These are chat archives for highfidelity/hifi

8th
Jul 2014
mpursley
@mpursley
Jul 08 2014 02:49
Hey All, I'm clicking on "Recover Password" in "Interface"... and it takes me to https://data.highfidelity.io/password...
But, when I put my email into https://data.highfidelity.io/password, it says "• Email not found"
Clément Brisset
@Atlante45
Jul 08 2014 02:50
Did you get an alpha invite?
mpursley
@mpursley
Jul 08 2014 02:52
Yes... well, maybe?
"High Fidelity Alpha: confirm your email, please"
"You're confirmed. You will be notified about the High Fidelity Alpha once we're ready. Thank you!"
Clément Brisset
@Atlante45
Jul 08 2014 02:56
@mpursley sent you a PM
mpursley
@mpursley
Jul 08 2014 02:58
Awesome, thanks! :)
Philip Rosedale
@PhilipRosedale
Jul 08 2014 04:03
@judasshuffle We decided to 1/2 the audio frequency and go with uncompressed signal. Humans above teenage years cannot hear about 12Khz (which is the limit at a 24,000 samples/sec rate), so we thought this was a very good idea - bring the raw sampling rate down but do no compression.
Judas
@judasshuffle
Jul 08 2014 06:24
@PhilipRosedale so beter than cd quality sound, works for me.Perhaps you should name the project to reflect this :) suggests -"Sounds goodity.wee oo"
Judas
@judasshuffle
Jul 08 2014 12:56
no mi maths is backwards that's a bit down from cd. Would compressing the audio with FLAC add a ton of delay?
Philip Rosedale
@PhilipRosedale
Jul 08 2014 15:17
@judasshuffle Yes that is something to consider... but... the trick is that we need to be most sensitive to the packet rate, to minimize latency. We are using packets containing 256 samples or 10.6 milliseconds of audio. Since there is one sample for each ear, and each sample is 16 bits, this means that we need 1024 bytes of data to send it (uncompressed). If we used FLAC to allow us to set the sampling rate to 48Khz, we'd need to guarantee that we could get 50% compression (2X) on each packet, to avoid making extra packets.
Thijs Wenker
@thoys
Jul 08 2014 15:18
@PhilipRosedale would it be something to make the encoding optional, for bad connections?
Philip Rosedale
@PhilipRosedale
Jul 08 2014 15:18
In many cases, the FLAC (or any other lossless) codec will fail to achieve that 50% compression (actually we do have a bit of overhead since a UDP packet can reliably be about 1500 bytes long, but longer packets are more frequently lost).
The other consideration is that of course it adds encode/decode CPU time to process those packets.
And... what we have been thinking is that ultimately what we probably want to do is actually transform the packets into the frequency domain at capture, so that we can more easily process things like echo/reverb and attentuaton of frequencies at distance. Which is much easier once in the frequency domain.
Judas
@judasshuffle
Jul 08 2014 15:20
As much as id love 24 bit 192, i know that's not going to happen 16/24 is more than good enough for what we need and compared to a telephone is night and day
Philip Rosedale
@PhilipRosedale
Jul 08 2014 15:21
An interesting question is whether the FLAC codec does a FFT (or MDCT) transform into the frequency domain as part of its processing.
What we ideally need (and haven't done yet) is something like that... a lossless and very fast transform into frequency space that could be done at the client on encode and decode.
Philip Rosedale
@PhilipRosedale
Jul 08 2014 18:14
Are our javascript error messages now going somewhere other than the log screen? @huffman Were you working on that? I've got a print() statement that failed because of an undefined variable and I got no output to the log to alert me.
Thijs Wenker
@thoys
Jul 08 2014 18:53
FYI, RakNet is opensourced by OculusVR https://github.com/OculusVR/RakNet
Ryan Huffman
@huffman
Jul 08 2014 19:15
@PhilipRosedale I haven't done anything related to that, although @ctrlaltdavid may have recently?
Thijs Wenker
@thoys
Jul 08 2014 19:21
omg, we discovered something funny, mindcity's normal gunscript from the examples makes my frisbees work like bullets
i can destroy voxels with them
Chris Collins
@cozza13
Jul 08 2014 19:22
tron
Discs of Tron
Thijs Wenker
@thoys
Jul 08 2014 19:24
have a look when u have some time
Chris Collins
@cozza13
Jul 08 2014 19:24
can you give me the script
Thijs Wenker
@thoys
Jul 08 2014 19:25
well it works with any particle script, try the ball script throw a ball at something while someone else has the gun script activated
Chris Collins
@cozza13
Jul 08 2014 19:27
Ohh. If I have the gun script, anyone near me using the ball script will now be able to destroy voxels?
Thijs Wenker
@thoys
Jul 08 2014 19:27
yep
dangerous right?
David Rowe
@ctrlaltdavid
Jul 08 2014 20:39
@PhilipRosedale I'll have a look at the script error output ... seems to ring a bell
David Rowe
@ctrlaltdavid
Jul 08 2014 20:57
@PhilipRosedale print(somevar); ... on Windows ... error is reported to: 1) Interface's Command window (the DOS-style one that launches the GUI); 2) the program log file; and 3) the Script Editor's debug log pane (if using the Script Editor) ... is this not what you see?
Philip Rosedale
@PhilipRosedale
Jul 08 2014 21:47
@ctrlaltdavid I just added exactly that... print(somevar); to my keypress detection code.... Controller.keyPressEvent.connect(function(event) {
if(event.text == "SHIFT") {
print(somevar);
And I get nothing in either the application log screen OR the xCode program output window (when using the dev build). Latest versions of all.
It looks like if I put that print command at global scope... at the start of the file... I get the logged error as expected. But when I use it inside a function as declared above, I get nothing.
David Rowe
@ctrlaltdavid
Jul 08 2014 22:05
@PhilipRosedale Thanks ... I can reproduce what you're (not) seeing ... will investigate further
David Rowe
@ctrlaltdavid
Jul 08 2014 23:13
@PhilipRosedale It seems to have been this way for quite some time (i.e., before I got my sticky hands on the code) ... for some reason the Script.update event is eating JavaScript error messages from other event functions ... i.e., comment out the Script.update.connect() and errors are reported.
Will see about fixing it
Philip Rosedale
@PhilipRosedale
Jul 08 2014 23:32
@ctrlaltdavid thanks!!
David Rowe
@ctrlaltdavid
Jul 08 2014 23:36
@PhilipRosedale Pull request submitted : )
Philip Rosedale
@PhilipRosedale
Jul 08 2014 23:56
@ctrlaltdavid Very cool, thank you so much!!