User:Zero Linden/Office Hours/2007 Nov 15

[7:38] Zero Linden: Welcome all
[7:38] Zha Ewry: Oooh. Tree, Windlight hates you :-)
[7:39] Burhop Piccard: Hey Zero.... I'm here this time !
[7:39] Zero Linden: Excellent
[7:39] Tree Kyomoon: could be, but it sure is awesome. Love the glow especially !
[7:39] JetZep Zabelin: Tao's hat is glowing =D
[7:39] Zha Ewry: Tree, on my client, you are eyeballs.
[7:39] JayR Cela: hi Tree / hi Zha / hi Sai / hi Tao / hi Zero / hi Morgaine :_)
[7:39] Zha Ewry: JUST eyeballs
[7:39] Zha Ewry: Hey JayR
[7:39] Tree Kyomoon: yes, ever since I got windlight my invisiprims have been screwed up
[7:40] Morgaine Dinova: waves
[7:40] SignpostMarv Martin's: invisi prims seem to be working fine
[7:40] Tao Takashi: Hi
[7:41] Zero Linden: My six year old son, who is watching over my shoulder, says he likes the skeleton the most! A point for Tree!
[7:41] Prokofy Neva: We should have the ability to completely shut off Windlight.
[7:41] Morgaine Dinova: Tree++
[7:41] Tree Kyomoon: yay!
[7:41] Wyn Galbraith: A son with good taste.
[7:41] Zha Ewry: grins Way to go tree!
[7:41] Dr Scofield: wonders whether skeletons are the avs of te future
[7:41] Tree Kyomoon: /waves
[7:41] Zero Linden: Okay - let's get on with it........
[7:41] Zero Linden: Welcome to my office hourse
[7:41] Zero Linden: Dnag
[7:42] Zero Linden: I always mistype that
[7:42] Zero Linden: My office hours
[7:42] Dr Scofield: after all we'll ultimately will all look like that
[7:42] Zero Linden: The transcript is published to the wiki (thank you Tree!) -
[7:42] Morgaine Dinova: We'll bring hay next time for the horse, it's only fair :-)
[7:42] Zero Linden: - so speak freely, speak in public
[7:42] Wyn Galbraith: Everything is recored speak openly speak freely.
[7:42] Tree Kyomoon: no problemo
[7:42] Arawn Spitteler: Would dislexics sit at Office Horses?
[7:42] Wyn Galbraith: *recorded
[7:43] Zero Linden: Indeed I'm sure I do!
[7:43] Tree Kyomoon: dyslexics untie!
[7:43] SignpostMarv Martin: Prok: wouldn't disabling the shaders in the graphics panel "shut off windlight" ?
[7:43] Zero Linden: Okay, Durhop has asked to put something on the agenda
[7:43] Burhop Piccard: Who is Durhop :-)
[7:43] Penny Wakowski: what the fuck is the matter
[7:43] Morgaine Dinova: A friend of Burhop's
[7:43] Penny Wakowski: can anyone see me?
[7:43] Zero Linden: Gosh - clearly NOT ENOUGHT COFFEE
[7:44] Zha Ewry: slips a shot of espresso into Zero's mug
[7:44] Zero Linden: Penny - we see you
[7:44] Coffee Mug: whispers: Ahh! Fresh Hot Coffee
[7:44] Penny Wakowski: i dont see me
[7:44] Penny Wakowski: shoot
[7:44] Burhop Piccard: So.... A quick geometry disucssion?
[7:44] JetZep Zabelin: i dont she can see us *yet*
[7:44] Arawn Spitteler: got coffee at the ministry of magic, but can't finish that first cup.
[7:44] Zha Ewry: In the context of infrastructrue and AWG work?
[7:44] Zero Linden: Yes - though somehow I don't think it will be quick....
[7:44] Wyn Galbraith: Who's the bighead?
[7:44] Burhop Piccard: Let me type a few lines first and then we can discuss....
[7:45] Burhop Piccard: The primary interest is to allow content creators to use better tools and support richer forms of 3D expression.
[7:45] JetZep Zabelin: Penny, give your client more time to load everything
[7:45] Zero Linden: But mostly becase it will be an instructive case study in how these sorts of things work in the AWG
[7:45] Multi Gadget: v1.52.0 by Timeless Prototype
[7:45] Penny Wakowski: thanks jetzep
[7:45] Burhop Piccard: SO this high level architecture view is really just the expression of this idea in terms the software architects and codes can (hopefully!) understand. [1]
[7:45] Wyn Galbraith: LOL It's SignpostMary!
[7:45] Burhop Piccard: This functional view cuts across several areas... obviously, most of this is out of scope of the AWG.
[7:46] Tree Kyomoon: aha, its "atmospheric shaders" that causes my skeleton to disappear
[7:46] Burhop Piccard: However, one key piece is the comunication of geometric data between server and client. So my interest in the AWG is the protocols and standards for Agent/Region/Viewer comunication will be able to support geometric data that is not currently in the Prim/Sculpted prim mold.
[7:46] JayR Cela: tree / i have same problem with walls dissapearing / using windlight and atmospheric shaders
[7:47] Dr Scofield: burhop, could you explain what special requirements you see for that com?
[7:47] Burhop Piccard: I think it is less "special requirements" and more a generalization of existing requirements.
[7:48] Saijanai Kuhn: in the long run, it may be "alternative representations" as well
[7:48] Saijanai Kuhn: back, btw
[7:48] Burhop Piccard: Specifily, that we can pipe any type of goemetric data between components.
[7:48] Dr Scofield: ok, but i assume you have something in mind...
[7:48] Zha Ewry: Beyond, being able to mark a chunk of data as needing a different renderer (and having a different encofing)do you see a big hit on protocol?
[7:48] Burhop Piccard: Inially, its just the prims/sculpties we have today.
[7:49] Burhop Piccard: The parallel is much like with a web Browser. A web browser supports HTML but depends on plugins for a number of specialized formats (i.e. PDF, AVI).
[7:50] Burhop Piccard: SO I have no format in mind - I just want a sent of sandards so others can add their own.
[7:50] Tree Kyomoon: I thought all aspects of rendering to the screen were plugins of some sort, some were just more integrated into the browser...(remembers gif plugins)
[7:50] Morgaine Dinova: In REST terms, if I understand it correctly, Burhop would like to be sure that either the noun vocabulary is extensible, or that the datatypes associated with existing noubs is not fixed.
[7:50] Zha Ewry: nods
[7:50] Zha Ewry: Mostly,I think, from the protocol layer, it should be that the datatypes be extensible
[7:50] Neas Bade: it's a bit different in this case however, as the server needs to fundamentally grok that shape as well, otherwise physics is borked
[7:51] Burhop Piccard: Essentially, I'm working backward from the generalized needs of content creators ... they need better tools. They need more flexibliby in being able to express themselves.
[7:51] Zha Ewry: And.. make sure we can derive a bounding box for
[7:51] Neas Bade: your webserver doesn't actually need to understand gif
[7:51] Tree Kyomoon: I agree zha
[7:51] Zha Ewry: glances as Neas...
[7:51] Zha Ewry: Bounding box
[7:51] Zha Ewry: Yeah
[7:51] Wyn Galbraith: groks Neas.
[7:51] Zero Linden: So, I see that we all agree that the AWG protocols need to allow for alternative geometric representations
[7:51] Burhop Piccard: So that means improvments to browser... improvements to physics.... improvements to protocol (AWG)
[7:51] Saijanai Kuhn: but as physics improves the "bouding box" will become more detailed. H4 supports proxy-meshes (or something like that) --lower detail shapes instead of just BBoxes
[7:52] Zero Linden: even if none are defined or implementat on day one
[7:52] Tree Kyomoon: Neas, unless it wants to optimize for storage and delivery of gifs..but yes
[7:52] Burhop Piccard: And Zero - that is all I think I want.... an awarness of this.
[7:52] Zero Linden: But.... there are several ways we could fit that
[7:52] Zero Linden: into the protocols
[7:52] Dr Scofield: hmm...do we have to really care about that level of detail on the AWG protocol/architecture level?
[7:52] Zha Ewry: Right, we assume that the current prim set, and the current meshes, are just what we need to do today, but a subset of what we need to do to future proof this
[7:52] Pixel Gausman: Dr Sco: y
[7:52] Zero Linden: Well, yes, we do have to... here's why
[7:52] Burhop Piccard: Zero - yes.... I think go after the viewer folks to think about the other parts :-)
[7:52] Saijanai Kuhn: we should allow for gracefully declining to display something, at least
[7:53] Morgaine Dinova: Zero: yep, it doesn't matter whether they ARE implemented on day one. What matters is that they CAN be implemented on day 2 without breaking day 1 protocols.
[7:53] SignpostMarv Martin: heh, optimization of delivery leads to "lossless" textures becoming lossy upon reciept
[7:53] Tree Kyomoon: we should only care if theres some advantage in terms of vectorizing storage/delivery righ?
[7:53] Neas Bade: though if you just want bounding box, you are back to non hollow prims. The more interesting physics issues come with more complicated shapes than boundy boxes
[7:53] Saijanai Kuhn: that's a bug, Sign
[7:53] Burhop Piccard: Morgaine - yes.
[7:53] Pixel Gausman: convex hulls for physics should be ok
[7:53] Zero Linden: I'm going to take it as a given that for a piece of geomtry to work the region or viewer handling it HAS to understand that format
[7:54] Zha Ewry: nods Ooh. Right. I tend to forget that we focus on bounduing boxes, due to the additive style today. Hulls, yes
[7:54] Saijanai Kuhn: This stuff is already i teh works for avatars I believe
[7:54] Zero Linden: And I'm going to assume that the Agent domain doesn't care a wit - it 's all just typed bits
[7:54] SignpostMarv Martin: Saijanai: a bug caused by the "optimised delivery process"
[7:54] Saijanai Kuhn: physics, though?
[7:54] Zero Linden: Yes, physics too
[7:54] Burhop Piccard: Yes... but physics is a bit easier. So the engine needs to know some material properties, mass, moment of inertia etc.
[7:54] Neas Bade: just bringing up some points, as we've been very much flogging getting ODE physics into shape in opensim, and the complexity isn't always immediatly obvious
[7:54] Dr Scofield: ok
[7:55] Neas Bade: zero: yeh, agreed, agent domain doesn't care
[7:55] Neas Bade: it's just shuffling bits
[7:55] Zero Linden: So now we have a fundimental decision
[7:55] Zha Ewry: Agent, and Asset, better not care
[7:55] Saijanai Kuhn: so agent domain is just about distribution to the proper sim?
[7:55] Tree Kyomoon: I wanted to bring up Streaming in MoCap data at some point..unless that is totally off topic
[7:55] Burhop Piccard: So, the server side would need some API to cumunicate this... but we are getting into implementation details.
[7:55] Zha Ewry: because, if they do, we're really pushing in stuff to the wrong places
[7:56] Arawn Spitteler: Improved functionality is normally a hit on client-side capacity.
[7:56] Zha Ewry: Server side, needs to have a way of saying "I can/can't deal with this graphics chunk" and.. if we want to be really clever, one might allow for an asset to have
[7:56] Zero Linden: Are graphical formats seen as alternate representations - one object can be in both SLPrims and X3D?
[7:56] Zha Ewry: several gemotreic presentations
[7:56] Zha Ewry: I really like that
[7:56] Zha Ewry: ?12 off
[7:56] Neas Bade: just using content negotiation?
[7:56] Zha Ewry: If we can have a "here's the full, base form"
[7:57] Wut Burt: That would provide a fallback for clients without X3d support
[7:57] Pixel Gausman: yes, we should allow for the server and client to maybe negotiate on prefered formats?
[7:57] Saijanai Kuhn: goes back to OpenDoc, of course ;-)
[7:57] Zero Linden: OR - do objects come in a format - and if you don't know that format you are SOL
[7:57] Zha Ewry: and "here's the fancy form for better sims and clients"
[7:57] Burhop Piccard: Zero - thats a bit tricker.... There for clearly file formats.
[7:57] Burhop Piccard: However, file formats do not always map well to comunication formats.
[7:57] Zero Linden: And then, once that is squared away, we then need to decide at what granularity the format is negotiated
[7:58] Zero Linden: Burhop - we can transmit any binary blob easily enough
[7:58] Zha Ewry: If at all possible, Zero, I thnk we're all happier, with allowing for multiple
[7:58] SignpostMarv Martin: XHTML 2 "solves" the multi-format issue by letting the client know what formats are available, and letting it pick which image to serve
[7:58] Dr Scofield: negotiate formats? request object & specify formats that you grok with priority?
[7:58] Pixel Gausman: i can see several representations of an object. maybe one SLPrim, maybe another collada-ish, maybe another OBJ, another X3D....
[7:58] Burhop Piccard: Yes... granualarity is an issue. You dont' want to send a 1 meg prim.
[7:58] Zha Ewry: Especially, if we imagine that we will have tons of old conent for ages, and want to bring in new content slowly
[7:58] Pixel Gausman: Dr Sco: y
[7:58] Neas Bade: signpostMarv: that's been in the HTTP spec since 1998 actually
[7:58] Zero Linden: SMM - why did they put that in XHTML 2 vs. using content negotiation that exists in HTTP ?
[7:59] Zero Linden: Or are you talking about the same thing?
[7:59] SignpostMarv Martin: different thing
[7:59] Pixel Gausman: sometimes you might want lighter weight prims, sometimes heavier
[7:59] SignpostMarv Martin: content negoation uses the HTTP_ACCEPT header right ?
[7:59] Zero Linden: Yes
[7:59] Neas Bade: yes, using accept
[7:59] Zero Linden: I'm guessing that decision has much more to do with authroing content than
[7:59] Zha Ewry: Interestingly, you've got a two part content negotiaion problem
[7:59] Burhop Piccard: Pixel - yes maybe a verigh tight OBJ. Collada is a bit differnt - more for moving assets between systems. You could use it for tha but sill might need a way tosend small pieces acorss the wire.
[7:59] SignpostMarv Martin: the thing I'm referring to uses an ambigious URL- e.g. an image URL without a file extension,
[7:59] Zha Ewry: (sim understanding it, and then the clients)
[7:59] Neas Bade: well, one is at layer 4 and one at layer 7
[8:00] SignpostMarv Martin: an attribute in the element indicates what formats are available,
[8:00] Neas Bade: layer 4 is prefered for all our work I would think
[8:00] Pixel Gausman: SignPostMarv: not sure we can cram all that into HTTP GET functionality, but the same idea, yes
[8:00] Zha Ewry: sighs
[8:00] Zero Linden: in most deployments, it is MUCH easier to control the XHTML content to specify the alternatives, than it is to configure the web server to do content negotiation
[8:00] Zha Ewry: Oooh. Ick. I dont' like that I said that.
[8:00] Tao Takashi: so if I send some object can I not simply also send an identifier whcih tells in which format it is in?
[8:00] Zha Ewry: You'd ideally like to be able to keep all the style around, and pass the "best" one down to clients
[8:00] Pixel Gausman: Burhop: thats why i said colloda-ish, maybe something like collada that is more runtime appropriate
[8:01] Neas Bade: my guess is the xhtml 2 spec additions are just because of tooling
[8:01] Tree Kyomoon: so like MIME types in the XML?
[8:01] Zero Linden: Well, this is veering more into the second question
[8:01] Zha Ewry: Because, you're oing to have several sets of clients looking into lots of sims
[8:01] Zero Linden: Do we
[8:01] Tree Kyomoon: doesnt like MIME
[8:01] Neas Bade: as the tool writers didn't think to just dav enbable their server and use http
[8:01] Burhop Piccard: Pixel - yes I thought we were on teh same page... just thought it was good for eveyone else to hear.
[8:01] Zero Linden: have the viewer and the region negotiate a single graphics format up front?
[8:01] Kurt Stringer: What about load - what if the client would want an alternative, to render on an older platform for instance
[8:01] Burhop Piccard: Zero - perhaps.
[8:01] Zha Ewry: Viewers
[8:01] Zero Linden: Do we have them negotiate on each object transfer?
[8:01] Zha Ewry: This gets into the question of concsensus view, and such
[8:02] Tao Takashi: ok, maybe upfront is sufficient
[8:02] Zero Linden: Do we have them negotiate a set of formates on connection, and an ordering
[8:02] Tao Takashi: might be lists of supported formats
[8:02] Zero Linden: and then server them that way?
[8:02] Zha Ewry: I wuold hope, we say "Send me type X, Y Z, ocntent"
[8:02] Zha Ewry: Yeah,. in order, and such
[8:02] Saijanai Kuhn: Howabout a list of acceptable conent, in order of preference?
[8:02] Zha Ewry: I'd like Intels new cool format est,
[8:02] Burhop Piccard: I think you would have multiple graphics formats if you wnat to evolve SL.
[8:02] Neas Bade: Saijanai that's how HTTP works
[8:02] Zha Ewry: then, tri-meshes,, and then, prims, if that's all you have
[8:02] Zero Linden: What do we do if there are objects in the world in only a format the viewer doesn't understand?
[8:02] Tao Takashi: in the future you might also want to have completely different clients, like games
[8:02] Zha Ewry: "MISSING!"
[8:03] Tao Takashi: it would be nice if they also can use their own format but still can run on that architecture
[8:03] Burhop Piccard: Zero - good question.... perhaps display a bounding box.
[8:03] Zero Linden: Well, missing geometry is much worse than missing textures
[8:03] Neas Bade: we could have the ubiquitous "broken prim" like you used to have on webpages
[8:03] Saijanai Kuhn: boudning box?
[8:03] Zha Ewry: winces at Neas
[8:03] Wut Burt: users probably won't like lots of jarring red "missing" cubes. i'd prefer an optional fallback mechanism
[8:03] Zha Ewry: Yeah
[8:03] Neas Bade: roughly same shape, but some ugly color
[8:03] Tree Kyomoon: Plugin requried...windows will now search for the player..
[8:03] Zero Linden: And solid bounding boxes can be pretty misleading and intrusive to other objects
[8:03] Saijanai Kuhn: use the proxy mesh researved for the physics engine?
[8:03] Zha Ewry: If you had a bounding box. Ick.
[8:03] Neas Bade: yeh, but what else you going to do?
[8:03] Burhop Piccard: So.... either the server does not allow certain types.... or it does and the view informs the user he needs a plug in to see that peice of geometry
[8:03] Neas Bade: just kick people off
[8:04] Neas Bade: ?
[8:04] SignpostMarv Martin: Zero: couldn't the viewer "reverse engineer" the physics shell for an object, rendering a grey object that's the right shape ?
[8:04] Zero Linden: Right - so everyone agrees the bounding box solution really isn't workable
[8:04] Neas Bade: plugins suck
[8:04] Neas Bade: for this
[8:04] Zero Linden: SMM - not if it doesn't understand the format of the data
[8:04] Neas Bade: because then you'll get content that you need vista to see because no one build the pluging for linux
[8:04] Zero Linden: Or do you mean before upload?
[8:04] Saijanai Kuhn: but the next step up is the bounding mesh used by H4 for avatar objects. If that is available (and it has to be), use tthat
[8:04] Neas Bade: so you'll fundamentally have everyone seeing different stuff
[8:04] Burhop Piccard: Neas - Not sure if you can open it up any other way without seting some new graphics standard.
[8:04] Zha Ewry: Note, tho, that, the architecture, gets to sor tout the tech support, and then, we work with the user expeirnece peole about that
[8:04] Saijanai Kuhn: it still will get you the right physics behavior
[8:05] SignpostMarv Martin: I'm saying as an alternative to not displaying anything, the physics shell could be rendered to the viewer in a format that is supposed to be supported by all viewers
[8:05] Pixel Gausman: Burhop: scandal!
[8:05] Pixel Gausman: :-)
[8:05] Zha Ewry: We're going to have to deal with evolving graphical elements
[8:05] Zero Linden: Well - rather than put the negotiation at the per-object level
[8:05] Tree Kyomoon: everyone sees different stuff anyway, some of us are colorblind
[8:05] Zha Ewry: Since, we konw, that this isn't going to be prims as is, for the next 10 years
[8:05] Tree Kyomoon: (not me though)\
[8:05] Zero Linden: we could have the simulator and viewer negotiate a set of formats where the contract is:
[8:05] SignpostMarv Martin: sort of like "view PDF as HTML"- you get the basic gist of the document, but it's nowhere near accurate
[8:05] Dr Scofield: some of us are not able to see graphics at all
[8:05] Zero Linden: viewer: "I can render any of these"
[8:06] Burhop Piccard: Again, its a bit like a web browser.
[8:06] Zero Linden: region: "I can represent EVERY object in at least on of these formats"
[8:06] Zero Linden: *one
[8:06] Zha Ewry: nods at Zero
[8:06] JetZep Zabelin: dont design for the lowest common denominator that just strinks
[8:06] Zha Ewry: CLose to content negotiation, as it is
[8:06] Burhop Piccard: Zero -yes could be.
[8:06] Tree Kyomoon: so who / where do we get the master table of formats? MIME?
[8:06] Zero Linden: MIME of course
[8:06] Zha Ewry: NO Mime
[8:06] Burhop Piccard: Consider HTML that gives some text if you won't display the GIF
[8:06] Zero Linden: :-)
[8:06] Dr Scofield: agrees...viewer would also be able to specify priority
[8:06] Zha Ewry: MIME
[8:07] Zha Ewry: Sigh
[8:07] Zha Ewry: yeah
[8:07] Zha Ewry: what else.
[8:07] Neas Bade: it's not MIME encoded, it's just MIME types
[8:07] Dr Scofield: ASN.1
[8:07] Pixel Gausman: p.s. we might also want to allow alternative representations for blind/low vision/deaf users
[8:07] Zha Ewry: So.. that's only being used to mark and negotiate, we're not MEM encoding, right?
[8:07] Zero Linden: MIME types = good, MIME encoding = scary
[8:07] Zha Ewry: *MIME
[8:07] Zha Ewry: Aye
[8:07] Wyn Galbraith: Cerious Sofftware who makes Thumbs Plus supports pretty much ever graphic format there is.
[8:07] Burhop Piccard: Dr Scolfield - yes. Thinking long term, SL on a cell phone probalby doesn't want to download a CAD file of a aircraft.
[8:07] Arawn Spitteler: wonders if the client would be more efficient, choosing which texure to download.
[8:07] Zha Ewry: We're all on the same page, there Zero
[8:07] Tree Kyomoon: we need to go down to MIME and wake them all up...tell them that some formats have been added since 1990.
[8:07] Zero Linden: Noooooooooo - no MIME encoding - you can take that as my first dicitorial decree to the AWG
[8:08] SignpostMarv Martin: HTML supports <object type="image/gif" [data="http://example.com/foo.gif">Your] browser cannot render GIFs</object>
[8:08] Neas Bade: x-shape/cylinder
[8:08] Zha Ewry: grins at Zero "I don't think you'll get too much argument from the AWG on the whole"
[8:08] Neas Bade: for instance
[8:08] JetZep Zabelin: cant render gifs? get off the net
[8:08] Dr Scofield: just use network byte order
[8:08] SignpostMarv Martin: it's more used for flash :-P
[8:08] Wyn Galbraith: Gifs are like the first graphic toy.
[8:08] Neas Bade: SignpostMarv: I still think it's so much better to do that in HTTP
[8:09] Burhop Piccard: JetZeb - (I used to have a 1200 baud modem)
[8:09] Zero Linden: But now, consider that region R has negotiated with viewer A for formats X & Y
[8:09] JetZep Zabelin: me too ;)
[8:09] Zero Linden: and negotaited with agent B for formats X & Z
[8:09] Zha Ewry: Oh, Burhop, you had a fast one?
[8:09] Saijanai Kuhn: 300 baud modem, and have used 150 accoustics
[8:09] SignpostMarv Martin: Neas: the object tag spec is meant for static content, not dynamic
[8:09] JetZep Zabelin: i had 16k on my first comp
[8:09] Burhop Piccard: Phone cradle!
[8:09] Zero Linden: Now agent A wants to rez something out of inventory
[8:09] Zero Linden: what format can the region allow?
[8:09] Arawn Spitteler: used to have a 50baud cradle
[8:10] Wyn Galbraith: has modem'd before. "I remember those, pain in the butt."
[8:10] Neas Bade: SignpostMarv: you can do content negotiation with static content with apache
[8:10] Burhop Piccard: Zero - now thats an interestin use case to explore.
[8:10] Zero Linden: It can only allow format X
[8:10] SignpostMarv Martin: Neas: you'd still need to serve something
[8:10] Neas Bade: yeh, it really seems that the sim must at some level set up rules in advance
[8:10] Saijanai Kuhn: and if format x doesn't exist in inventory...
[8:11] Burhop Piccard: SO again, you may have to fall back to an alternate representation.
[8:11] Wut Burt: "could not rez item - sim does not support format X"
[8:11] Arawn Spitteler: thinks algorhyth might be similar, to allowing grid space to content legal in only a few cultures
[8:11] Zero Linden: can't rez, 'cause one or the other viewer won't be able to see it
[8:11] SignpostMarv Martin: refusing to rez an object because someone can't see it would be a very, very bad thing to enforce
[8:11] Wut Burt: but the trap door is that you ALWAYS provide a prim fallback representation, which if you're lazy is a bright red cube
[8:11] Kurt Stringer: I see no difference whether from inventory or scene
[8:11] Tao Takashi: well, they can use a difference viewer then
[8:11] Zero Linden: Well, perhaps - but not all fall backs are useful
[8:11] SignpostMarv Martin: you could log a bot into a sim that tells the region it supports nothing
[8:11] Neas Bade: but if you base it on viewers then you just end up ping of death-ing all the other users, push your content, then they can't see it when the re log in
[8:12] Tao Takashi: if I use Netscape 1.0 I might also not be able to see or use that many web pages
[8:12] JetZep Zabelin: an incorrect representation would be better than none
[8:12] SignpostMarv Martin: that would prevent anything from being rezzed
[8:12] Zero Linden: The fall back of Quick time movies to the QUicktime logo
[8:12] Zero Linden: isn't actually a viable alternative
[8:12] Zero Linden: you are effectively shut out of the content
[8:12] Neas Bade: basing what you can push based on who is on the sim at any time doesn't seem to make sense
[8:12] Burhop Piccard: The tricky part (and what makes the web easier) is that you always have to display something.
[8:12] Wyn Galbraith: Would it be any different than substituing a generic font for a font used that's not available on someone's system?
[8:12] Zero Linden: Naes, no I don't think it does
[8:12] Tao Takashi: I am not sure this will be a problem in practice for the most part
[8:12] Burhop Piccard: On the web, if you don't have adobe, you will just have abroken link
[8:12] Morgaine Dinova: Zero: no no, rezzing isn't viewer-related. Rezzing happens in the region. If a viewer can't handle it, it simply doesn't render it, or only minimally, or it can negociate to not get data sent at all.
[8:12] Zha Ewry: nods "I think, here, you have to accept that some people are going to get cut out from time to time, or else, you end up with peole always at the LCD, which is sort of pointless"
[8:12] Zero Linden: Wyn, I suspect that fonts have MUCH more in common than graphic formats
[8:12] SignpostMarv Martin: on the web, if you don't have adobe, your browser will download the PDF
[8:13] Saijanai Kuhn: yeah, but store owners won't appreciate the broken link image for their goods, for example
[8:13] Zero Linden: I can substitute almost any font and still get the content
[8:13] Neas Bade: yes, look at how long it took to get .png out into the net in a reasonable way
[8:13] Zero Linden: subbing a prim even for a sculpty basically loses all shape information if not all semantic content
[8:13] Wyn Galbraith: That's true, Zero, but the basic idea is the same, mostly.
[8:13] JetZep Zabelin: give the prim an alt source
[8:14] Zha Ewry: Alts help, only in some cases, Suppose, someone allows a 3dmax model. Which might be 500 prims worht
[8:14] Wyn Galbraith: If you can't find the torus, use a sphere, sort of thing.
[8:14] Kurt Stringer: I'd much rather see simple prims than be shut out...
[8:14] Zha Ewry: You'd have to have a full ilinkset, and all, to get that out into an alt represenation
[8:14] JetZep Zabelin: the alt could be specified by the creator
[8:14] Tao Takashi: if you have a torus, you might have a complex mesh
[8:14] Zero Linden: So, I think we'll have to see that region operators will have to set a minimal set of formats
[8:14] SignpostMarv Martin: the Mozilla peeps came up with an alternative to MNG that has the first frame as normal PNG data, and the remaining frames in metadata (i think)- the basic gist is it's a new format that has built-in fallback if the viewer supports PNG but not the new animated PNGs
[8:14] Neas Bade: but the issue is that it is relatively easy to go prim -> cool 3d mesh
[8:14] Zero Linden: every object in the region will have to be expressible in one of these formats
[8:14] Tao Takashi: I guess people want different formats at some point anyway
[8:14] Zero Linden: and every viewer in the region will have to render all of those formats
[8:15] Neas Bade: degrading the other way is an unsolved problem in the generic case
[8:15] Zero Linden: then the viewer and region can negotiate additional alternatives on top of that set
[8:15] Tao Takashi: it might mean that I have to use different viewers for different grids
[8:15] Neas Bade: you can't easily degrade a 3dmax shrek into 7 prims
[8:15] Wut Burt: I'd say you need some kind of fallback rather than zero rendering at all, else you might rez some unsupported file type and never be able to return it to inventory...
[8:15] Zha Ewry: Something at that level, feels plausible
[8:15] SignpostMarv Martin: again, this is situation where content negotation isn't used, but what are the possibilities of embedding alternate formats into the default prim format ?
[8:15] Tao Takashi: but IMHO that's ok as long as I at least have one account
[8:15] Burhop Piccard: So, defining graphics formats is a very slippery slope. Getting back to AWG, I'd like to see the structure setup to support any format regardless of what it mgiht be.
[8:15] Saijanai Kuhn: as I said, H4 will allow/require mesh proxies for avatars for collision purposes. Why not use those if all else fails?
[8:16] Tao Takashi: I also think we should at least support some version/format information in the protocol
[8:16] Zha Ewry: The format and data passing, looks easy, Burhup, and for people who have special early needs, things like Zero is desctribing will work pretty well
[8:16] Tree Kyomoon: so would we support a client uploading a mystery type to the server, then the client would be responsible for building a render or handler for that mystery type?
[8:16] Tao Takashi: region owners also might set policies
[8:16] Zero Linden: The only thing I'll say about graphic formats themselves is that for live internaction, they tend to be highly compressed and optomized -
[8:16] Zero Linden: we can send down prim info in as little as 13 bytes (I thin)
[8:16] Tao Takashi: of course
[8:16] Zero Linden: there isn't much room for internal extension flags and data
[8:16] Tao Takashi: in 10 years it might look different though ;-)
[8:16] Morgaine Dinova: If the "least common denominator" that is sent down to client is just position, type and bounding box, fine. Although client should be able to negociate even that away if not interested.
[8:16] Burhop Piccard: Zero yes..... So there is some serious work that would be needed to keep things usuable.
[8:17] Zero Linden: so the negotation or alternative structure will have to be done outside the context of the format
[8:17] Tao Takashi: ah, yes, probably a good idea
[8:17] Pixel Gausman: a lot of tools work in triangles.
[8:17] Burhop Piccard: Zero - as a user, you don't want someone to be able to force too much down your throat.
[8:17] Venice Georgia: hi
[8:18] Venice Georgia: can i ask where can i go to apply for work?
[8:18] Zha Ewry: Yeah, it's going to be "I can render X, in format Y"
[8:18] Burhop Piccard: I should be able to say- I'm not accepting any of your bloated content.
[8:18] Arawn Spitteler: Pooley
[8:18] Neas Bade: though all those tools presupose that you deliver the content on compressed DVDs :)
[8:18] Zha Ewry: That sort of scope
[8:18] Zero Linden: I think, Zha, there is a big difference between "I'm not interseted in geometry" and "I want only location, size and average color" and "I want minimal format X"
[8:18] Neas Bade: not via the interwebs
[8:18] Zha Ewry: Yep
[8:18] Zero Linden: I think the first two are alternative message sets
[8:18] Morgaine Dinova: On a cellphone, I might want to say "Send me NOTHING other than people's positions, and landstream URL".
[8:18] Zero Linden: the later is negotiation for what format the full geo comes in
[8:19] Burhop Piccard: Morgaine - exactly!
[8:19] Kurt Stringer: But the data and the thunking algorithm are seperate things
[8:19] Tao Takashi: the thing is: if you accept multiple formats you somehow need to specify the format when sending an object
[8:19] Zha Ewry: That sounds about right zero, with some concern for not making it too messy in the simple, default cases
[8:19] Tao Takashi: so maybe it can only be one format which gets negotiated
[8:19] Burhop Piccard: But If I'm Joe Engineer showing off my new NASA Rocket peolple should be able to go see it if they want.
[8:19] Zero Linden: wonders why one year-olds find pulling the CDs off the shelf one by one so fascinating.....
[8:20] Tree Kyomoon: mabey we should just have a raw data server and leave everything up to the client, so the server doesnt care whether data represents geometry or toy boats
[8:20] Zero Linden: watches the pile on the floor by his desk grow....
[8:20] Ace Jungsten: hiyi people
[8:20] JetZep Zabelin: Zero, to get your attention :D
[8:20] Zha Ewry: grins "Ask why 2Yos try to put coins into the television set
[8:20] Saijanai Kuhn: guys, one more time:
[8:21] Tree Kyomoon: anyone ever get peanut butter sandwitch in the VCR?
[8:21] Burhop Piccard: Crayons is the dryer is better.
[8:21] Zha Ewry: So, Zero. You're saying "Format and content" vie negotiation and.. some level of focus on allowing fallback?
[8:21] Saijanai Kuhn: by the time this stuff is implemented, bounding boxes will give way to bounding meshes. Those are, by definition, simpler than the complex geometries they represent
[8:21] JayR Cela: treee :_))))) yes / kids and PJ samiches in the vcr / LOL :_)
[8:21] Ace Jungsten: whats this party on about people ..?
[8:21] Saijanai Kuhn: so... the fallback can be the geometry the physics engine uses to represent the object
[8:21] Zero Linden: Sai - gosh - I hope we get this out and working before then!
[8:22] Zero Linden: For now, we couldn't afford to send a bounding mesh
[8:22] Wyn Galbraith: They learn physics that way Zero, the 1 year old. :D
[8:22] Tree Kyomoon: and SL will be rendering hypercubes :)
[8:22] Burhop Piccard: Zero - its just one requirement --- universal geomerty support :-)
[8:22] Zero Linden: wonders if we should give the simulator a shelf of CDs so it can learn physics too....
[8:22] Wyn Galbraith: had finger paints on carpet once.
[8:22] Morgaine Dinova: Dismantling CDs is the start of analysis.
[8:22] Saijanai Kuhn: are you saying that 2 years fr now, SL won't be using H4 with collisions for avatar bits?
[8:22] Wyn Galbraith: Good idea, Zero.
[8:23] Neas Bade: Saijanai, you are confusing internal representation and wire representation
[8:23] Zha Ewry: At least DVDs, if not blueray's Zero. Might as well have it learn state of the art ;-)
[8:23] Wyn Galbraith: Like in "When Harley was One".
[8:23] Zero Linden: No - I'm saying that 2 years from now the average bandwidth to the viewers will not have appreciably increased
[8:23] Neas Bade: on the internal side of things, bandwidth and latency is something you don't worry about
[8:23] Tree Kyomoon: good point saij...makes it seem almost futile to try to decide standards for technology 2 years from now
[8:23] Neas Bade: but getting it out ot folks is much harder
[8:23] Tree Kyomoon: :)
[8:23] Zero Linden: So, here's another poser -
[8:24] Burhop Piccard: Zero - yes... however, SL may have moved into other domains....Intranets for example.
[8:24] Burhop Piccard: Local networks, etc.
[8:24] Zero Linden: If I'm in Region R, and I've negotiated formats X & Y that I must be able to render
[8:24] Tree Kyomoon: rocket packs
[8:24] Neas Bade: The way studios move 3d content for 3d movies is via suitcases filled with DATs on airplanes, not over the network
[8:24] Zero Linden: and then Region S, adjacent to me and in my field of view
[8:24] Neas Bade: for the same reasons more or less :)
[8:24] Zero Linden: insists on format Z, which I can't render
[8:25] Zero Linden: Do I not see the region that I'm standing against?
[8:25] Zha Ewry: nods
[8:25] Dr Scofield: shouldn't S offer the fallback?
[8:25] Zha Ewry: This is another vairation on the intimate nature of edges
[8:25] Morgaine Dinova: Server-side bandwidth will have increased only slightly in 2 years' time, but population growth X client-side bandwidth will have increased a lot more. In time, the only *massive* bandwidth is going to be in P2P. Not there yet though.
[8:25] Zero Linden: Put another way - does format negotiation have to be consistent across a connected set of regions?
[8:25] Zha Ewry: (We talk about this, a bit, in a few office hours back)
[8:25] SignpostMarv Martin: S can offer a fallback, but what happens if the viewer doesn't understand the fallback format ?
[8:25] Zero Linden: Dr. - yes, but I'm talking about the negotiated fall back set
[8:26] Zero Linden: not the optional ones
[8:26] Neas Bade: morgaine, other than fios, no one is seeing a big client side bw bump
[8:26] Burhop Piccard: So you could always have an S format fall back.
[8:26] Neas Bade: and fios is relatively limitted areas
[8:26] Zero Linden: If the fall back set is different in two adjacent regions
[8:26] Zero Linden: then we have this problem
[8:26] Zero Linden: a viewer that supports the smaller set only will not be able to see across the region boundry into the other
[8:26] Burhop Piccard: So, S format is maybe like HTML for me.
[8:26] Neas Bade: universities are still operating on 40 Mbs agregate for the campus
[8:26] Zha Ewry: I could easily be sold that you need uniform capabitlties to be adjacent.. but.. That's ugly when you want to roll new versions
[8:27] Morgaine Dinova: Neas: in the US, you're right. But broadband takeup isn't slowing in the rest of the world :P
[8:27] Neas Bade: or in that order of magnitude
[8:27] Dr Scofield: so, define mandatory fallback format
[8:27] Zero Linden: Well - the problem is there is no good universal fall-back like there is for the web - which by the way is TEXT, not HTML!
[8:27] Neas Bade: fair enough, but if you start excluding parge parts of the global population, you loos eyou metcaff's law effect
[8:27] Morgaine Dinova: text++ :P
[8:27] Wyn Galbraith: Obama says if we make him prez we'll all have broadband.
[8:27] Dr Scofield: there is: just use the existing SL format
[8:27] Burhop Piccard: Ok, so bad analogy.
[8:27] Zero Linden: OH - well, that solves that then.... :-)
[8:27] Dr Scofield: no?
[8:28] JetZep Zabelin: OH then all the broadband providor companies arent going to vote for him ;)
[8:28] Tree Kyomoon: Obama wants coal powered cars
[8:28] Neas Bade: Dr Scofield: please convert shrek to 7 prims via machine algorithm ;)
[8:28] Zha Ewry: Of course, we need to convince him to take over the world, since SL, and the AWG have worldwide aspirations
[8:28] Neas Bade: I'm sure there is at least a couple of PhDs in that exercise
[8:28] Arawn Spitteler: remembers Zero speaking of data degrade for Distance: Data Degrade, could also be used across Sim and Access Lines, which would be handy if the neighboring Sim were on a Server with less connection.
[8:28] Zero Linden: Well - for example, I don't think many people would consider prims to be a good fall back for sculpties (imaginging that scuplties were not in the base set)
[8:28] Dr Scofield: i'm not claiming that it will be beautiful
[8:29] Dr Scofield: well, there are now
[8:29] Saijanai Kuhn: Its easy enough. Use the leggo drawing routine that fills convex hulls with stacked leggos
[8:29] Zha Ewry: (And as soon as Nasa, or soeoen gets moving, galactic)
[8:29] Saijanai Kuhn: just use BIG leggos
[8:29] Dr Scofield: duplos
[8:29] Wyn Galbraith: Invite him into SL (hint hint)
[8:29] Zero Linden: Anyone know how HLA solved this?
[8:29] Tree Kyomoon: HLA?
[8:30] Tao Takashi: I gotta go, Second Liffe Roundtable in Cologne is calling :)
[8:30] Tao Takashi: cya!
[8:30] Saijanai Kuhn: later Tao
[8:30] Burhop Piccard: Zero - yes, I agree. Sculpties are also not a greate fallback for their MAYA/Blender/CAD content (but still better than nothing)
[8:30] Tree Kyomoon: ciao tao
[8:30] Zero Linden: HLA is a sort of military simluator interoperability protocol
[8:30] Dr Scofield: cu tao!
[8:30] JayR Cela: byee Tao
[8:30] Zero Linden: where the objects in the simulation can be driven by different hardware/software at different locations
[8:31] Zero Linden: I have someone at NASA that is going to give me a run down on it soon....
[8:31] Zha Ewry: Zero, next week, I'd like to chew on capabitlitie and how we want to group nouns, and such,, REST style.. and your thoguhts. We had a huge chat on that yesterday, and I'll try and post it as the seeed for the discussion
[8:31] Burhop Piccard: Zero : interesting thought, but does it fit your curent archtiecture?
[8:31] Zero Linden: I'll ask, as they have undoubtly addressed this issue - but theymight have done so with bandwidth (
[8:31] Arawn Spitteler: is pretty sure that some programmer at The Presidio would like to pump Zero also.
[8:32] Saijanai Kuhn: [2]
[8:32] Morgaine Dinova: Zha's chat of yesterday was excellent, we got a hell of a lot covered
[8:32] Zha Ewry: Physics, with some inputs from off world?
[8:32] Zha Ewry: (off SL world)
[8:32] Zero Linden: (mesh is a fine fallback if you can pump lots of bits)
[8:33] Neas Bade: yeh the problem gets a lot easier if it is fundamentally only meant for 100 Mbs LANs
[8:33] Burhop Piccard: Zero - yes.... tricky part is with so many different bandwidth profiles the best fallback changes.
[8:33] Zha Ewry: Just, reliable 1MBit is not safe to assume, tho
[8:33] Tree Kyomoon: Higher Level Architecture
[8:33] Neas Bade: 768kbs down / 128 kbs up DSL it works not so well
[8:35] Zero Linden: Well - that looks like a lot of lovely reading......
[8:35] Neas Bade: very good discussion, very interesting
[8:35] Zero Linden: okay - our hour is up and other work awaits me
[8:35] Saijanai Kuhn: way bloated for our use. Might be a useful pattern in there (or 20)
[8:35] Zero Linden: as I'm sure it does for all of you
[8:35] Wyn Galbraith: Always a learning session at Zero's.
[8:35] Zha Ewry: Heh. Trying to get a good picture of capabiltiies for other people, for one, Zero
[8:35] Dr Scofield: cu, zero & take care!
[8:35] Tree Kyomoon: hopes we can discuss MoCap streams next time
[8:36] Zero Linden: I think this kind of thought process is very instrucive for how we can plan for a more open future while
[8:36] Zha Ewry: Thanks Zero. Great session as always
[8:36] Morgaine Dinova: Yep, so much hinges on caps that we have to get it right.
[8:36] Neas Bade: yes, thanks much zero
[8:36] Zero Linden: getting the architure right for today
[8:36] Zero Linden: thanks again
[8:36] Zha Ewry: Balance
[8:36] Tree Kyomoon: thanks Zero!
[8:36] Wyn Galbraith: Go rearrange your CDs.
[8:36] Zha Ewry: That's part of te challange
[8:36] Burhop Piccard: Later all. I've got a meeting too. SO you AWG people, watch out for the content creators :-)

User:Zero Linden/Office Hours/2007 Nov 15

Navigation menu

Search