Nightfire: I think he means it shouldn't, heh
[12:51] Lexbot Sinister: Simon- i bet it's not planned that way on the grand scheme, but it might be the effect, none the less.
[12:51] TankMaster Finesmith: it certainly would be neat to do mesh building within SL
[12:52] Kennylex Luckless: Converting things to convex hulls can make things go from 7 prims to 130
[12:52] Vincent Nacon: maybe Second Life is long overdue for a sequel.... Third Life
[12:52] Mona Eberhardt: OK, colour me stupid here, but why doesn't a tiny prim that uses a tiny bit of a 1024x1024 texture get penalised and instead we see scripts being calculated into LI?
[12:52] Levio Serenity: i can see adding some to the LI for larger numbers of scripts, but 1 prim mesh with 1 script bumping up to 2-3 LI is just lame
[12:53] Kitto Flora: Haha - what Vincent said
[12:53] Whirly Fizzle: heh yeah, stay penalising use of 1024 textures.
[12:53] Whirly Fizzle: *start
[12:53] Mona Eberhardt: I've seen thousands of objects that are full of small prims, each one with its own 1024x1024 texture.
[12:53] Levio Serenity: peeps will just continue to use 1 prim sculpts which are lame
[12:54] Faust Vollmar: They cant go in and mess with existing builds, they can only make new features force LI, else alot of existing stuff would either break or suddenly use alot more than before
[12:54] Lexbot Sinister: not 1024 textures... but size of object vs size of texture.
[12:54] Rex Cronon: i might understand if an prim had over 100 scripts in it, but for one scipt u increase li by one?
[12:54] Mona Eberhardt: Yes Lex, exactly that.
[12:54] Faust Vollmar: Its either content damage or continued lack of efficiency
[12:54] ANSI Soderstrom: there should also be a "Maximum" on the LI, prims with a LI of 15.000and more will block Sandboxes just with ONE MicroPrim
[12:54] Kennylex Luckless: I cold not have it in correct size here
[12:54] Mona Eberhardt: A tiny object with a ridiculously large texture.
[12:54] Andrew Linden: On the 1024 texture issue... I think the viewer should just be made smarter -- it shouldn't download higher resolutions when it doesn't need them
[12:54] Simon Linden: There are a lot of different variables for "cost" ... and we've had a lot of discussions on how to account for them. Geometry, texture load, bandwidth, script cost, physics shape ... I think those are the big 5
[12:54] Duckie Dickins: breach had some guns that used to lag you when they came into view because they were full of high res textures on tiny prims.
[12:55] Lucia Nightfire: ANSI a jira was filed to request overflow not carry over into other owner parcels
[12:55] Kennylex Luckless: https://my.secondlife.com/kennylex.luckless/snapshots/51d30714cb530c460a000001
[12:55] Qie Niangao: ANSI, yeah, I think something is genuinely broken at the high end of the LI calculations. They simply don't seem credible.
[12:55] Vincent Nacon: doesn't the texture have LOD?
[12:55] Andrew Linden: so that really big objects could use 1024 and the viewer would download that resolution, but very small objects with high-res textures would only be downloaded at lower resolutions.
[12:55] ANSI Soderstrom: griefers will not longer rez selfreplicable objects, the hide a microprim and the sandbox is closed
[12:55] Meeter: Timecheck : User Group is almost over
[12:55] Rex Cronon: andrew. i think it should download them but as a resized version:)
[12:55] Levio Serenity: i asked the firestorm team for a TI calculation for finding a "texture impact" on objects
[12:55] Simon Linden: Your viewer chooses the LOD it requests, but I don't think that's tied to the render size, but instead distance (?)
[12:56] Whirly Fizzle: Mona, I can show you a gorgeous mesh house build that every single texture is a 1024, even on the tiny teacups. It locks most people up dead in there. If you're anywhere near it, all textures just blur/sharpen/blur/sharpen over & over because it fills texture mem up. Its death trap of a build lol
[12:56] Andrew Linden: That is already the idea... but the code to figure out what resolution to download, and getting that resolution, is complicated (or so I understand from what other devs have said).
[12:56] Kennylex Luckless: Strange, I got other LI results here than I got in Natoma.
[12:56] Vincent Nacon: I thought LOD were based on the scale of the object on the screen than actual scale nor distance
[12:56] Mona Eberhardt: Of course, Whirly. Each texture is, what, 4MB? Imagine having to download 100 such textures.
[12:57] Vincent Nacon: smaller than that, Mona
[12:57] Andrew Linden: And storing them all in RAM.
[12:57] Vincent Nacon: it's using JPEG2000
[12:57] TankMaster Finesmith: may, ive got 6GB on my vid card :P
[12:57] Lexbot Sinister: Either way, with materials forcing a pretty alien Li calculation onto regular prims, it makes the generic curious SL newbie apprehensive of trying to build, and trying out new features. long term, productive residents nearly all started the same- poking some prims, trying some easy scripts, etc. Make that difficult, or hard, and you lose out on a whole lot of new future customers.
[12:58] Andrew Linden: Hrm... good point Lexbot.
[12:58] Whirly Fizzle: Shes right, its 4Mb (for 32 bit)
[12:58] Vincent Nacon: oh right... 32
[12:58] Lucia Nightfire: not a single Guy Faulkes mask among those protestors, shames
[12:58] Mona Eberhardt: Whirly, up to how much RAM does the viewer allocate to textures?
[12:58] Vincent Nacon: this is a protest?
[12:59] Lucia Nightfire: vinc, the prim protest effigy on teh ground
[12:59] Vincent Nacon: oh them
[12:59] Simon Linden first thought it was a slow rez griefer attack
[12:59] Qie Niangao: Maybe should have kept the Materials interface script-only, to keep the innocent from getting hurt by it. ;)
[12:59] Vincent Nacon: they don't count
[12:59] Whirly Fizzle: Yexture sizes/memory: http://community.secondlife.com/t5/image/serverpage/image-id/721i430BA9577AA33BE6/image-size/original?v=mpbl-1&px=-1
[12:59] Vincent Nacon: :P
[12:59] Levio Serenity: occupy prim street! whoo !
[12:59] Kennylex Luckless: I made that when Mesh did come to be funny
[12:59] Whirly Fizzle: *texture
[12:59] ANSI Soderstrom: lol simon
[12:59] Mona Eberhardt: Thanks.
[12:59] Duckie Dickins: or materials for mesh objects only?
[12:59] ANSI Soderstrom: a creafully grief
[12:59] Qie Niangao: well, materials are actually very nice for simple box prims, etc.
[12:59] Kennylex Luckless: They are free to copy now
[12:59] ANSI Soderstrom: *care
[13:00] Lexbot Sinister: I mean, we all know that building in prims isn't effective, or the best way to generate content for SL, but, it starts people on wanting, and then climbing upwards to the point where they want to learn software that will help making better content. Building in prims also creates better understanding of the effort put into making something, and gives a better understanding of pricing of goods others have for sale, etc.
[13:00] Ima Mechanique: Mona, depends on your Graphics card
[13:00] Simon Linden: It seems better to be able to slap materials onto a cube and look at them ... that lets you do basic walls and such
[13:00] Qie Niangao: bad, bad news for stuff like hollow spheres, tori, etc.
[13:00] ANSI Soderstrom: i'll still build the next 20 years with prims, i have no need (and no mind) about meshes.... but i dont want to "forget" the most nice prim, which is a torus
[13:00] Vincent Nacon: orr.... I got a better idea
[13:00] Meeter: Thank you for coming to the Server User Group
[13:00] Vincent Nacon: get rid of regular Prims
[13:00] Vincent Nacon: sculpty too
[13:00] Vincent Nacon: :D
[13:00] ANSI Soderstrom: nope
[13:01] Lucia Nightfire: naked city
[13:01] Lucia Nightfire: I see what you did ther
[13:01] Simon Linden: Not gonna happen in SL, I"m sure
[13:01] Qie Niangao: the worst part of it is that Materials contaminates the *entire* linkset, not just the thing to which it applies.
[13:01] Kennylex Luckless: The last "Save the prim " cube is a mesh
[13:01] Lexbot Sinister: Agreed, Quie
[13:01] Vincent Nacon: yeah not gonna happen but it's an ideal way to do it in a "video game" contents
[13:01] Lexbot Sinister: it's not predictible, for a regular curious user.
[13:02] Faust Vollmar: You drop prims, you drop the core of SL. Sculpties on the other hand... Its a shame theres so many years worth of content built on them, they are monstrous little things that basically can't be got rid of now.
[13:02] Faust Vollmar: SL Tribbles as it were
[13:02] Vincent Nacon: basically what I'm really saying is...
[13:02] Rex Cronon: faust. automatic sculptie to mesh conversion:)
[13:02] Vincent Nacon: Third Life
[13:03] ANSI Soderstrom: iam lucky that iam not a newbie, i would get ILL about the inconsistence about some calculations of LI..... cutting a 1LI torus, 200LI, pressing CTRL-Z, 108LI
[13:03] Faust Vollmar: Rex thats complicated by the sculptie-specific modelling tricks to get LODs right and whatnot
[13:03] Yuzuru Jewell: Rex, My tool can convert sculpt to mesh.
[13:03] Lexbot Sinister: Well if LL comes up with a new mesh building system inworld that is both logical and useful, i'm of course all up for that...
[13:03] Andrew Linden: Third Life would have to be much better... but migrating content from SL to TL would be... very hard if not impossible.
[13:03] Rex Cronon: i have tool than can convert sculpties to objects. which can be easily converted to mesh:)
[13:04] Andrew Linden: The shapes and appearances could be migrated (probably) but not scripts.
[13:04] Lucia Nightfire: TL is already patented, its the dreamworld
[13:04] Kennylex Luckless: Prims are creative, so are material, so I rather has a bit lag than lose ability to use peims with materials
[13:04] Andrew Linden: That is, the changes required to fix the script engine would make it impossible to migrate scripts to a new system while keeping them functional.
[13:04] Simon Linden: I'm going to leave with a Rod Humble quote:
[13:04] Simon Linden: Q: Is there going to be a Second Life 2?
A: It's not going to be for a few years, but it'll be something next gen. What we'll call it, I don't know.
[13:05] Qie Niangao: Well, Mesh in SL is awfully crippled by the representation, locking the model geometry to the object itself. So can't just tell an object to assume a different Mesh geometry, as with prims & sculpties.
[13:05] ANSI Soderstrom: i cant alter a mesh with a script, so there is ALWAYS a need for prims in addition with working on scripts
[13:05] Ima Mechanique: I'll wait for 3rd Dimension with Rift
[13:05] Adamburp Adamczyk: hey Simon, can i ask yo ua quicl questin?
[13:05] Rex Cronon: tc simon
[13:05] Adamburp Adamczyk: quick question*
[13:05] Simon Linden: fwiw from http://www.sfchronicle.com/business/article/Will-Second-Life-have-a-second-life-4616560.php
[13:05] Simon Linden: sure, what's up?
[13:05] Adamburp Adamczyk: is tha whiteboard still in boston reception?
[13:05] Adamburp Adamczyk: the one where resiodents who visited could sign
[13:06] Adamburp Adamczyk: sf, not boston
[13:06] Kennylex Luckless: Whatever happens, OpenSom and that Cloud junk can not beat SL.
[13:06] Faust Vollmar: Heh. I hope those fixes include the utf-8 truncation that makes hmac functions impossible without base64 hijinks, Andrew
[13:06] Simon Linden: ah, right, I've never been to the Boston office
[13:06] Adamburp Adamczyk: sorry, i meant sf
[13:06] Adamburp Adamczyk: not boston
[13:06] Simon Linden: You know, I'm not sure where that is
[13:06] Andrew Linden: The whiteboard still exists, but it is not in a spot that is easily accessible by visitors.
[13:06] Adamburp Adamczyk: *CENSORED*
[13:06] Ima Mechanique: I want HMAC in an LSL function :-(
[13:07] Adamburp Adamczyk: it was, at one point, the star attraction i think
[13:07] Adamburp Adamczyk: that, and joppa's ballet dancing
[13:07] Duckie Dickins: You know that storage area at the end of raiders of the lost ark......that's where the whiteboard went.....along with phil linden's avatar. :P
[13:07] Faust Vollmar: Ima there is an implimentation, it's just not fast.
[13:07] Adamburp Adamczyk: Get to the Joppa!
[13:07] Lexbot Sinister: I'd like to thank the Lindens listening, while i was on my soapbox ;)
[13:08] Faust Vollmar: For my purposes I'd be just as satisfied with SHA-256 as I would with HMAC-SHA1 to be honest
[13:08] Ima Mechanique: Faust, I mean a singl llHMAC() type function
[13:08] ANSI Soderstrom: oki, thanks to the lindens for listening to residents things
[13:08] Simon Linden: Thanks everyone for comign today
[13:08] Mona Eberhardt: Thanks Simon.
[13:08] ANSI Soderstrom: and not laughing
[13:08] Mona Eberhardt: Thank you Lindens.
[13:08] Rex Cronon: tc simon
[13:08] ANSI Soderstrom: at
[13:08] Faust Vollmar: Ima yeah it would be nice but I've been told its impossible due to deep rooted stuff in LSL
[13:08] Ima Mechanique: thanks guys
[13:08] Qie Niangao: Thanks Simon, Lindens... have fun all.
[13:08] Duckie Dickins: now go eat your lunch lindens. :P
[13:08] Kennylex Luckless: Tc and have fun,.
| Prev 2013.06.25 | Next 2013.07.09 |