(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
[11:11 AM PT] Ubit Umarov: welcome :)
[11:11 AM PT] Ubit Umarov: So last week code changes..
[11:12 AM PT] Ubit Umarov: i decide to add a fucntion to give color from color temperature
[11:12 AM PT] Ubit Umarov: so added osTemperature2sRGB( temp )
[11:13 AM PT] Gavin.Hird @grid.xmir.org:8002: 2300K = ?
[11:13 AM PT] Ubit Umarov: given temp in kelvin, it returns a very aproximated "blackbody" sRGB coler
[11:13 AM PT] Ubit Umarov: color.. based on D65 white point
[11:14 AM PT] Ubit Umarov: and standard gama of 2.4 ?
[11:14 AM PT] Ubit Umarov: ofc looking for example about LED color temperature we see a mess
[11:15 AM PT] Gavin.Hird @grid.xmir.org:8002: I thought the standard viewer gamma is 2.2
[11:15 AM PT] Ubit Umarov: each using its own idea of whitepoint
[11:16 AM PT] Ubit Umarov: seems 2.4 on math provides a look of 2.2.. whatever
[11:16 AM PT] Ubit Umarov: color seems to have a lot of pseudo science
[11:16 AM PT] Gavin.Hird @grid.xmir.org:8002: const F64 SCREE_GAMMA = 2.2
[11:16 AM PT] Vincent.Sylvester @hg.zetaworlds.com: A science as exact as herding cats
[11:16 AM PT] Gavin.Hird @grid.xmir.org:8002: set in llpngwrapper.cpp
[11:17 AM PT] Gavin.Hird @grid.xmir.org:8002: SCREEN_GAMMA
[11:17 AM PT] Ubit Umarov: in fact makes sense,. since color is in fact a creation of brain
[11:18 AM PT] Ubit Umarov: well that function is just a set of fits to a table of values
[11:18 AM PT] Ubit Umarov: table that is also made from aproximations..
[11:19 AM PT] Ubit Umarov: so well it does return a color from dark orange ( red ) to blue
[11:19 AM PT] Ubit Umarov: with white at abotu 6500K ( D65)
[11:20 AM PT] Ubit Umarov: well not much and not that uesfull :)
[11:21 AM PT] Ubit Umarov: and did waste a lot of time on it .. had to redo several of the fits...
[11:21 AM PT] Ubit Umarov: bc if you google you get lie 3 versions based on fits from same table..
[11:22 AM PT] Ubit Umarov: and wrong on one range or other..
[11:22 AM PT] Gavin.Hird @grid.xmir.org:8002: might be useful to feed RBG values for a modern light bulb given in Kelvin into a loght script
[11:22 AM PT] Ubit Umarov: so i added a 4th :P
[11:22 AM PT] Gavin.Hird @grid.xmir.org:8002: light
[11:22 AM PT] Gavin.Hird @grid.xmir.org:8002: for example
[11:22 AM PT] Ubit Umarov: well problem is that LEDs use other white point
[11:22 AM PT] Joe Magarac: glTF has lights with intensities in real units. Thinking about that yet?
[11:23 AM PT] Ubit Umarov: if you google you see diferent ideas of where white is
[11:23 AM PT] Gavin.Hird @grid.xmir.org:8002: google these days...
[11:23 AM PT] Vincent.Sylvester @hg.zetaworlds.com: Go to a hardware store and pick out two different lamps with same color they look different, not exactly exact in their measurements either
[11:23 AM PT] Ubit Umarov: well not a opensim issue..
[11:24 AM PT] Vincent.Sylvester @hg.zetaworlds.com: Don't go shopping for lamps if you have ocd xD
[11:24 AM PT] Ubit Umarov: this is jsut a little complement to the SRGB/linear convertion funstions
[11:24 AM PT] Ubit Umarov: and conversion ;)
[11:24 AM PT] Gavin.Hird @grid.xmir.org:8002: I am sure it will be useful, so thanks
[11:25 AM PT] Ubit Umarov: well some monitors also have settigns to define the temperature of white
[11:25 AM PT] Gavin.Hird @grid.xmir.org:8002: There is this document Reproducing Real World Light on the net that I have used to get values to make light scripts
[11:26 AM PT] Ubit Umarov: there is also a LL script somewhere, that tries to use deeper approximations
[11:26 AM PT] Vincent.Sylvester @hg.zetaworlds.com: Talk about over-engineering
[11:27 AM PT] Ubit Umarov: like others i used the values from http://www.vendian.org/mncharity/dir3/blackbody/
[11:28 AM PT] Ubit Umarov: the ones from the 10º aproximation, D65
[11:29 AM PT] Ubit Umarov: ann valid ranges are from 1000 to 40000 ofc like on the table
[11:30 AM PT] Ubit Umarov: well those where the code changes :P
[11:30 AM PT] Ubit Umarov: meanwhile i did find a region using z [11:30 AM PT] Kayaker Magic: I gota run, have a lunch meeting soon.
[11:30 AM PT] Ubit Umarov: and.. well .. we do get light.. but.. not good
[11:31 AM PT] Kayaker Magic: Buy all!
[11:31 AM PT] Ubit Umarov: at z == 0 we see normal dark enviroment.. and a set down we get a lot more light
[11:32 AM PT] Ubit Umarov: he did offset the min cameraZ to a higher value
[11:32 AM PT] Ubit Umarov: based on our min z settign for phsycis
[11:32 AM PT] Ubit Umarov: that is not right
[11:33 AM PT] Ubit Umarov: viewer should instead clamp to like z = 0.1
[11:33 AM PT] Ubit Umarov: so light stays coerent as we cross z = 0
[11:34 AM PT] Ubit Umarov: well just my 2cents to you viewer devs :)=
[11:34 AM PT] Gavin.Hird @grid.xmir.org:8002: all right
[11:34 AM PT] Vincent.Sylvester @hg.zetaworlds.com: I don't get why can't the shader be two sided and just do the same below 0 as we do above. Granted that's not great either, but better than this mess
[11:34 AM PT] Ubit Umarov: yhe region is...hg.osgrid.org:80/Marineville
[11:35 AM PT] Ubit Umarov: well that is Marineville here
[11:35 AM PT] Ubit Umarov: anyways the spec is z > 0
[11:35 AM PT] Ubit Umarov: so all should just keep terrain at Z >= 0
[11:36 AM PT] Joe Magarac: Yes.
[11:36 AM PT] Ubit Umarov: our -100 is only for legacy regions
[11:36 AM PT] Ubit Umarov: like that oen
[11:37 AM PT] Joe Magarac: I would like to avoid Z[11:37 AM PT] Ubit Umarov: so instead of adopting that path, just make sure viewer does not blow up on z < 0
[11:37 AM PT] Ubit Umarov: that patch i mean
[11:37 AM PT] Gavin.Hird @grid.xmir.org:8002: if the server finds z < 0, just shift everything up till it is z >0 :-)
[11:37 AM PT] Vincent.Sylvester @hg.zetaworlds.com: Skyboxes at 4000 meters up, fine, deeper oceans to sink a Titanic in, impossible...
[11:37 AM PT] Ubit Umarov: just make it z = 0 or 0.1 etc..
[11:38 AM PT] Ubit Umarov: but not where tim did it
[11:38 AM PT] Ubit Umarov: i caps camera position
[11:38 AM PT] Ubit Umarov: he..
[11:39 AM PT] Joe Magarac: I just tried. Sharpview renders just fine for negative Z values. If you check the "Unlock camera from avatar" box, you can go underground. This is not useful but works.
[11:39 AM PT] Ubit Umarov: cool
[11:39 AM PT] Ubit Umarov: imo at z [11:40 AM PT] Ubit Umarov: so that is a clamp.. but somewhere else
[11:40 AM PT] Gavin.Hird @grid.xmir.org:8002: it could be useful to move the camera lower to peak up at the terrain above
[11:40 AM PT] Joe Magarac: Yeah, I'd clamp cameras to Z>=0.
[11:40 AM PT] Joe Magarac: When you're below one-sided terrain, it dsappears.
[11:40 AM PT] Ubit Umarov: ( possilbe clamp a bit above to it is not that dark
[11:40 AM PT] Vincent.Sylvester @hg.zetaworlds.com: Realistically water level should be 0 and terrain extend down under that. The silly 20 meter standard water leaves very little room for anything. Changing the water height only makes sense if you do it everywhere, even then the shader has a height too so the higher you make the terrain the more you need to adjust the altitude of that
[11:40 AM PT] Gavin.Hird @grid.xmir.org:8002: particularly if you look for buried objects
[11:41 AM PT] Ubit Umarov: vincent region coords are not really as rl ones
[11:41 AM PT] Joe Magarac: Water at 20m is a convention, not a system property. New Babbage in SL has water at 100m and extensive underwater works.
[11:42 AM PT] Vincent.Sylvester @hg.zetaworlds.com: Obviously not, but still. Negative coordinates being such a problem I find a bit odd
[11:42 AM PT] Lyr Lobo: yes
[11:42 AM PT] Ubit Umarov: yeah at sl ones needs to set water level to make a underwater world
[11:42 AM PT] Ubit Umarov: well for this viewers..
[11:42 AM PT] Vincent.Sylvester @hg.zetaworlds.com: Viewers full of hacks, tricking a shader into thinking it is somewhere else hardly the worst offense either
[11:42 AM PT] Joe Magarac: Beq Janus wrote a tool for SL that moves everything in a region vertically. This was used when a new region joined New Babbage and needed a vertical move.
[11:43 AM PT] Ubit Umarov: we do have tools on oar load etc
[11:43 AM PT] Vincent.Sylvester @hg.zetaworlds.com: You can translate an entire scene with console commands
[11:43 AM PT] Joe Magarac: Right. It's easier in OS. You can process an OAR file and add an offset.
[11:43 AM PT] Ubit Umarov: but well..bc this or that ppl refused adn force me to allow -100 :o
[11:44 AM PT] Ubit Umarov: ( i did enforce z >0 for a month or so :P
[11:44 AM PT] Ubit Umarov: in region code...
[11:44 AM PT] Ubit Umarov: well not a ig thing
[11:44 AM PT] Ubit Umarov: just mk sure viewers do not blowup
[11:45 AM PT] Ubit Umarov: and if you can let some light in at z < 0
[11:45 AM PT] Vincent.Sylvester @hg.zetaworlds.com: My problem with that was always more a shader thing with how it showed mesh as if it was full bright. It being dark down there never bothered me, that's expected at those depths, but it should be universally dark
[11:45 AM PT] Ubit Umarov: you can vist that land with fs 74246 to see what tim patch does
[11:46 AM PT] Ubit Umarov: ( ofc we do thank Tim for the try )
[11:47 AM PT] Ubit Umarov: so.. what news do you have? :)
[11:47 AM PT] Ubit Umarov: err about opensim first, please :)
[11:49 AM PT] Vincent.Sylvester @hg.zetaworlds.com: You also changed out Clamp
[11:49 AM PT] Ubit Umarov: ahh yeah mionor change
[11:49 AM PT] Joe Magarac: Brief note on the Sharpview front. I've been working on sit/stand/walk/run/fly, and hope to have a release out in a week. Animatsalt here can do all that now. Also, fixed two major crash bugs, and I think we got the random jaggy triangle bug fixed.
[11:50 AM PT] Gavin.Hird @grid.xmir.org:8002: only thing I've done this week is ugrade the grid to latest dotnet 8
[11:51 AM PT] Ubit Umarov: and that rendering lib? still "dead" developement ?
[11:52 AM PT] Joe Magarac: Rend3? I'm going to have to take that over, looks like.
[11:52 AM PT] Ubit Umarov: ( we had a generic type clamp at Util.cs and a floats one at Utils ( libomv)
[11:52 AM PT] Joe Magarac: It's not a whole renderer. It's just a glue layer that does GPU space allocation and some other stuff.
[11:52 AM PT] Ubit Umarov: now dotnet math as full expansion per all supported types
[11:53 AM PT] Ubit Umarov: with code similar to Utils
[11:54 AM PT] Ubit Umarov: Util one did use x.Compareto(y) that is a lot heavier than simple ifs
[11:54 AM PT] Ubit Umarov: so i removed its use keeping only for any module outthere
[11:55 AM PT] Ubit Umarov: not a big thing... just a few ns here and there
[11:55 AM PT] Ubit Umarov: ahh also the util one could not get inlined by jit
[11:55 AM PT] Ubit Umarov: like the Utils and Math can
[11:58 AM PT] animatsalt resident: Gotta go, next meeting.
[11:58 AM PT] Gavin.Hird @grid.xmir.org:8002: bye
[11:58 AM PT] Ubit Umarov: have fun
[11:59 AM PT] Ubit Umarov: wel we are at the hour also
[11:59 AM PT] Ubit Umarov: any other issue ?
[12:00 PM PT] Ubit Umarov: ok.. so lets end here,,, hope to see you all next week :)
