#virtual-reality
1 messages · Page 77 of 1
this new batch is impressive.
K, well, I appreciate your opinion Fantis. Hopefully, there's a place for my controller in the market. I've used it, and so have 100+ other developers at GDC last year, and they seemed to love it.
well, can you understand our sceptisism and respect that at least? If there is anything I want you to leave our conversation with is that yaw turning is not how we do things anymore
"we" isn't you
"we" is the bigger community
you personally don't get to say , this isn't how "we" do things. And that's what I think you don't get
You haven't tied it, and you're making wild speculations.
mate, how long you been making games?
uhm, out of what, 800 released games for vive on Steam, how many has yaw turning?
Like, what experience are you standing on?
They don't have yaw turning because there's no useful controller that doesn't make you sick.
You never answered my question: are you expecting people to stand up, and use your controller instead of actually turning their body around?
No, I wasn't
So it's for sitting games
its mostly used seated , as you can see in the video
That works fine and all, but as a VR enthusiast I find it holding back innovation by keeping VR seated
so, what are your thoughts on the Omni for example?
You're right that there's room for seated experiences as well, but it's not what I'm developing for
It sucks
And I've actually tried it
phone, 1 sec
Your product doesn't seem to be developed with motioncontrollers in mind, which makes it a little 2015
Or mobile, but we've seen plenty of motioncontroller solutions for mobile as well
gotcha. Thanks.
@ember comet For your success, atleast showing the concept of how a user would use your product and motioncontrollers would make your page more interesting
Gamepad and mouse isn't how most of us play VR, atleast not on PC
👍
Fantis: I think it would take too much time for me to explain the whole thing to you.
You know when someone has ideas about stuff, and changing those ideas would require too much effort?
Nah I know exactly what it is, still can't get around yaw turning
Like I said, you'll be at the show, come try it.
When you actually have your body/swivel chair
if you don't like it, don't buy one.
What will you be selling it for?
A lot less than the medical bills when you choke to death from cables wrapped around your throat swiveling your chair to all hell and back.
sure
lol, nonsense
right, cuz cable's don't exist
they do, but no one strangulated themselves with HMD cables
Yea, it was meant to be a joke you see...
some one can fall when operating you pad and hit their head and die..
well, I saw no emojies there
Although, I'm really interested to know, you guys are hobbyists, or you work for studios?
what difference does it make? P. Luckey was hobbyist, Notch was hobbyist, etc. etc.
gotta start somewhere and starting in the garage can be very beneficial
The difference is this: Fantis keeps talking about how "we" do things and how it's done these days. I'm really wondering what his basis is for making claims like that. Is the basis that he's working at a studio of 50+ people funded by a publisher who researches this sort of thing extensively, or is he taking his opinion and spreading out across lots of people where he has no real basis to do so.
So, it's got nothing to do with looking down at anyone. I just want to know the perspective being presented to me.
The SeattleVR community?
There's more VR here than anywhere else in the country
Probably the world
k, loading the website up. You sayinf you work there?
I co-founded the studioo
By your standard, I'm probably considered a hobbyist, but no1 was making this stuff so I did it myself
err
well, for once there are plenty of things are being researches and talked about. So no need to go too far to see what solutions work and what solutions don't work. I believe even Carmack said that turning with controllers is 100% wrong in VR. Just because you say "every one is making round wheels, but I will make a square wheel" doesn't mean your solution works for majority of people.
okay, cool
that loads
k, anyway, thanks for your opinions. Hope you'll give it a shot if you're ever able. ttyl.
🖖
Has anyone tried RE7 for PSVR? I think that may have yaw turning, but I havn't heard if it worked well or not. For me any camera take over doesn't feel good on fast rotation. Eve Valkyrie wasn't bad, but its not real locked camera yaw.
it does
well, I am guessing no verdict will be reached today either
btw, PSVR costs $517 in Walmart
$500 on Amazon (from Amazon)
$533 - $550 from other sellers
- shipping
sooo, not to different from Rift
(those are just HMD prices, nothing else)
Rift is $599 + free shipping
the only difference I guess that more people have PS4 than VR-ready PC
and psvr is easier to use
but that looks too much mony for me
maybe becouse its on really high demand?
we got plenty of unsold psvr here in spain 😃
heh, don't know really.. But with those prices PSVR no longer looks like affordable option
I wonder if S8 can handle better graphics than S7. Maybe it can give Gear VR better footing against PSVR
it's a bump up on the CPU but it's nothing major
Oh, I missed the puck controller dude.
And I think I get how he ended up making the puck controller too 😛 wouldn't take no for an answer.
how did you miss the puck controller??
hey peoples 😃 I've got a bit of a rush request... is there a BP template that I can throw into my scene that lets me teliport around the place (not using the default ue4 navigation system) like.. definable areas and teliport locations... Rift compatable would be best
also.. having problems getting ue4 to recognise the occulus remote for anything
@restive blade You're describing the VRtemplate in the unreal engine.
no.. I'm really not... 😛 it use's the navigation system
You mean the navmesh?
yeah
I have nothing but problems with that thing (don't have a lot of time to fix my scene)
Oh... you want something nav mesh independent, but still definable in where you can go?
... but navmesh is how you define where things can go 😛
teach me o wise master!
What's wrong with yours? You can just use multiple nav mesh volumes
and or nav mesh blocking volumes
to shape the defined area.
Also the larger you make the nav mesh, the longer it'll take to update on change.
You can hit 'P' on the keyboard in the editor to see where the nav mesh has defined
But.
Hi everyone
heya
@pearl tangle but.. but.. S8 sounds like a beast on the paper (compare to S6/S7)
Yeah minor upgrades though not like the 900 to 1000 series GPU update
oh, definitely not
I wonder if it will be powerful enough to allow dynamic lighting (a few stationary lights that cast dynamic shadows)
nah definitely not. much better to fake shadows in any VR regardless of what platform it is
@restive blade you just have to disable the check for navmesh and tell it to work on any static mesh or whatever else you want it to check against
it's literally changing 1 blueprint node in the teleport function. pretty sure epic points out which 1 it is in a blog post or something somewhere
@pearl tangle check for normals, 90 degrees or so as a limit I think
Yeah that's if you only want to go on flat ground or something
Looks like they updated the SteamVR stuff... so it doesn't dump you back into compositor on load.
That's nice.
Although not entirely smooth in operation yet.
so whats the loading show then?
It's a blue gradient background
with a window of the application that's loading.
and a little loading device to show it's loading.
Sometimes it turns into a black background. Not 100% sure.
(not 100% sure as it could've been the specific application doing its own thing, but somehow mixing up the compositor loading background)
sounds terrible
definitely overriding it with level streaming and custom load screen is the way to go
Well. It's not ideal. But chiefly, it shows players that the game hasn't crashed.
Which is better than what it was before.
yeah gotta be a better way for sure
hi vr people, im about to try implement a mechanic where you can push yourself around in a vacuum (no gravity environment) with your hands. Right now im trying to detect collision with hands and surfaces, and allow you stop movement with hands, but also push off surfaces and kinda move yourself forward as if your hands had friction if touching surfaces
anyone done something similar to this?
im currently having a bit trouble finding the movement vectors reliably and making sure the hands dont end up in walls
so my idea right now is to use physical animation on top of the full body solve, which obv will cause some disconnection between real movement and avatar movement
but this also has some problems since then i wont always have completely predictable collisions and those collisions will only be brief then because the physical animation will push off the collision bodies and make it hard to determine direction and speed to move
so right now i lean towards trying to implement a physical approximation of this mechanic myself, rather than relying on actual physics simulation... how does that sound?
We won't know until we try it.
I wouldn't disconnect the location of the controllers from the actual location, that would throw people off pretty badly. Maybe some kind of ghosting mechanic to show real position vs model position
You guys knew, htc published this? https://dl.vive.com/Tracker/Guideline/HTC_Vive_Tracker_Developer_Guidelines_v1.3.pdf
yep. had a chat with them last week about sorting out the unreal stuff and got some updates from Epic on the UDN to do with it
So who's to blame for the fact that Unreal games are still not properly detecting left/right controller?
Basically makes Unreal games look a little dumb every time.
@wicked oak So what is unity doing differently?
Because unity games match the controller icons in the Steam VR menu 100% of the time whereas Unreal VR is basically a crap shoot
oh yeah i have noticed that sometimes it comes in screwy
I think that bit is a fix for it
From SteamVRFunctionLibrary.h:
/
* Returns an array of the currently tracked device IDs
*
* @param DeviceType Which class of device (e.g. controller, tracking devices) to get Device Ids for
* @param OutTrackedDeviceIds (out) Array containing the ID of each device that's currently tracked
*/
UFUNCTION(BlueprintPure, Category="SteamVR")
static void GetValidTrackedDeviceIds(ESteamVRTrackedDeviceType DeviceType, TArray<int32>& OutTrackedDeviceIds);
/
* Gets the orientation and position (in device space) of the device with the specified ID
*
* @param DeviceId Id of the device to get tracking info for
* @param OutPosition (out) Current position of the device
* @param OutOrientation (out) Current orientation of the device
* @return True if the specified device id had a valid tracking pose this frame, false otherwise
*/
UFUNCTION(BlueprintPure, Category = "SteamVR")
static bool GetTrackedDevicePositionAndOrientation(int32 DeviceId, FVector& OutPosition, FRotator& OutOrientation);
this is the bit that should make it handy the different tracker objects since it has a DeviceType in there now. Just need to make sure it's pulling the device ID from steamVR rather than generating it itself
Cheers @pearl tangle Will have a poke at that in a bit
Not sure if it will change anything on the controller type but maybe they spit out a left hand controller or right hand controller since it should handle the touch controllers
i cant get tooltips working with 3d umg using the umg pointer interaction
has someone manage to get it working?
Is the signup for the vive tracker over?
Does anyone know how to access the daydream laser pointer?
I want to do line tracing along said pointer, but have no idea how to get the start position / orientation
The form is still online. https://www.vive.com/uk/tracker_form/
@hard light oh, thanks for reminding me about daydream...I have a Google Pixel XL now and Daydream HMD, how are you finding dev for it so far?
it's an absolute fucking nightmare to set up
lol
Well, that does not sound promising at all
Have you done any market research on Daydream users by any chance?
I would really not recommend it to anyone - getting the android SDKs etc in the proper configuration (which is incorrectly documented) was no small feat and took quite a lot of effort
I am just curious what the stats are if any are out
then there's the fact that Google aren't keeping their plugin up to date, so I had to compile my own version for 4.14
now that I've got that far, I can't work out how to access basic information about the motion controller (because it is all entirely self-contained within their plugin and the blueprint nodes are a bit lacking)
but no, I had no data because I'm not interested in developing for mobile VR
@hard light I think 4.15 has daydream stuff updated
well, I'm 4.14 and 4.15 isn't stable, so it isn't very helpful, heh
4.15 is stable
previews are usually very stable already
I am on 4.15 since November (then it was master) and it was stable all the time
?
Previews are usually NOT stable
Maybe before 4.12
but 4.13+ it's been shitstorm after shitstorm
And then release is just as bad
you seem to use a different software than I do
ue4 was really buggy before 4.10 and 4.11, but since then it is solid, including previews
yes, I said before
4.11 was the last solid release
everything since has been pretty sketchy even outside of preview
Yup
give me an example?
?
but either way, it isn't my choice, so it's irrelevant
4.13 crashes and freezes on me on a regular basis
5 hot fixes...
hotfixes make it more stable, so hotfixes are good
non-preview releases shouldn't need hotfixes, that's kind of the point of 'stable release' branches
The point is...
what are you doing for making it crash?
WTF are they doing out in release if they aren't stable?
just using PIE is enough to cause it to crash sometimes
yup
I have to rush back to editor and stop playing around with shit in there
to stop it
but I usually forget
So they freeze
strange
But yeah @hard light , to stop it, you just alt tab back to editor and start touching BPs or whatever
and let the new window open up
that sounds bizarre
that's how I've been "mitigating"
It's so stupid
It's probably a windows/ue4 bug
I havent really used versions like 4.13 and 4.14 I think, I was most of the time on master
some conflict with resources
but master should be less stable than stable versions
?
so if I dont notice any crashes in master then I would have thought the official releases are even better
master has the latest fixes though
yeah, it has
along with latest features
but also new bugs it might have
But before we get too off topic, @hard light so current Daydream implementation in UE4 is poop and you wouldn't recommend at all?
not for serious projects, no
getting it set up is quite the hurdle
I blame the reliance on codeworks etc for that
by the time you're in UE4 you're using a plugin on top of the engine on top of an SDK on top of another SDK
and if something fails in that stack, the whole thing topples and most of the time you get a fairly useless generic error message
damn
well, daydream is still new
so some old engine version like 4.14 doesnt really need to support it perfectly yet
So
I have a shitton of art in my damn project. I already told the artists to look into ISMC whatever they can for environment art.
What other suggestions are there for this multi-storied (this is going to be 2-3 times taller) "arena" platformer/shooter?
is it for mobile ?
It's usually 90FPS, but there's been some drops down to 50 when the full scene is in view, shader complexity is stupid.
nm
Nah, vive
90FPS 😉
Anyone have any suggestions on how to get this to a manageable point?
Hmmmmmmmm, I checked most of the mats
It's about 500 instructs for the heavier ones
I don't think we have any transparencies
that's quite a lot for VR
basically in VR it's better to have atlased textures and simply use straight images for diffuse, normal, etc. Don't make some crazy noodles in materials.
the geometry looks fine, but 500 odd instructions is really heavy
I would take a look at the VR quick start guides and set up your ini config like they suggest (geared more towards mobile rendering)
also, avoid dynamic lighting and bake as much as possible
but Robo Recall uses dynamic (stationary) lighting :/
500 is way way way too much
a regular material has something like 35 pixel shader instructions
2 small lights for spells can take 3 ms of rendering in my game
i have to make sure to not have A SINGLE ONE dynamic light
For anyone interested in my locomotion work - I've released an update:
https://www.reddit.com/r/Vive/comments/5rg44o/freedom_locomotion_vr_updated_110_released_with/
Link to the demo: https://hugerobot.itch.io/freedom-locomotion-vr Thank you all for your great feedback to the Freedom Locomotion VR demo. All...
check this out
red is Base Pass, turquoise is 2 lights
TWO
i disabled those
btw, no shadowcasting
lightgreen is the reflection capture
wich takes way too much btw
and then the postprocess stuff
bizarre
sounds like UE4 has unoptimized lighting :/
Remember that Doom 3 BFG runs 90 fps in VR
and it's all real-time. There are no baked lights at all.
but why?! You'd think lighting is the first thing to be optimized :/
because unreal engine was traditionally an engine for static stuff I think
@wicked oak do you have performance issues with lighting in deferred or forward ?
deferred
i cant handle the forward basepass
with the amount of drawcalls i have, the depth prepass takes too much
I see.. I recall seeing something about dynamic lights being more performant in forward
What kind of hardware are you running on ?
(assuming they don't overlap much and there only a few in the view and only one or two cast shadows)
@wicked oak forward is slower for you? depth prepass is so relevant?
sounds like you are going backward about this 😃 gotta optimize your art and use forward 😉
I think forward is way faster in my case...
haven't tested too much though since I still have to wait for ssao support
Forward works best when you have simple-ish geometry
if you have massive amounts of drawcalls, there is too much overhead on the depth prepass
and i do 😦
Anyone here got any ideas for optimizing heat/battery life when doing stuff for the GearVR? We have made a simple 360 stereo panorama viewer using unreal. Using it on a Galaxy S6, it overheats/runs of battery incredibly fast
yeah that was my thoughts too. It's an external developer we had inhouse doing the app - now the client complains about it and they've given me the task to see if there is anything we can do besides upgrade to a Galaxy S7 which should be much better
What confuses me is the Oculus native 360 stereo viewer runs at a much higher quality than we can achieve with UE4, but with much lower energy/heat costs
but might be because they build their own engine/renderer for that viewer
yeah. We have a single sphere. A material that switches between 2 jpgs depending on screen position(hence getting stereo effect)
we are pushing 2x4k jpgs though, but anything under that is an unacceptable quality - especially compared the 18k we use in Oculus Stereo Viewer
you should use Stereo Layer of cubemap type instead of sky sphere
it's timewarped and should provide exact the same quality as Oculus' native 360 deg. pics viewer
Hmmmm!
interesting
You got any link for some documentation? Seems like the only I can find is some blueprint API calls
it's Stereo Layer component
unfortunately as with everything VR it's a limited documentation, but enough to get you started
also check Oculus dev forums
that is very interesting
seems like people are also using it for HUDs inside the Rift/vive
yeah
Thank you very much! I will have to dive into it and see if we can get it working 😃
np 😃
@tawdry dragon sounds quite heavy to use the full ue4 just for showing a 360 panorama
What do you mean by Full UE4?
well unreal engine 4 has many features and many are enabled by default without any way to disable them, but you surely dont need them for a 360 panorama viewer
well, there is also some interface/product selection stuff going on top of viewing 360 images
ah
But we needed to use a platform that allowed us to make a custom app for the client and also add products/functionallity later. External devs suggested Unity, but update costs were simply to high. Since I know a bit of unreeal, the idea was that we got an external dev to create the base system and then we would do app updates inhouse
If Motorseps suggestions works out, then I will send him hugs and kisses!
🙀
👍
@hard light I've roamed the forums, and somehow people manage to use it with 4.14
yeah, getting the plugin to work was not a big deal
Hopefully I am allowed to show off the Oculus/Vive thing we're doing for a client atm, as soon as its done of course 😃 It's gonna be used in some large conferences in Las Vegas and New Orleans 😄
I just want to know how to get the relevant transforms so I can actually do stuff with the laser pointer, since Google's plugin is very heavy-handedly self contained
does nvidia post in any forum threads or anything on status/plans for the vrworks branch?
is it just dead at 4.13? I'm sure 4.14 had a ton of merge conflicts due to the forward renderer
they posted an article not to long ago about how to integrate lens matching shading into UE4
sounds like Robo Recall will be free
hah, that's cool
Ha, Nolo VR raised double of what they asked for in 2 days
new DWVR update
gonna record a video now
as the new sword mechanics are really cool, but might be slightly unbalanced OP
but who cares at the moment, let the players blow skeletons into orbit
YESSSS! Oculus wins!
Oooh dang!
but they agreed Palmer violated NDA
so I am guessing Palmer's new job was to be a scapegoat
I am guessing money will exchange hands anyway
hopefully not even 2B
$500M is what Oculus will be paying Zeni :/
hopefully get some of that facebook money to help pay for it so it doesn't hit vr too hard
Palmer: Make DK1, get ton of money, haters gonna sue, still gg
$500M is still a lot of money to pay for violation of NDA, especially since trade secrets were not misappropriated
I bet Fb/Oculus will appeal
I could have gone into dev grant if anything 😦
maybe it will shut Zeni for good and they won't do any lawsuits against Fb/Oculus afterward
"Of the $500 million, Oculus is paying out $200 million for breaking the NDA and $50 million for copyright infringement. Oculus and Luckey each have to pay $50 million for false designation. And Iribe has to pay $150 million for the same, final count."
oh boy
so as a company, Oculus loses "only" 250M
Carmack is cleared (unless some other news come out later) and that's important !
Well, I'd like to say I am going to sleep well and continue with Gear VR development since now Oculus is out of the woods with the tech 😃
STATS!
What's zenis angle with the case? they just want some money?
cash grab
HDRF sound?
FRDH sound
WTF?!
how is that possible from legal perspective ?!
this could be enough to do some serious damage to the Rift
managed to do a poc this evening on the pushing around. Result is here: https://www.youtube.com/watch?v=KIfoacQ0iGM
lemme know what you think of the idea
@hard light but if it was decided that Oculus did not misappropriate technology and didn't steal anything, how can someone make sales of Rift stop ?
@mighty carbon they said the jury found that John did take source code and files
You may want to read the article again - the court awarded Zenimax half a billion dollars because Oculus infringed on copyrights and broke NDA - but it suggests they still believe Zenimax VR or game code may be in use within Oculus software. I suspect they may be forcing an audit
They used "DOOM" trademark early on
that's all
so, Oculus did violate that trademark and copyright
for DK1
not for CV1
not even for DK2
so, jury's verdict doesn't count I suppose, if jury said they didn't steal the tech, but ZeniMax still saying they did
"the jury also found that id Software co-founder and current Oculus CTO John Carmack took Rage source code and thousands of electronic files on a USB storage device which contained ZeniMax VR technology when he left the company."
they haven't awarded any damages in this instance, because it has yet to be proven that Oculus is using it - the next step is to stop Rift sales and audit Oculus
Oculus are in trouble
hmm.. sucks
if you don't play by the rules, you run the risk of getting caught and having it fuck you over
¯_(ツ)_/¯
So what you're saying is I should hurry up and buy an oculus before they stop sales? lol
well, if they can order Oculus never sell Rift, then why on earth would you ever buy it
Why buy something that may die number two
eeh? Oculus is not inferior hardware
Vive > Rift
Just get a Vive and be in bliss
better tracking solution and the hardware is evolving more quickly
And know that Vive had everything from the start
because closed source encourages higher quality products and now quick dorky tech demos Steam is swamped with
walled garden that is
That seems the opposite IMHo
Rift and Vive games largely operate on the same open platform, which is why most Rift games can be hacked to work on the Vive
says life.. Look at Steam and how much shit is on it
Open Source would have more competition
And would expand on ideas already well established
open source is utopia
Steam is not open sourced?
I don't get your argument
when it was walled garden it was soooo much better
Steam is not open source - you don't get to see Steam's source code. Only API is exposed
Hmmm, when only triple A shit came out?
also, open source != walled garden
I know, we were talking about open sourced and you said Steam.. so
what does open source have to do with hardware ?
OSVR is open source and such a shitty VR system
Steam VR is open source and is inferior to Oculus Software
open source doesn't encourage competition - it encourages being lazy and borrowing from existing solutions
look at AMD and Nvidia. Nvidia has closed source state of art drivers, while AMD has open source garbage drivers.
Hey VR folks, has anyone had any success layering cameras in VR?
@heavy tiger layering?
Using multiple cameras to render different parts of the scene.
as in you want to place a camera for each eye?
or are you just saying you want them to be able to see the output of another camera like on a monitor or gun scope or something?
why would you want to do a camera for each eye? That's already what the Oculus and steam plugins do but they have convergence properly figured out and whatnot
man, I am not gonna sleep well without knowing if Zeni can halt sales of Gear VR :/
@hard light I agree vive > rift but rift's controllers destroy vive's
how Vive beats Rift ? The only thing that is going for Vive is better 360 deg tracking (room scale)
vblanco is the one who put a breakdown here (and on Fb) for all 3 systems
Rift has better ergonomics, better software, better controllers and better support than Vive.
(and I can vouch for support part personally)
and of course for open source fanatics Vive is better simply because of Steam VR being open source
actually, Open VR, not Steam VR
just looked for Open VR and Steam VR and both have only API code open to runtimes. Runtimes are closed source.
With a third tracker for the Rift their 360 deg room scale tracking feels just as good as the Vives.
there ya go
but then you have to buy two additional trackers to get the same functionality of vive out of box
Rift unfortunately is plagued with reliability issues for room scale tracking.
You only need three with the Rift
It'd be nice to say that 3 camera Rift would get you equal Vive functionality... but it's really a gamble.
Some people yeah, no probs. Others... well... maybe 4th ones a charm?
yea 3 with rift vs 2 with vive. but doesn't rift only come with 1 tracker?
1 for the HMD, 1 for the controllers
Ahh true, I think it does
And 1-2 more you have to buy seperately.
Are you sure the trackers for for separate use?
As a developer though, you simply can't assume that Rift users will have room scale. Which is a bummer.
On the flipside, having a turning solution is useful for everyone.
I don't have room for room scale VR, but I can turn 360 deg 😃
You can't or you can?
So even with Gear VR + Nolo VR I could only stand in place
And why can't you turn 360?
Haha
Particularly relevant because I'm working on a walking on the spot method of locomotion 😃
It'll work well with the PSVR too!
I am fine with teleporting. I'd walk if it was natural walk, like on those omni-threadmills
It's a pretty natural walk. Truth of the matter is, either solution (in place or treadmill) is going to require some time to get used to.
Like you might think treadmill is good because you get full hip swing. But then you're not accounting for the weirdness of the harness and the feeling on your feet (sliding, and possibly sliding along a concave dish).
true, but if I use treadmill, I am not going to walk into anything by accident 😉
With the boundary stuff, it's pretty hard to accidentally walk into stuff as well.
Either the custom boundaries I use, or the normal guardian/chaperone boundaries that the Rift/Vive uses.
I still manage to punch almost every thing around me with the boundaries up, haha.
@green ice Fair call - but one of the main problems with normal chaperone/guardian boundaries is that they draw over everything irrespective of depth.
my custom boundaries are actually occludable in game.
So you can still tell when your hands are infront the boundaries. Also, they can be altered independently of the chaperone stuff, so you can use them as 'advanced warning' (i.e. set them back a foot from the wall).
Ahh cool, I haven't had a chance to test the system out on full room scale, stuck behind a desk with about half a step each way.
- Normal system that is.
😃 need a link to the demo, or have you already downloaded it?
Which demo?
My locomotion demo 😃
For the Vive only right now though:
https://hugerobot.itch.io/freedom-locomotion-vr
Ahh won't be able to test it out, moved over to Rift for now. But that walking actually looks pretty cool.
Cheers. But yeah, going to start working on Rift compatibility
Just flicking through the vid, that movement where you are kinda jolting forward, wouldn't that make you feel really sick?
around 3:51, I think I spoke too soon though.
But also, not really - walking in place solutions cause stimulation of the vestibular system
which adds a lot of noise and in many users drowns out the expected signal of forward motion.
The system has a direct relationship between how much you move and how much you move forward in VR - so that the more you physically move and push your vestibular system around, the more you move in VR.
It doesn't work as well for some people as others. But I recently added a comfort mode that should help to further reduce the worst of the motion sickness that might occur.
Oh sorry.
You're talking about Dash Step!
Yeah, that's the one.
Yeah, Dash Step operates on the basis of vestibular lag.
i.e. it takes time for your vestibular system to kick in and detect the mismatch.
If you have visual motion under this window, the vestibular system doesn't kick in.
Oh, that's cool.
It varies from person to person. But generally I think 50ms should cover most people.
But you can alter it in the demo.
vive/ htc support is way better than Oculus. Oculus has absolutely terrible support. Doing a project for Samsung about their Gear VR even and can never get any proper support from Oculus. whereas I was on the phone for an hour with HTC's head of VR in europe last week just talking shit. Epic gets way better support from HTC than they do from Oculus as well
i don't have any business headsets or anything. just consumer 1s
i had major issues with their general support when i was buying the headsets thats for sure. their support was just as bad as oculus.
I can't imagine a random guy getting in contact with a VP.
My own experience with HTC support was - eh. They seemed relatively polite and prompt via the web chat service I went through.
epic just put me in touch with their guys and they were great. the oculus 1s were no help. was trying to do a project for a client to release on Gear VR, dozens of emails with support. I needed to check if I could actually release something on the oculus store since the client was Johnnie Walker. their best response was "just build it and try and publish it and see what happens"... that would really go well with getting hundreds of grand from a client to build something on the principle of wait and see...
But they did manage to send back the same controller with 0 fixes, so I had to send it back. Also had 1 week delay in response time, prompting me to contact them again (they said they'd call back).
oh their web stuff and phone support stuff was horrible, but the Vive stuff directly, not HTC is quite helpful now
Yeah.
So it seems like their dev outreach pretty good. Direct consumer customer service kinda crap.
@pearl tangle I actually got a lot of help from Oculus with Gear VR stuff
at the beginning it was shit, but now it's pretty good.
haven't experienced publishing side of it yet
btw, do you think Zenimax can also halt Gear VR sales/support if they manage to do that to Rift ?
@mighty carbon sure, they will have to halt everything 😈
for this fight with oculus I now love zenimax. even if they don't win, they greatly damage the reputation of oculus and that's great for the vive
But the Vive needs the oculus so they are forced to keep making better stuff.
gear vr is using Oculus hardware and software so i guess that would be possible
it would only be in the US though anyway if something like that happened
@green ice yes, but oculus won't die, they will still want to improve the vive
@mighty carbon bullshit
a good company can make progress without competition
SpaceX had no business reason to spend money on developing landing rockets, but they still did, just because of Elons vision
and steamvr is an open platform, so there's automatically competition between different hardware manufacturers for that
eeh, SpaceX has a business reason - to reduce cost of flight
not because Musk just wanted to make it happen
they are already the cheapest by far
how is babby formed?
@odd garnet By putting in the peepee with the poopoo
On it
need to do way with instain mother
@mighty carbon spaceX is actually doing all of their visualization with Unreal engine btw. They are also using the vive to actually do a lot of industrial design and test out all their spaces and everything
also pretty sure microsofts open standard for VR and AR will easily eclipse Oculus sales within the year.
How do you guys test your game without access to the VR headsets and controllers?
@heavy tiger Make a non-VR pawn?
Kinda hard for me given that everything I do is related to the pawn 😛
if you are doing a VR game with motion controllers you can't just fake it during dev and assume it will work
@pearl tangle Depends on the thing you're testing. But yeah, not advisable to fake too much of it.
Only for basic convenience
anything with motion controls you can't really just assume
even sizing of objects and stuff i na scene you can't build it like a normal game either, things are very different for VR dev. that's why so much of the stuff just hacked in for VR is absolute shit
Yeah, there's really no way to develop for VR without having access to VR gear and regular use for it.
just signed off another Vive project with a client 😉 if 1 of the other 3 proposals sitting on my desk goes through definitely hiring a new 3D guy if anybody feels like moving to singapore 😃
sounds like VR is booming out your way, hehe
over here not so much, I suspect because there's so much hardware out here now that everyone and their aunt has their own and does their stuff internally now
I've got a background in 3D! I want to be in Singapore.
But dammit, I'm doing my thing!
I forgot how to import movies all of a sudden
Lol... I'd be happy to be a 3d guy for you..But Singapore might be to strange for me
Spent the day working out how to fill in textboxes from keyboard while in VR ( to facilitate naming GestureTracker gestures during development, not for end users )
Pain Is Fun, managed to work it out
@simrak#0804 Do you mean typing into a physical keyboard, or typing into a virtual keyboard with motion controllers?
Going with physical keyboard for this one since we will have a 'controller' coordinating the thing.
I figured out how to make a virtual keyboard but didnt see a need for it for the moment
Fair enough 😛
anyone has any examples on MotionController replication that works? I had a prototype working using "Get Hand Transformation" - but for some reason that doesnt work with Oculus Touch
Im currently using one that reads the motioncontroller location locally and then uses RepNotify to send it to a server and update the location on all other clients - however for some reason it bugs out now and adds some wierd offset to your controllers on the local client, but not on the network
@tawdry dragon Did Stereo Layer work for you?
@mighty carbon hadnt had the time to try it out yet - will do in the weekend
ah, np
right now I simply cant get my head around that replication problem
i just update with the motion controller component transform
i make sure to keep the player index to -1
unles its the locally controlled pawn
then its 0
and then on clients i just interpolate to the last transform
Carmack's outlook on the trial
RPC
from the local pawn into the server
the server just updates the replicated_transform variables i have
and then every other client just interpolates to that
btw the transform is on local space
not world space
could be improved by storing it in a more compact way
not a full trasnform but a simple 16 bit location + low precision rotation
hmm, will have to try it out
atm I am loosing my hair :p
we have a milestone dilevery tomorrow and for some stupid reason, it stopped working
Nothing like that 'something breaks the day before' feeling
@tawdry dragon I have a thing that does mulitplayer vr stuff
@silk lodge BP or C++?
@silk lodge Can You show me the graphs where You replicate motioncontroller transform?
is there a good virtual keyboard system available for VR ? (marketplace maybe)
or is it easy enough to make it from scratch ?
@silk lodge Sorry, something just came up and have to go. Maybe I Can catch You another time
ill be around!
Sweet, thanks 😊
anyone figured out a good way to grab items not using physics handles?
but still respecting collisions
conceptually i want to grab an item and move the item using physics, and have my hands follow the held item perfectly (without lag)
kinda like with attachtocomponent just the other way around (Attach players hand to item)
is there a way to do this reliably using some kind of chain of attachtocomponent?
im thinking that if i recorded the grab offset (which will be a delta in worldspace), i would somehow have to convert that offset into a local space
and then move the item offset by that grab offset
but math is killing my head
Check out mitch's VR demo. Sweep the object each frame to the location of the controller.
hmm i'll try that out, i guess it will just stop if it hits something otw
That depends on the masses of the grabbed actor and hit actor. It's still doing physics collisions.
i'll try and see how it behaves
oh nice i see he already did all the hard work with the manual calculation of custom physics, so nice of him 😃
Is ' Mitch's VR demo ' something different from the VR template, or is it referring to his youtube series?
"Mitch's VR Lab, an introduction to VR in UE4" ?
i just wanted to thank him 😃
... and ask some math questions 😛
i see he still allow the rotation to change due to interaction with the world
so he doesn't snap back to the original grab rotation
but should be possible with the stuff he did
Yeah, I noticed that too. I changed that for my project. My grabbing is based on his logic
Same as for the location. Save the initial rotation, then set the rotation each frame.
but i go blackout and faint when i see Quats 😄
do u interp or something to the initial rotation using those Quats?
Nope, just setting it to what it's supposed to be. You never notice it rotating, no interping/lerping needed.
just reading through the code
so when the object is picked up, he's using AttachToComponent with an object that has simulate physics on
and then move it manually too in the "Handle Bounce Back"
technically i thought if you used AttachToComponent u couldn't use Simulate Physics = on too
It gets attached to a scene component. IIrc that's different than attaching it to a primitive component
I don't think attaching is obligatory in this case. You could just save a reference to the "attached" actor in a variable instead of attaching it.
true, looks like it
will have to try that out
seems kinda unnecessary to me
how do you derive the relative rotation at pickup?
i mean the corresponding initial rotation as this
Get the actor rotation and use the transform and inverse transform nodes to get the local world rotation
Ah, hmm, let me check
ty man
It's been a few months since I did this. Maybe there's a nicer way to do it than to break the rotator into axes and transform them individually. Not sure.
oh i hate the math
looks similar to what i did with the procedural tree stuff i made back a while
i was also breaking up into axes
but didn't realize that's the same here
i'll try it out thanx alot
Np, caused me a lot of headache. Nobody else needs to break his head on this 😃
very much appreciated
@eternal inlet the way the template does physics at the same time as low latency is by setting a really high damping value
*mitch's template
It had an issue with not inheriting angular velocity when you threw. I was helping mitch figure out how to do it and he ended up had to implement this whole quaternion library in blueprints because not enough stuff was exposed to get the right axis angle calculation for the angular velocity impulse call
i see, it's totally over my head 😄
yeah, if you have a non-physics object (kinematic), the engine will automatically calculate the velocity and angular velocity when you release for a throw and enable physics (epic template does this)
but if you have physics on before it gets messed up, the code doesn't calculate it, can't remember all the details
but I think Mitch's template does things the best way for now
makes sort of sense
yeah and add on @graceful junco's rotation stuff into it, and it's really the best i'd say
There is one issue though when you grab an actor (like a cube) and place a second actor on top of it. When you move your hand, the actor on top stays where it is and the one in your hand slides, like there's no friction. There are workarounds for that too, maybe a bit messy.
There's another VR template (VRExpansion_something Iirc, which tries to fix that in c++). That kinda works, but I prefer the feel of my method.
ah yes i saw that.. the mordental something plugin?
I think so.
im shit at c++ though
In my project I'm checking if an actor gets placed on top, then I'm releasing the actor and reattaching it with a strong physics constraint. It then lags a bit behind the controller but for my case that's no issue, since you're moving slowly anyway, trying to balance the two actors. When the top one falls off, I'm switching back to the Mitch grab style.
sounds like an intricate workaround 😃
Yes, took a lot of time implementing and tweaking that. But it's a vital part of my game, had to be done.
yeah it's these details that can make or break a game i suppose
what game are you working on btw?
i may have asked already, but im so forgetful 😐
VR physics puzzle. Building towers and balancing objects. I'm not really actively posting stuff yet, I prefer spending the time working on the game. https://twitter.com/VROverstuffed
There's no video yet showing what the game actually is like.
😃 Thanks
has anyone used "r.MobileContentScaleFactor" with any success on GearVR?
never tried it, but you can't scale resolution on Gear VR
I guess gearvr overrides it, right?
I'm having thermal issues right now. Game will only last 20 minutes until I get the warning. Now it's my understanding it needs to last 45 minutes
don't really know.. It's just how it is no matter what engine you use for Gear VR. Resolution is fixed to native.
if you publish on Gallery, 20 min might be okay
:/
or make it for S7/S8 and drop S6/Note 4 support.. S7 doesn't heat up as badly as S6
I'm having both S7 and S6 last 20 minutes
yeah.. I don't really see how to fix overheating issue
I bet even with blank scene UE4 will heat up phone badly
As Carmack said, UE4 is not a good engine for mobile VR
is there any unreal game out there that lasts 45 minutes?
that game looks nice
Silly project my coworker built on the side -> https://medium.com/shopify-vr/marionettes-in-vr-6b596620c3ca
I'm guessing it helps to have a giant unlit shader in the background that has like 1 instruction
that too
that's very cool!
@raven halo are you under 100k tris and 60 drawcalls in the view ?
yeah
I have like 20K triangles
and maybe 10 draw calls
I'm going a bit far with the shaders
like I mentioned the other day, I'm doing parallax corrected cubemaps, tonemapper and all that in an unlit material
I think I'm just going to have to optimize the shaders somehow
or go with old school materials - just plain textures, nothing fancy
you can't expect to go all out with shaders and not overheat phone 😃
maybe S8 can handle all that for 45 min, but definitely not current gen S6/S7
I'm not doing anything I haven't done like 4 years ago on an iphone 5
well, it wasn't VR and iPhone has better hardware afaik
of course there are many more pixels to process
that too
but yeah, it's insane how hot it gets
(by hardware I mean SoC)
I've stress tested iphones gpu until they run out of battery, and I've never seen it crash or anything due to the thermals
I think it's fair to compare Exynos/Qualcomm vs whatever Apple uses as AMD vs Intel
rock solid 60fps too
That marionette thing was reaally cool
am i right in thinking that switching to forward rendering would affect how emissive materials work?
i no longer see the "glow" around the material with FR
wasn't "glow" done as bloom ?
@spring pond Looks like fun!
Vive controls not intuitive compared to touch WTF BS is that
lol
@wicked oak Tooltips in UMG doesn't work in world space, it's screen space only
I'm faking it with MEnu anchors
So I'm reading that it's possible to throttle the CPU and GPU with GearVR in Unity, but I can find no information at all about this feature in unreal
well, Oculus docs say there are BP nodes for that, yet there are no nodes
hah
really?
have you checked out Oculus repo?
maybe it's in their branch
I've just asked
hopefully I'll get some answer
because I think that's what I need
heh, just found out what you were referring to:
"Added Blueprints plugin that supports calls for Gear VR functionality such as GPU/GPU throttling, querying headphone connection, querying device temperature, and more."
no, I haven't
ok, I will let you know if I happen to find it there
but I am considering using it in the future since Epic takes forever with updating launcher engine version
thx
by the way
I see your name pop up all the time
is it just me, or are there very few people asking questions about gearvr development?
very few people
99% of people use Unity for Gear VR
so, it does seem like I am the only one asking questions
also it shows to both Epic and Oculus that people don't give a damn, so they don't push Gear VR for UE4 too hard
TBH, I bet Unity has the majority of VR devs in general
I think I actually saw soemthing a few months ago, some chart that had like 75% unity
or something pretty high like that
I don't see myself going into Unity.. I love BP and UE4 tools in general.
I guess I'd give up mobile VR before I give up UE4 😉
I think just give up mobile in general rahahahaha
I'm with you on that motorstep
well, vblanco released on Gallery, so I assume it's no problem to release on Gallery even if it overheats in 20 min
I guess if there is no way to improve the overheating I have no choice but to release on gallery
found it by the way
it's only in the Oculus branch
Right, I don't see those in the 4.14.3
Oculus says they should be in stock UE4
heh
but they aren't
they are definitely not there
maybe in 4.99 ... 🙄
lol
unity gets a lot more Gear VR attention than Unreal does since UE4 they are trying to integrate in the native stuff rather than just have a plugin so it's a bit slower to adapt in that case
plus gear vr and cardboard are the cheaper vr platforms for developers to start with and unity gets more beginners than unreal does so makes sense
@pearl tangle doesnt have to be cheaper
I dont think I would spend more or less money if I would make my game for gearvr and not for vive
i meant if some kid is trying to get into game development and wants to do vr then it's a lower barrier to entry to buy an s6 and a gear vr 2nd hand than to get a top end pc and vive
and those are a lot of the developers on unity whereas larger portion of devs on unreal are a bit more experienced developers
@mighty carbon my programmer is looking into the gear VR plugin and the missing BP nodes, he says that for some reason it doesn't come compiled with the oficial release lol. I think they might have forgotten. We are trying to get it to work from the official release version. I'll keep you posted
We would rather stay away from oculus branch due to obvious reasons
Ah man. Vanishing of Ethan Carter is a beautiful game
marred by VR 0.5 control scheme
I wanna play it in VR, but get motion sick from playing in it.
@mighty carbon hah, just managed to double the time before overheat in GearVR by changing the CPU value from 2 to 0
@raven halo why would you want to stay away from Oculus branch? It's in sync with Epic's and has latest Gear VR stuff.
There is 4.15pre3 on Oculus github
my problem with oculus branch is that it crashes on Vive
at least it did last time
but i wnat to try their oculus forward renderer
its faster than ue4 official one as it doesnt do a depth prepass
but it has less features
and its worse if you have plenty of lights
i did test it, it was faster on my game
it also crashed on vive
that back in the 4.12 or so version
I don't think 4.14.3 and 4.15previews have anything different than official UE4 except up to date SDKs
They have experimental branches though
But official branches are identical to stock UE4 as far as engine features go
@raven halo how do you know that throttling CPU down won't bring performance down?
yes
in my case it has not been a problem
it is very thight though
I assume that there is no parallelization of game time and draw time, right?
both added together reach 15ms
also
GPU I've brought it down from default value of 3 to 2
I've gone from 15 minutes on the S6 to 28
I'm now testing on the S7
(where I have the expensive shaders)
and it's already at 32 minutes
and keeps going
on mobile too?
and ender thread uses extra worksers in some cases
yes
but there is 0 parallelization on the actual game thread
for your Ticks
good to know
interestingly enough, S6 seems to overheat at 35º more or less
I am already at 39º on S7 and isn't complaining yet
My s6 overheats constantly
Nice
and 30 minutes on the S6
I guess Oculus won't complain?
or do they also ask for 45 minutes on the S6?
I just wonder how to apply throttling in a game where full CPU and GPU power are required most of the time
45 min on all supported phones
So, if you throw out Note 4, S6 should run for 45 min.
is Note 4 worse than S6?
can I throw out specific devices? o_O
is there any stats where I can see how big of a percentage that is from gearvr total users?
keep in mind that in my case S6 doesn't get to 45 minutes despite halving the shader instructions! There is not much more I can remove in my specific case
My s6 overheats constantly
Note 4 is sooo ancient :)
Google specs of Note 4 and S6 and it will become clear
Afaik you can only request to drop Note 4
mmm, whats up people
@graceful junco @eternal inlet saw you guys mention my plugin, the reason that sliding "kind of" works with it is because i'm using an actual physics constraint to hold the objects. If you set the friction higher on the top object it should stay pretty reliably, minus up/down quick movements of course. For stacking games that likely isn't good enough to account for user error but a seperate constraint of the contact points would work like Happy is already doing.
Yeah Zap, figured I should probably take a look in here
It's not a bad place for UE4 VR devs to hang 😃
sometimes busy, other times not so much.
Some colorful characters in here like our resident Gear VR UE4 fan.
lots of down time at work anyway, figured I could at least sit here sometimes :p
o/
yeah i talked to @graceful junco about it
ah there he is
well, nice to see you @tired tree, and nice job on the VR template stuff
Yes, cheers indeed.
appears happy was looking for unnatural friction for his game, which is fine, if you are looking for natural friction amounts you might want to use physics constraints for your grip jonas
i'm not a big fan of mitches implementation for interactive holds, since he is applying fake gravity towards the hand it has unnatural results and also can "orbit" around objects
Anyone interested in haptic vests? Like this kinda stuff:
http://www.woojer.com/
Woojer's amazing Vest™ & Strap™ will totally blow your mind! www.woojer.com #BornToFeel #VR #Music #HoH #DJ #Fitness
yeah i tried that, but im gonna experiment a bit more with alternatives, since i can't get it to behave reliably
I agree, Mitch's approach has some shortcomings for sure
Actually quite like mitches implementation... although yeah, needs limits to sort out the oddness.
it will need some improvements
But it's a good starting point.
agree Zap
Also needs bendiness.
like Gorn?
Because it's only positional at this point.
Er... not the rubbery type of bendiness 😃
hehe
Just rotational push
stick your hand into a pole holding a block in the template
But actually quite like Gorn too!
ah, because it tries to find a free location
yeah
@tired tree Haha... yeah.
i tried adding in @graceful junco's suggestion with lock of rotation
it still does funky stuff, but behaves a bit better actually
What happens?
big problem is the lack of friction
its a shame that the late update path isn't exposed to blueprint, you could also fix the interpenetration of held objects then too
late update path is a late tick of a sort?
Late tick for hands
in the render thread, it moves all attached objects to a new position
so if the controller is moving fast it still appears locked on
ah
the problem is if the object is colliding, it will move into the object visually, because of this
the collision isn't re-ticked in the render thread for obvious reasons
I turn off the late update for a gripped object when it is in collision, but thats not currently possible with blueprint only
But he had the 'real hand' without late update it looked like.
Does turning off the late update make a big difference?
yes
oho
like.. 10s of cm? 😛
since frames are 90fps intended, and you can move what..like 20 feet per second easily? thats 20ft / 90 =2.66 inches of offset per frame
+/- depending on the arm speed
at 45fps its muuuch worse
Hmmm.
i have a similar issue with the fullbody solving from IKinema
Tricky stuff.
it's lagging behind all the time
yeah, it will always be that half a frame or more behind
Maybe you can er... shift the collision in the direction of motion
it's pretty noticable tho
to compensate.
Ik is hard to fix too, because you don't want to run IK calcs in the render thread
for just arms you can late update them to the controllers and let them shift at the shoulders
but full body doesn't really work
ok, not even sure IKinema allow to change any of that
you could sample last frame and adjust the ik locations if you wanted
it would be more accurate over all
only quick velocity changes would be slightly worse
but the normal operation would be better
right now i just pull out transforms from the controllers at tick and feed into the ik solver
but im so far only bp
so if it has to be done in c++, i will be fucked hehe
yeah if you took the velocity from the last frame, and adjusted the current location by that, could probably get a better result
could be BP
would have to take tick time into consideration, since you would only be half a tick or so off
i've seen that if i calculate velocity by lastpos-currentpos/deltatime
i will sometimes get some weird spikes
i've tested this on normal 3rd person template examples too
delta time is rock solid
if you get a tracking hitch the velocity goes haywire
but for some reason world position some times gets offset
ue4 doesn't keep the controller from flying out into nowhere for a frame....