#virtual-reality
1 messages ยท Page 45 of 1
it's the answer for shared computation
unless you are doing a mmo
i still think not
becouse the costs increase for you as a dev
you get less profit
unless it's built into the framework
we talked about doing cloud stuff for realtime rendering to mobile devices at the unity conference
well, for a case like that, maybe
even just-in-time stuff is handy
if someone makes a new multiplayer map, why NOT upload it to a datacenter, train the AI on it ridiculously fast and send it back?
you'd have most of the work done already
server costs are getting cheaper and cheaper
well, but then the guy who created that c an just let it calculate for a day on his pc
could do
but it's like building lighting
one small change, needs to be redone
my hope is that my lightmass replacement will be hooked into our cloud services too
until lots of players like that and spam you
get your level lighting back instantly
nah, just charge a microtransaction fee
or throw it into a queue
yup
if you dont pay you calculate it yourself
i still think cloud is just too limited in general for stuff
latency, the fact that you have to upload stuff
that kind of thing
Crackdown 3 is going to use cloud for destruction
and it has some absolutely incredible destruction
neat
Can you use cloud for real time light mapping out in the distance?
no
titanfall 1 used microserviced VMs to spool up more computing power for intense multiplayer games
i.e. reduce the number of dynamic lights
just being able to divide work by design is amazing
cloud computing would be a reasonable solution for agent pathfinding
it's not once the hardware is paid for
we have a data center filled with the last gen of titans
sweet
since you only need to really know the start points and path points for each agent, network traffic is pretty minimal, and all the processing can be done elsewhere
you mining bitcoins on that?
Ooh. Recycle GPUs into data centers.
the real issue is obviously cost, hehe
it cost a lot to set up but once you're in there it's not so bad I imagine
energy costs
and let's face it, the gaming entertainment industry is worth far more than even the movie industry
money is not a problem
for the right people ๐
a hundred players connected to a datacenter can easily have the datacenter on 24/7
calculating light
that can get expensive
the main issue datacenters have now is energy
and heat
as for bitcoins, we had 300 million invested into us a few months ago by one of the media companies. it's in one of our press releases.
Yeah.
we're all about the graphics baby
becsoue you can just buy MOAR cores
but how do you give energy to that?
some of the biggest supercomputers have a DEDICATED reactor
nice
i think china has a huge one that is powered by a damn nuclear plant
that's awesome
Is the gaming industry worth more than the movie industry? Or are gaming receipts more than theatre receipts? Because those are very different things ๐
small improvements in datacenters are big money
like moving to alaska for the cold air
thats why people is doing less energy
or kentucky for cheap power
its better to buy moar cores that use less energy
yup
running costs are not always predictable
yes, it is not as powerful as x86
but you can buy a bunch of them
and set them on parallel
GPUs are also very good at energy
So data centers are moving towards a neural net model of processing?
they can do a shit ton of math on low energy
Lots and lots of small cores
I assume you've deleted the message?
Because I don't know what you're talking about specifically ๐
Cool
Yay! My first bit of insider information. And I have no clue what it was! ๐
Probably not
Haha... I'm a believer of the cloud!
In sharp contrast to a lot of gamers that I used to hang around.
But it's good to get an idea of its limitations and how it can and can't (that well) be used
what did it do?
that's the main reason they couldn't go to SA. inflexible data centers.
ok so imagine you're in a heated battle and suddenly everyone gets mechs at once and leaves them on auto, plus a fuckton of grunts come in and all the fighting focuses on one point of the map
either you need to have accounted for that and you dedicated a lot of resources to each game, or you segregate out the AI by cluster into a new VM that runs in parallel
titanfall's servers would dynamically spin up more resources and offload to them in order to hit performance targets
it made it cheaper to run
Ah yeah
so say you're at capacity and all the titans drop along with some grunts. the grunts get shifted off as they're fairly low priority in this narrative
I watched an Amazon data center about dynamic load distro that's similar to what you're describing a long time ago
but the grunts continue to do good AI for the one noob who is still fighting them
yeah
it's just nobody had bothered to do it for multiplayer game AI because it's hugely complicated
Basically instead of renting computing per machine
it's rented per computing unit cycle
basically yeah
it was all on azure
azure's very good at creating new instances to the specs you need instantly
a lot of VMs are core-bound
azure is not
it's a stop-gap measure
it may never happen again
it'll be useful for all sorts of non-game related things though
I dunno if it's the WHOLE data center, but in that rack yes
Fair call.
don't have any more details other than that
So titan 2 not using Azure?
Fair enough.
it's still a strongly-AI-driven MP experience
Hmmm. You mentioned it'd be awesome in a MMO?
You're envisioning like a dynamic battlefield with a lot of different AI behaviour states?
Sounds like it'd make a great marketing bullet point. Wonder if players would actually end up liking it.
In the same way that players like the idea of 'good AI', but don't actually like been challenged that much ๐
original wow servers were REALLY hardware bound
the game was optimised for that exact hardware
then as things got cheaper they moved to VMs, now they merged servers because they have so much free computing power they want higher pop servers
good AI is challenging until the end then loses for you
in a hardcore game sometimes it wins
just like a WoW fight, yup
what's fun is learning how to beat something then implementing it
then doing it in different settings and combinations
if every time you fought an enemy you had to beat a unique puzzle you'd never grind
But WoW AI is largely scripted isn't it?
Just straight up phases that you memorize
yeah and zones of attacks
Classic gaming stuff.
the flip side to that is the tools you have to engage that fight are so incredibly complex and the teamwork required is so complex that scripted is OK
it's HARD to get people to do the right thing all together
Yeah, the challenge is coordination.
that's why you have things like players becoming AOE bombs
they already know exactly how to fight
now they need to know what to do to win THIS fight using what they already have
The last time I heard gaming hype about AI in a single player game was Half Life 2.
Revolutionary at the time. Like the use of bones to animate meshes
nah that'd been around since quake 2
hahaha
yeah half life 1 was quite cool
having enemies with ANY smarts was great
rather than just pathing straight at you and shooting all the time
haha
that's why I've been saying that AI needs to be offloaded to a neural network
with a massive toolkit of animations
What about instead of animations
I think you can do SOME procedural animation, but not a lot
nah dunno about that
that's why you use a neural net
To train it
Wait... I've seen evolutionary animation systems. It's not bad.
I probably wouldn't bother training animation
But not quite natural either.
I'd use it to vary the animation instead
the witcher 3 has so much dialogue that they used a procedural system to lay the base animations for a scene down
then artists came in and tweaked it
those kinds of things work better
so in my system I'd use a neural network to figure out lots of reasonable parameters for variation of animation and bone movements
I'd use the NN to generate them and then parametize them
then I'd do a separate set of animations based on stance and bodylanguage
then maybe use the NN to do genetic weighting and I'd have the final say on what looked good
so that you're not doing work, you're just reviewing what it made
Ok. So kinda like having this particular movement - but then increasing the range in these areas (speed, angle, etc) suggests a more 'aggressive' state
Or drunken - staggering around, less control over movement
take all of those animations and weight parameters and ranges into UE and then hook that up to a neural network that has been designed to create busywork and stuff
hook in feelings to activities
yeah exactly
each thing on its own is boring
but using an NN to bind them together in the right way saves you a lot of work and gives you a complex model as a result
Yeah
oh it turns out someone HAS done the audio NN we were talking abuot one night
a thing called WaveNet
you can download it
it's called parametric TTS
Were we talking about audio NN?
Haha... I remember talking to you about NN before for sure.
neural networks are fantastic for iteratively testing things with minimal human oversight
so if you want a wide variety of animations and to figure out how much they can deviate and still look real, you let a network do the work for you
I need to stop talking about it and get into it
So you work on NN stuff at work?
no
Or just read up a lot about it
I'm just a script monkey
and are excited about it
Fair call.
I'm not like you guys with your trust funds and savings who can quit your jobs to do UE ๐
๐ I'd love to be a trust fund baby.
so many places to use a neural network. you could take the transition between every animation in your library and use the neural network to generate iterations that you say "more" or "less"
by the end you'd barely need to tell it to do anything
but you'd end up with nice, realistic interim animations
you just have to give it some ideas about what to do
The real trick will be to get NNs talking to each other.
Kinda like how the human brain does it...
One recognizes human like poses and feeds that to the system been trained to create animations.
there's probably some sane limit, but you could use a network to figure out the best waiting between two existing NN indexes
or you could use the data from two networks to train an amalgamation
Yeah. Pretty much!
one NN could be "this is how you physically do things like walk across the room and get a cup" and the other network is "this is how animations look good"
since the animation is fairly additive you just procedurally combine the two
or you have a third NN that just takes animations and takes objectives and smoothly integrates them somehow
I'm having trouble describing it
Well, they're not easy things to describe ๐
TTS?
ah ok
This post presents WaveNet, a deep generative model of raw audio waveforms. We show that WaveNets are able to generate speech which mimics any human voice and which sounds more natural than the best existing Text-to-Speech systems, reducing the gap with human performance by over 50%. We also demonstrate that the same network can be used to synthesize other audio signals such as music, and present some striking samples of automatically generated piano pieces.
text to speech!
hehe
really good TTS
I want to be able to train it with a voice actor and have it turn my voice into the dialogue I need
Yeah, we were suggesting that we could generate new lines of spoken speech from Audio NN.
using the other voice
(did the text to speech work in Discord, or does nobody have headphones in xD?)
... and of course after that conversation, Adobe came out with their thing
you could do the same thing with visuals and newsreaders would never need to stop smoking or get makeup again
I used the TTS command on that last line
did you see that 'audio photoshop' that they showed off?
adobe's thing is rubbish
it's word boundary identification
designers are easily wowed
๐
I have TTS off ambershee ๐
me too, lol
but otherwise you spammed 584 people with it
but yes, the Adobe thing was really not that impressive, IMHO
at least some of them sleep with their computers in their rooms with the sound on
But it's not just straight up boundary and splicing. It's doing a pretty good job inteporlating between different words and phonemes
A lot of the speech sounds plausibly spoken by the person.
vocaloids have been doing plausible spoken word for years
the whole point of wavenet is it could change inflection
not just word order
it could make it angry, terse, happy
without the actor needing to re-record dialogue you could change it as much as what you needed, even in-game to suit the mood
as they point out, it occasionally throws in breaths and things that didn't exist in the original dialogue
this is like the skyrim face-creator of voices
there's probably room to be the mixamo of it too
take otherwise unconnected technologies and build a company out of it and wait for adobe to buy you
Heh ๐
Fair enough. Didn't realize NZ was that far ahead
in the time zone
night
8gmt here
West
that's quite the gap
We're in the butthole of the planet.
Not a bad place to be... but as far as trade goes, Perth is the ass!
Lol
nite
@hard light https://www.reddit.com/r/Vive/comments/5cdy6p/some_informations_about_vive_wireless_adapter/d9vqssy/
http://www.tpcast.cn/htcvive.html Of note (through Google Translate): * Video format: HDMI2.0 * Video Resolution: 2k (2160 * 1200) *...
Propagation delay <1 ms
Bloody-hell if true.
impossible
shrug I'm not the one making the claims ๐
their own website claims 15ms
It's lies
which sounds way better
(well, way more accurate)
'propogation delay' is likely the hardware-hardware latency between headset and peripheral
But nothing about wireless sender to receiver
Valve's own tests with Netero put latency at 8ms
Have you heard of Netero?
nope
@pearl tangle https://www.facebook.com/help/1484407385200471?helpref=related and https://code.facebook.com/posts/1126354007399553/next-generation-video-encoding-techniques-for-360-video-and-vr/
To watch a 360 video on Facebook in VR you'll need a Samsung Gear VR headset, a compatible Samsung Galaxy mobile phone and the latest version of the...
.1 ms to 15 ms is going to be noticeable
i passed through perth years ago, nothing there
cant believe it used to be scotlands capital
Perth was never the capital?
Scottish Perth is not the Perth I live in ๐
it was a royal burgh at one point in time
It's funny to remember that a lot of anglo cities are named after old world ones.
Colonies tend to be named for the place people departed from
They stick a 'new' in front of it.
Why's this Perth not New Perth
Seems like the answer is; because whoever founded this place didn't feel like sticking the new in front of it ๐
I mean... that's kind of an obvious answer in retrospect.
Is that Vive adapter using 802.11ad? I saw someone mention it running at 60Ghz but didnt see a source.
probably wireless HDMI
I wonder what effect it will have on human's brain
Anyone else with the touch find the haptics unreliable? Like sometimes they don't fire?
specifically using the waveform type
I find that in UE4 using pretty much any controller
vibration / haptics, all totally unreliable
@mighty carbon That's an article from one dude from '98. Substantial research into wireless radiation has since been undertaken
and pretty much revealed that it's simply not at the level that can cause cellular harm.
if wireless radiation was even potentially dangerous, our brains would all be mulch by now
Much less the level that might be actually dangerous
the sheer volume of wireless signals you're subjected to on a daily basis is enormous
Our forcefeedback on the vive was 100%, but the touch doesn't support force feedback so I'm converting everything to haptics
So... you're been bathed in visible radiation all day everydayyyyyy
In the most literal sense possible!
@clever sky Read until the very end
@mighty carbon I can only see a single page.
last page is from 2016
Anyway, your best bet is to be skeptical of individuals crafting theories about this stuff - even if you don't understand the science behind it. Not necessarily to reject what they're saying outright - but simply wait for more information and dissemination of information before paying it much mind.
Because there's plenty of motivation for people to buck conventional wisdom and appeal to the fears and ignorance of many.
so, the less we exposed to EMF, then less chance of developing cancer we have
But the question is to what extent? Is the risk reduction worth the effort?
Like... if we spent X hundred billion, we could create an asteroid shield that could block 10% of potential asteroids. Is that worth the effort?
I am sure 60Ghz WiFi is not best thing to put on your head
But is it worse than a banana?
And here I thought you read the article
... I can't because it doesn't load?
No thanks. I'm just gonna sleep and not worry too much about this stuff ๐
here at work no PDF can be loaded in Chrome, so it automatically downloads
Like have a look at the chart I linked. thoroughly researched, using many sources of information.
And it's quite interesting what amount of radiation normal things produce.
Like... the sunlight.
Or a flight from one city to another.
this is not a chart. It's a paper with sources that describes how EMF causes DNA damage and so on
sure, but does it talk about how much exposure and to what kind of EMF is required to actually do any damage worth caring about?
Sunlight causes skin cancer, if you are exposed too much to it
Yes, that's right.
right, @hard light , there is no saying how much exposure is dangerous, but common sense should tell people the least you exposed to it, the better.
But we don't fear sunlight. We simply take reasonable precautions.
but come on, using phones is one thing. Strapping 60Ghz device to the head is another.
did you know tomatoes are poisonous?
And sunlight is much more substantial as far as ionizing radiation goes then most things.
you still eat them though
Why does it matter if it's 60GHz?
because the damage is so minor that it just doesn't matter
Rather than say... the normal 2.4GHz?
and yeah, the fact it's 60GHz doesn't really matter?
I get that it might seem like a bigger scarier number... but that's not the component that actually matters in terms of damage done.
What matters is total energy transmission.
sounds like you are one of those people who believe doing drugs in small dozes is ok too
lol
recreational drugs*
...
do you drink coffee or coke?
nope
but poisons are only poisons in the correct dosage.
you can poison yourself with water
can you?
You can also consume cyanide and be absolutely fine.
pure oxygen will kill you
neat
Oxygen toxicity is a condition resulting from the harmful effects of breathing molecular oxygen (O
2) at increased partial pressures. It is also known as oxygen toxicity syndrome, oxygen intoxication, and oxygen poisoning. Historically, the central nervous system condition was called the Paul Bert effect, and the pulmonary condition the Lorrain Smith effect, after the researchers who pioneered its discovery and description in the late 19th century. Severe cases can result in cell damage and death, with effe...
youd probably feel amazing for a few breaths though
Infact, oxygen was the original destroyer of worlds.
Until life forms adapted to start using it.
but yeah, you probably consume cyanide on a fairly regular basis
Well, the apple seeds that are famous for Cyanide.
the reality is that the dose is so small, it doesn't matter
You'd have to consume something like... 2 truck load of apples in a single day.
i.e. enough so that the other things about the apple will kill you well before that point.
you get cyanide in lots of nuts (particularly almonds), and also leafy greens like spinach, cassava and some beans
I think we just blew Motorsep's mind here.
I need to print that radiation chart.
My mum's like... "Get away from that (LCD) monitor! You'll get sick from radiation!"
Side note. Oxidation is one of the biggest causes of cancer. Also fires are caused by rapid oxidative action. Oxygen is... the catalyst for releasing energy.
But the release of energy has the tendency to release... ionizing radiation (free radicals)!
So when you exercise... it's a balance between oxidative release and improving overall physiological fitness that reduces other forms of damages.
Like the build up of visceral fat that impedes the function and flow of blood vessels.
no, I just have work to do and don't have time for pointless discussions with folks who are so deep into tech that they don't care for long term health issues
so the only thing that can blow my mind is 60Ghz strapped to my head
๐
Serious question motorsep. Have you ever admitted to been wrong?
I care enough about my long term health that I actually bothered to read up on it.
At least, in the last 10 years?
I'm simply curious as to what sort of circumstances would get you to rethink your previous beliefs.
On any subject matter... not necessarily about this issue in particular.
I always believed that we have a way too much EMF surrounding us, which might affect our health. Some folks with good genetics and heredity might never get affected. While others would be.
you realise, that the only reason you're alive, is because you're permanently surrounded by an EM Field, right?
The paper I linked does say people will get DNA damage in 100% of cases, or cancer in humans directly linked to EMF exposure. However, it shows direct link between EMF and cancer in mice, and describes researched mechanism of DNA damage by EMF. It's not a chart someone pulled out of their ass.
That being said, you can interpret the data however you please
yes, but here's the kicker - it's talking about exposure WAY above normal exposures
it's like the apple comparison again
if you eat enough apples, you can get cyanide poisoning - but eating an apple every day does not mean you will get cyanide poisoning
in the same manner, exposure to sunlight can give you skin cancer
but being exposed to the sun every day is perfectly normal
apples don't cause DNA damage
if you get poisoned with apples, you are going to recover
DNA damage is irreversible
oh, you also forget that apples have arsenic (at least in USA)
๐
and speaking of exposure amount, it's different for everyone
DNA repair is a collection of processes by which a cell identifies and corrects damage to the DNA molecules that encode its genome. In human cells, both normal metabolic activities and environmental factors such as radiation can cause DNA damage, resulting in as many as 1 million individual molecular lesions per cell per day. Many of these lesions cause structural damage to the DNA molecule and can alter or eliminate the cell's ability to transcribe the gene that the affected DNA encodes. Other lesions indu...
"In human cells, both normal metabolic activities and environmental factors such as radiation can cause DNA damage, resulting in as many as 1 million individual molecular lesions per cell per day"
...and yet we live on.
but long story short, DNA damage is generally reversible.
(and damage that isn't is generally ignorable / harmless)
also, if you really want the good news - cancer is inevitable; if something else doesn't kill you first
Well actually.
There's a root cause mechanism common to all cancers.
Which is that they don't die.
Their apoptosis (suicide switch) is turned off.
Allowing them to propagate at a rate greater than healthy or near healthy cells that do die off
cell lines
not individual cells.
everyone will develop cancer in their lifetime should they survive long enough; it's a genetic defect and completely unavoidable
So. If we could find a method to target that differential or to ensure apoptosis... we could theoretically cure cancer.
It's just not something achievable at our current level of technology.
though yes, if you could target the mechanisms by which it develops, I guess you could stop it
I mean, other forms of death prior to this one were much more likely, and we found ways to push that stuff back ๐
And so we must treat all causes of death - as things that have causes and potential solutions.
Anyway, night.
Headed to bed. fun chat.
I am good with wired desktop VR for now ๐ As tech develops and there is no negative effects reported (that's if this thing goes into becoming consumer product), I'll switch with that thing. I bet FCC won't let it pass to US market ๐
What is the best way to record VR gameplay to put into a game? Taking screenshots is easy enough, but video?
I use Shadowplay
works super well for this kind of thing
upload straight to youtube from SP, pretty barebones, you have to edit once it's uploaded, etc. but it's overall solid and doesn't impact performance
you also get the files in Videos folder
at the risk fo asking the lion's den
how happy is everyone for UE for VR vs. Unity?
i'm pretty happy with UE, I was potentially going to give unity a run.
UE4: More potential for CPU side optimization due to C++, higher graphics, inbuilt shaders and tools better than unity. the vr is also abstracted, so same code works on all headsets
it seems like 4.14 makes things a lot better.
Unity: Way more bare-bones, wich means it runs faster by default
no need to disable literally everything you can
the component system is good for fast coding, but performance might suffer
tools are nowhere near UE4 ones
but you have the asset store
btw, ive heard from devs that unity for psvr is a disaster
yeah, I mostly do C++ right now both professionally and with UE4.
oh wow. that's unfortunate.
I just got my vive so I'm expanding my horizons a bit.
the biggest complaints I've seen with UE4 + VR was around forward rendering.
wich is now better on 4.14
Ue4 forward is WAY more heavy thajn unity forward
but unity forward has huge problems with lights
ue4 forward is lots more flexible
and much better graphics
yeah, mostly I'm comfortable with the whole Unreal paradigm. its nice though I don't have to switch off for VR. i wouldn't mind just grabbing my SDKs and doing my own thing at some point, but this is such a good start.
check the upcoming high graphics vr games
psvr and Oculust Touch
most are unreal
most indies use unity becouse they were using unity before
sure. that's fair.
i'll probably check out blueprint at some point too, but I'm pretty happy with just C++ with resharper and visual assist.
what an awesome email to wake up to! got one of the remaining vives
it is here silly ๐
shhh already am
i've been a bit hesitant to move forward without a headset, i found myself making too many gameplay decisions based on gamepad
but it's a good opportunity to write my own movement component etc from scratch
that way i can just pay someone at a later time to implement all the prediction stuff, i can do it but at the end of the day it would have holes in it and i want to make a quality game, that and audio/music are the two areas i can't get the level of result that i want on my own
@wicked oak ue4 forward is way more heavy than ue4 forward? why?
and how can you say unity has problems with lights while ue4 doesn't support dynamic movable shadows?
ue4's forward is in it's infancy
so not supporting dynamic movable shadows is only a problem if they're never going to add it
its also a problem if it takes like half a year to get added
who designs these things?
why would you put a high power battery one someones head when you could put it on a belt clip and be safer and 2x bigger battery size.
throw in a couple of holsters for the touch controls
that would be dope as a rope
given how heavy batteries are I suspect the battery isn't even on the HMD
the thing on top is a transmitter
the thing on the back is the rest of the hardware
I believe in the interview they gave, it was self-contained on the HMD
to be honest I'd like to see HTC put out something like the GearVr except it works with lighthouses and motion controllers. not sure of the viability though
ah
bet someone mods it to be further down fast
as well, I saw
I can't imagine why anyone would want to add more mass to the HMD
hopefully vive2 will move more to a belt pack and have the headset be more like big wraparound glasses
not a box stuck to your face
most of the box is just the screen
the hardware is negligible
screen advances will have multi-density pixels on curved surfaces that wrap around your eyeballs
they'll be able to be smaller, closer and more in focus
no lenses
that's when they'll become sunglasses
and at that point they'll go wireless too
the problem right now isn't the screen and ability to do curvature, we've got that down-pat
plus there's already existing battery advances that will let you have the same battery as this addon does but at a quarter of the size
it's really in the manufacturing of the optics
lol
with just a screen you could flex it to better accommodate each person's vision requirements too
can't do that with a rigid lens
wait no what you were serious?
of course
how else are they going to be smaller?
it's a long way off though, I wouldn't even guess
not the way we're doing displays currently
we're getting much closer
they weren't doing flexible see-through displays just so they can do cutesy round phone screens
you need to defocus the light out as if it were 1m away, your lens in your eye can't collimate anything closer than a couple of inches
that as a problem on its own is easily solved. when you're also trying to be a big fuckoff lens in general that gets harder and your solution narrows
yeah
this is how magic leap works
so it's on its way
I imagine the popularity of vive and existing HMD designs has pushed the goalposts out for them a bit
yes
how do you know this about Magic Leap
and why aren't you getting slapped the fuck up by some NDA
I got a bit of insider info a couple of years back that then got released though, I felt so special
because they released that themselves
cool.
beyond that we'll have to keep working on resolution and transmission
all I can say is I wish that I didn't sign any NDAs, and rather heard things from these kinds of insider spies
but I imagine a hybrid solution with some computing power worn on the user will be good
eh, NDAs are fine
it's like I can't talk about half of a good future solution because it's commercially sensitive
you can mentally fill the gaps
nods unconspicuously
hehe
everyone seen this?
โค team17
@winged shale Mass isn't the be all and end all of HMD comfort. Balance and weight distro is very important.
I mean motorcycle helmets are much heavier, but no one complains about how heavy those things are ๐
mass affects how fast you can turn your head though
It's the angular momentum you gotta be careful of with mass. If you distro it so it sits on top of your spine, it'll feel a lot more comfortable.
but it's coming. just stay alive for four or five more years and you'll be in VR mecca
what matters is the inertia tensor
As opposed to out on the front of your face
where the mass will torque everything a lot more.
gaze interaction is OK
Well, it's ok in small amounts. Just as a main UI functionality is a no no.
not for every UI, maybe for a start-menu, or peripheral thing
as long as you have gaze tracking then cool
but if it's just pointing the head a la GearVR
๐
Or things like NPC interaction - using the look at dot product to see if the player is looking at an NPC. That's fine.
also yay Team17 doing something non-worms-rehashed
yeah obviously, just anything involving gaze that you wouldn't do in your daily life to do stuff
Kills me when people suggest that gaze with eye tracking might become a primary interaction method.
I think I was reading about some people doing research into that area proclaiming that it was aweesssomme.
research and design
What's Lethal?
there is, however, information in the gaze microsaccades that can be useful
VR thing? Or just a T17 thing?
both
Link?
it's what Eyefluence is talking about: "transforming intent into action"
Antidamage linked it some 20 posts ago
Nice
yeah, looks like the features of an early access game in full-release
imo
cute though, maybe it's really fun
I'll have to try it
Could be worthwhile if you're looking for a polished shooting gallery.
But I think I've had my fill ๐
reading the blog
I think the issue with the last video was no engagement, besides my heavy breathing
so I've hybridized it into a blog/vlog
Sounds pretty bad ass (the whole momentum preservation thing)
Like that you fixed the moving platform issue this way...
Something I haven't dealt with, with my own grabby stuff ๐
bows deeply
I was busy making it so it could procedurally detect climbable surfaces
based on the angle of the hand hold
Haha ๐
I'm trying to make all the systems in my game emergent
so simulating everything I can without making players feel uneasy
allows for emergent awesomeness in gameplay
like how in the end of the video I kind of just float out into space because I misjudged a push
Ah yeah.
That's right... you want to create those sorts of possibilities for creative players
so they can show everyone else 'what you can do'
While still having enough structure to the real game to pull your standard player along.
or how the systems end up just working like they would on the ISS, like with letting go after grabbing on to a moving/rotating item, how you keep going as if nothing had happened, except for centripetal tangents
so really players could practically set up an artificial gravity with a spinning chamber
and yeah that's the idea
yeah, just thinking about that, I should have a standing mechanic
for when there's sufficient acceleration dot head orientation
Ah yeah, that'd be a good idea.
Move between space station with art grav and space ship that generally lacks that
yeah, like Interstellar
Or have like a spinning cylinder in a cap ship where the rotating cylinder is designed to be permeable between the non-spinning sections
and players have to time entry and exit
๐
hehehe
Man... AAA VR space games are going to be amazing in a decade.
Thanks to the sort of groundwork you're laying down here ๐
'motor' hulls that spin etc
perhaps I'll get launched into a AAA developer from this โบ
all there is to do is to keep working
Yep ๐
anyone know an s7 case you don't need to remove to put the s7 into the gearvr?
a gearvr isn't going to fit in my pocket
So my Vive has been kind of a pain for syncing since I've gotten it. I've just kinda dealt with it and expected to troubleshoot the thing for an hour everytime I wanted to do something
I finally decided today to try a PCI Express USB 3 card to see if it was maybe the USB controllers on the mobo causing issues (USB 3 didn't work at all on the Vive on my mobo)
For the first time in I don't know how long, the Vive and controllers synced immediately
Yeah, Vive stuff is super super tempremental
I've been having issues on and off again with my Vive.
Ranging from having to kill steam and reboot it in order to get compositor working
To plugging and unplugging and trying different ports
Goddammit Vive, why are you so good when you work??
So my moral is that if you have this mobo: http://www.newegg.com/Product/Product.aspx?Item=N82E16813128651 or another in it's line and are having sync issues, try buying an external USB controller
Buy GIGABYTE GA-970A-UD3P (rev. 2.0) AM3+/AM3 AMD 970 SATA 6Gb/s USB 3.0 ATX AMD Motherboard with fast shipping and top-rated customer service. Once you know, you Newegg!
Just downloaded Onward so I could try its much vaunted locomotion system
Yeah, it's shit D:
Good in that it allows freedom of movement.
But it's not exactly motion sickness reducing like many are claiming.
I can see why people would like it compared to what's out there though.
indeed
Also the voice is always on.
First thing I said when I connected to other people was
'Fucking shit, what the fuck is wrong with this fucking game'
Because I was also having connectivity issues!
Ah the delights of multiplayer ๐
Then I got booted. Fair enough.
Or maybe it was the connectivity issues.
evening ladies and gents
question, trying to setup a scope in VR. whats the best way to do that? i looking into it and found something called render target? but very few tutorials
i say very few... i found none actually. can anybody point me in the right direction?
Would like to help. But actually need a bit of help in that area myself. Let me know what you find!
There's a VR template
That you miight want to try. That has a render target mirror.
Have a look at how it works.
which one is that?
... Whatever link on the UE forums that says VR template that has the most clicks/views/posts
In the VR section
on the vive controller, is grip button a single input? or is there both grip1 and grip2
it's singular
and it's also not an axis
those values are from SteamVR API so that hardware manufacturers can get a little more creative
shrug
maybe for top two fingers, and then grip2 is for bottom two fingers?
like the new SteamVR prototype controller is able to detect grip percentage by you physically dropping the controller
so it uses axial stuff
they just really want to cover all their bases for this shit, because they're thinking along the lines of HW manufacturers making things like, physical putters, or table tennis paddles
so maybe there will be a use for grip2
that we haven't thought of
Oh ok. I thought grip 1 was the left button and grip 2 the right ๐
And I've been wiring both up the entire time.
nah, lol there's actually one physical switch inside
it just gets pressed from both sides
ridiculous piece of interaction hardware
Yeah, that's what it feels like.
I was thinking... this is strange.... I mean, ok, 2 buttons means 2 switches... but the way they designed them makes it very very hard to use one only.
super confusing
when I got VDK1 I was pretty excited about how there were grip axes, which meant to me that I would have info on both sides of the grip's force
Heh ๐
sad day
Well. Oculus has a grip axis.
yeah
What would you do with that mechanic?
I'm using grip as more of a shift button than anything.
well, in my game the grabbing of things is entirely driven by the physical motion of your fingers
so when I get a grip button pressed, I start interpolating their rotation down
I use trigger for grabbing.
and the other way for released
Grip buttons despite their name do not feel good been used to grip things.
grip or grab.
I don't use trigger because I want you to be able to pick up things and then use the trigger to pass through to the device's physical trigger
tradeoff really
so I use the grips for everything to be consistent
Ah. Yeah I can see that. But I've architected my system to essentially detect for that.
I've got a handle component that describes a place on an object where it is grabbable, etc.
the component passes through input from the controller to the object using delegates
i.e. if you have nothing, grip is grab. If it's a thing you want to throw, then releasing trigger drops it.
like the VR template.
But, and here's the trick; if it's an object you want to hold onto like a gun, it'll stick to your hand, and trigger then uses it.
for me, if you grip a handle, all it takes is one click, no holding. But if you're grabbing on anything else, you need to hold down
Question is; how do you drop a sticky gun?
You hold grip...
Then press trigger.
And it'll drop.
Tah-dah!
Grip as a shift button is working out well.
I don't want to have to suddenly randomly switch between grip and trigger for people who are used to trigger for everything
but that's a good idea
Because it's kinda easy to press
in the heat of motion.
Like swinging swords and shit.
But it's much harder to press grip AND something else unintentionally.
In fact all the buttons are easy to press unintentionally during the heat of motion.
you know I like that a lot, I'll experiment with it
I'm also doing this for the future of the SteamVR controller with its physical float grip
Alright. Gimme credit if it works out! ๐
will do
But it'll also appear in the demo I'm putting out, so you can try it then if you haven't already expreimented with it.
cool!
I wonder how that physical float grip works.
Bunch of cap touch sensors?
And you have to physically touch the grip to close the grip?
here's my guesses: proximity sensor with light broadcast+receive, sonar, distance sensor using one or both of them
Intriguing. Are light diodes sufficient as proximity sensor?
i.e. detecting the relatively level of darkness/brightness
if you broadcast from an IR led out, then receive with an IR photodiode you can get great resolution
I did a project that was an IR heart rate monitor in an early EE class
shone an IR led through a finger, and was able to read fluctuations in blood pressure by fluctuations in the IR phototransistor on the other side of the finger
Wow.
it's all in the filtering
That's fancy as fuck.
actually it wasn't, the output was on an O'scope
Heart rate sensor as a freebie in VR would be pretty cool
all analog ๐ easily digitized though
at that point all I knew was analog circuits
๐
but yeah heart rate would be neat
I started on a project trying to use a microsoft Band to provide data that I would turn into "comfort data" about an experience as a research project
didn't get anywhere because the goddamn MSband doesn't interface with PC directly
so I would have had to make a GD passthrough app for my phone
and at that point I had an iPhone......
moved on to greener pastures
yeah
I got it to connect to the PC with BT
but the library MS provided would not ever recognize it
Yeah, but the team that made that would've been; eh, no one's going to connect this shit to their PC, what's the point?
"And no one has BT on the PCs anyway."
Hehe ๐
pretty nice that Intel provides a BT radio on their 802.11 cards nowadays
Oh nice.
pretty weird that you have to plug in a thing to get it to work though..
Like an antenna?
there's this weird wire that is included in the package
yeah but it goes from the card down to a USB header
not floating like an ANT
you plug it literally into TX and RX on a USB header
Huh... so they thought that it'd be worth including the chip on board in case people wanted BT
and it magically works
but didn't think that people would want it enough to just build it onto the chipset?
I have no idea
because the way that it's connected means that they needed a USB port
Ah ok.
which makes no sense since the damn thing is connected to a PCIe bus
Yeah. I wasn't sure how that made sense either.
I guess it makes a little sense because maybe they last-minute were all, "gee, it sure would be nice to add BT 4.0LE to this, let's just slap on a chip
"oh god, it only interfaces with FTDI chips fuck fuck fuck"
"uhhhh, how about a new 'feature?'"
Hahaha.
instead of embedding a USB controller
It's easy to forget as a consumer that they're knuckle heads behind tech ๐
It's so shiny. It must've been created by geniuses!
And now I'm on this side of the curtain.
No worries.
how do you handle both left and right inputs @winged shale
having press and release for both is quite a mess
@sullen stirrup what do you mean left and right inputs?
Left and right controllers?
Your best bet is to examine the VR Template and how they do Left Grab vs Right Grab.
But the short answer is, create two functions or two actors; a left and right version.
And send the function call to the right actor on input.
er... the correct actor.
not the right side
yea but that ends up being a mess
Well there are techniques to reduce the mess. Encapsulation, enums, etc.
But it's just going to be more complex than doing a single controller.
As a matter of course.
@dry fjord Oh man. I properly get that frame timing issue now that we were talking about yesterday
It's because when you have variable frame rates with code that attempts to work according to delta time...
It's because delta time CAN'T be the same every time!!!!
Fixed frame = the exact time of 3 frames = 50ms
i handle left and right imputs becouse i have left and right hand actor
so left imputs go to the left hand and right imputs go to the rihgt hand
the Hand itself is symmetrical
Delta = 50ms +- closest delta time difference...
and has the typical TriggerPressed events and so on
for the last frame to complete.
So that's where the slop comes from!
Unless you use physics sub-stepping.
Right, ok, I get it.
Unless I don't. But it feels like I do ๐
@wicked oak When you say trigger pressed events... you mean in the hand actor?
How do you pass the inputs to those actors? i.e. get player controller to input to multiple actors of your choosing?
very simple, i have them in the player controller
and the Pawn has a GetHand function
Ok. So player controller refrences the hands
so the player controller gets a TriggerPressed(Left) event
And that means the hands can now use inputs?
and does a Pawn->GetHand(Left)->TriggerPRessed()
yes, becouse i defined them
the cool part is that they get to be completelly symetrical
Ah.... ok.
wich means my hands do exactly the same, both of them
I'd been defining inputs in pawn
and passing the inputs to the hands as function calls.
i dont code for left or right hand, i do them as one "Hand" actor
and just use different meshes
depending on being left or right
same thing, then
Yeah.
its recomended to do that in the actual player controller
less messing up
i recently moved it to the player controller
Hmmm. Ok. I might look into it.
was doing odd shit on MP
better to keep the "input" part in the controller
its what its for anyway
Just the VR Template is set up like how I described.
so I just built on top of what was already there.
i ignore the vr template lol
Yeah ๐
It's a useful feature.
just implemented it on C++ for my game
Ah yeah
Ok. Well, as long as my shit works, it's valid - is what I can figure!
Night
Anyone available next week for a VR Game Dev Battle ?
@winged shale sadly no ๐ญ
But, you get to be on stream and see which one of you can produce the best product by the end of the countdown!
Time Constraints, pressure, competition
This will be round 3
I'd love to but I've got so much crap to do on my own things..
maybe I'd do a Daydream VR game jam when the free code hits my inbox
I found that the best way to record VR footige is through the steam VR mirror, but if I want a standard non-stereo render with all the VR controllers etc, what do you do?
you can see the VR controllers in the mirror
they're just actors in the end
there's an engine source change you can do to get widescreen output too
neat
it should default to right eye. I don't know what left-eye-dominant mutant at epic made it the left eye.
I'm also annoyed
they probably did it so they could have the texture first from the render pipeline, and start on display compute in parallel while eye 2 (right) renders next
Advice on defining a body/hip area for a VR pawn?
@empty sundial for IK?