| --- Log opened Sun Sep 05 00:00:16 2021 |
| 00:29 | <&ToxicFrog> | Argh |
| 00:29 | <&ToxicFrog> | Ok, the quaternion [0 1 1 1] should be identity -- it's a 0 rotation. |
| 00:30 | <&ToxicFrog> | My understanding is that, to rotate a vector by a quaternion, you do: |
| 00:30 | <&ToxicFrog> | V = [0 x y z] |
| 00:30 | <&ToxicFrog> | let R be the rotation quat |
| 00:30 | <&ToxicFrog> | let R' be its inverse (negated x y z, same theta) |
| 00:30 | <&ToxicFrog> | let H be the Hamiltonian product of two quaternions |
| 00:31 | <&ToxicFrog> | rotate(V, R) = H(H(R, V), R') |
| 00:31 | <&ToxicFrog> | And if R=[0 1 1 1] rotate(V, R) = V |
| 00:33 | <&ToxicFrog> | Oh, wait, ok, the vector part of R has to be a unit vector |
| 00:33 | <&ToxicFrog> | So that explains *some* of the weirdness I'm seeing |
| 00:34 | <&ToxicFrog> | But like |
| 00:35 | <&ToxicFrog> | If I have R = [0 0.707 0 0.707] -- that should be a null rotation around a unit vector in the xz plane, right? |
| 00:36 | <&ToxicFrog> | But H(H(R, [0 1 0 0]), R') = [0 0 0 1], i.e. my input vector that used to point x-wards is now pointing z-wards |
| 00:39 | <&ToxicFrog> | And if R = [0 0 1 0] (i.e. null rotation around the y axis) I get my input vector back out with the X and Z elements negated. |
| 00:43 | <&ToxicFrog> | Oh wait shit |
| 00:44 | <&ToxicFrog> | the first element is meant to be cos(theta/2) |
| 00:45 | <&ToxicFrog> | Ok, so {1,0,0,0} is the identity quaternion |
| 00:47 | <&ToxicFrog> | Ok, with this sorted out I think my implementation of qmul and qneg is correct |
| 00:48 | <&ToxicFrog> | but that doesn't explain why my planes are still floating into the air ;.; |
| 01:03 | <&ToxicFrog> | Aha |
| 01:03 | <&ToxicFrog> | Ok, my math is wrong because SupComm quaternions are stored theta last, not theta first |
| 01:08 | | Degi [Degi@Nightstar-5tv1s1.dyn.telefonica.de] has quit [Operation timed out] |
| 01:12 | <&ToxicFrog> | Even correcting for that everything is still wrong though! It's just wrong in different ways. Argh. |
| 01:12 | | Degi [Degi@Nightstar-fa08bu.dyn.telefonica.de] has joined #code |
| 02:52 | <~Vornicus> | Oh, good, you got there |
| 02:52 | <~Vornicus> | sort of |
| 02:54 | <~Vornicus> | the entire quaternion has to be unit |
| 02:55 | <~Vornicus> | it's not just the axis part but the whole thing, the imaginary part is unit axis * sin(theta/2) |
| 04:23 | <&ToxicFrog> | I did in fact get all the way there in the end |
| 04:23 | <&ToxicFrog> | "quat has to be unit" was only relevant for testing, in the end, because the quats I get back from the engine are always unit |
| 04:23 | <&ToxicFrog> | The key insights that got me there in the end were: |
| 04:24 | <&ToxicFrog> | - MohoEngine stores quats as [x y z w] rather than [w x y z], so I have to take that into account when implementing vec3<->quat conversions, qneg, and qmul |
| 04:25 | <&ToxicFrog> | - if I have a unit rotation R and a vector V and I want to translate that vector from unitspace to worldspace, I need R'*V*R rather than R*V*R', because I need to "unrotate" it via the unit's orientation quat |
| 04:26 | <&ToxicFrog> | I still haven't figured out why planes don't bubble when sinking but Oh Well, everything else is working and I've released a new version of `TF_NavalWreckage.scd` for the first time in 14 years |
| 04:28 | <~Vornicus> | nice |
| 04:31 | <&ToxicFrog> | (changes from 2.0: support added for planes, hovercraft, and naval buildings; sinking units bubble and tumble slowly; code almost completely rewritten to be cleaner and easier to extend) |
| 05:03 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has quit [Connection closed] |
| 05:15 | | Vorntastic [uid293981@Nightstar-phvupn.irccloud.com] has joined #code |
| 05:15 | | mode/#code [+qo Vorntastic Vorntastic] by ChanServ |
| 06:43 | | Kindamoody[zZz] is now known as Kindamoody |
| 07:07 | | catalyst_ [catalyst@Nightstar-ejd4sd.cable.virginm.net] has joined #code |
| 07:07 | | catalyst [catalyst@Nightstar-ejd4sd.cable.virginm.net] has quit [Connection closed] |
| 07:38 | | abudhabi_ [abudhabi@Nightstar-ghfs0l.adsl.tpnet.pl] has joined #code |
| 07:40 | | abudhabi__ [abudhabi@Nightstar-edpsl2.adsl.tpnet.pl] has quit [Operation timed out] |
| 07:41 | | abudhabi [abudhabi@Nightstar-edpsl2.adsl.tpnet.pl] has quit [Ping timeout: 121 seconds] |
| 07:42 | | abudhabi [abudhabi@Nightstar-ghfs0l.adsl.tpnet.pl] has joined #code |
| 09:38 | | Reiver [quassel@Nightstar-avheq5.zen.co.uk] has quit [[NS] Quit: Reblooting.] |
| 09:39 | | Reiver [quassel@Nightstar-ksqup0.co.uk] has joined #code |
| 09:39 | | mode/#code [+ao Reiver Reiver] by ChanServ |
| 11:05 | | Emmy [Emmy@Nightstar-l49opt.fixed.kpn.net] has joined #code |
| 11:16 | | catalyst [catalyst@Nightstar-ejd4sd.cable.virginm.net] has joined #code |
| 11:16 | | catalyst_ [catalyst@Nightstar-ejd4sd.cable.virginm.net] has quit [Connection reset by peer] |
| 13:50 | | Kindamoody is now known as Kindamoody|out |
| 13:50 | | gnolam__ [quassel@Nightstar-j9ajs0.priv.bahnhof.se] has joined #code |
| 13:54 | | gnolam_ [quassel@Nightstar-j9ajs0.priv.bahnhof.se] has quit [Ping timeout: 121 seconds] |
| 15:09 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has joined #code |
| 15:09 | | mode/#code [+qo Vornicus Vornicus] by ChanServ |
| 16:03 | | gnolam__ [quassel@Nightstar-j9ajs0.priv.bahnhof.se] has quit [Ping timeout: 121 seconds] |
| 17:15 | | Vorntastic [uid293981@Nightstar-phvupn.irccloud.com] has quit [[NS] Quit: Connection closed for inactivity] |
| 20:05 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has quit [Connection closed] |
| 21:19 | | Kindamoody|out is now known as Kindamoody |
| 22:41 | | Kindamoody is now known as Kindamoody[zZz] |
| 23:24 | | Emmy [Emmy@Nightstar-l49opt.fixed.kpn.net] has quit [Ping timeout: 121 seconds] |
| 23:59 | < [RMS]> | Wait, they renamed CIFS back to SMB again? |
| --- Log closed Mon Sep 06 00:00:17 2021 |