#virtual-reality

1 messages ยท Page 141 of 1

glad plank
#

only 17?

tired tree
#

but still, new hardware, new users

glad plank
#

Have there been any new Vive/Oculus numbers?

#

Ive seen shipped stuff for PSVR

mighty carbon
#

only 0.71% of RX480 users

sly elk
#

i have guesses

mighty carbon
#

most of AMD users can not run VR on their GPUs

#

so it's pretty much safe to say if you don't cater to AMD GPU owners, you don't lose too many sales

#

back to my question about UE4 and people who worked at Epic or promoted UE4 heavily. Just seems like the passion has wound down for UE4 among the initial wave of promoters ๐Ÿ˜ฆ

#

Any idea what happened? There gotta be some kind of event that led to exodus and drop of support

sly elk
#

what do you mean by that? Lots of people are happily using UE4

mighty carbon
#

yeah

#

but a lot of people who worked on Ue4 from day one are no longer working at Epic

glad plank
#

Who left besides cliffy?

mighty carbon
#

and a lot of people who made training materials for UE4 either stopped doing it or moved onto Unity

glad plank
#

Unity has many more users

#

UE4 is a professional tool

#

Fortnite BR is about to bring A LOT of money into Epic again

#

They are making bank off of PUBG too

#

maybe they will hire more people, I dont think the engine has suffered though

#

its still offers more than unity

sly elk
#

I think partly UE4 is just in a different phase. Its not new now and the staff needs around it are not the same. Also, from where I sit epic is doing a phenomenal job providing training content. Of all of the game engines I have used, dating back to gamebryo, nobody has put this much time and money into providing documentation and training content.

glad plank
#

I agree with @sly elk . Would add that the engine features just keep on coming!

sly elk
#

also, there are likely developers here who know rift or vive sales numbers but are NDAd on them. It would be nice if that data was more easily available.

sly elk
#

that is some good news

#

I wish valve would make the bar for entry onto steam a lot higher. I think the last thing indie devs needed was a lot more noise and cynicism

wicked oak
#

prototype starts to actually be something

#

ive made the autogenerator also export collision geometry (per poly collision, but WAY simpler than the display mesh)

mighty carbon
#

@sly elk it's good and then it's not. By the same token, they might just start banning games that use assets from asset stores.

#

@wicked oak is this for PSVR ?

wicked oak
#

will be

glossy agate
#

Wow, what button did you press to jump to full screen?

wicked oak
#

alt enter

alpine torrent
#

@mighty carbon Samsung joined to WIndows mixed reality HMD what you posted earlier

mighty carbon
#

yeah, they did.. but, $499 with crappy controllers and no Steam VR support ?!

#

btw, Gear VR is done.. Carmack was asking what people want to see improved, but only in software

#

which makes me think that no more Gear VR evolution is going to happen

alpine torrent
#

Samsung HMD Odyssey

mighty carbon
#

yeh

mighty carbon
#

Microsoft acquired Altspace VR

sly elk
#

@mighty carbon selfishly that would be good for me. I doubt they would do that though.

alpine torrent
#

Stream VR preview on it @mighty carbon

mighty carbon
#

@sly elk that doesn't solve the saturation issue. There are tons of games being released with all custom assets and those aren't any better than store bought assets (and the games are also designed to rake up money from trading cards)

real needle
#

@mighty carbon the drift was mainly about google cardboard ๐Ÿ˜›

slim raft
#

So many 'VR Experiences' but so little legitimate games coming from the AAA guys

glossy agate
#

This holiday season we are getting a few good remasters. Doom, FO4, Skyrim. Less risky than an all new IP.

#

Hopefully they are super good so it drives HMD sales higher

alpine torrent
#

getting Windows Mixed reality HMD for holiday time it's good timing if those remasters are coming

glossy agate
#

Anyone here have the windows HMD dev kit?

dusk vigil
#

Getting it to our dev hub tomorrow : )

mighty carbon
#

@real needle have you tried Gear VR?

glossy agate
#

Cool. I'm curious if the controller transform is the same as touch or vive. Hopefully it's similar to one so we don't need to make a 3rd gripping option.

dusk vigil
#

Will try to remember to report back on it if nobody else can tell

real needle
#

I have at best buy, was pretty cool ๐Ÿ˜›

mighty carbon
#

aye.. rotational tracking is spot on, as good as Rift/Vive

#

mobile VR is shit anyway

#

(at least the way it's done today)

uneven moon
#

how do you guys usually implement a setting for a 'left handed' option?

#

if the gun is a part of my pawn, can I just invert the hands or something?

iron talon
#

does anyone know of good tutorials for VR Radial menus with vive?

uneven moon
#

@tired tree That suggestion you gave me earlier with the LookAt for the hand -- I need to snap the hand mesh to the gun so it looks like you hold it firmly, no mesh overlaps

#

as rotation happens

hollow socket
#

Hi guys, is there people familiar with the VR expansion plugin for UE4 here?

tired tree
#

@uneven moon then attach the hand mesh to the gun not to the controller during the secondary grip

#

@hollow socket yeah...I made it, whats up

hollow socket
#

Haha well that was unexpected

uneven moon
#

@tired tree I tried attaching the left hand mesh to the gun via a socket but now it makes my gun spin around while this is active. When I let go, the gun snaps back into place

#

I tried detaching the hand from the left motion controller so it can stick while the left MC controls the aim, then on let go, re-attach

#

still spins around

tired tree
#

are you rotating towards the hand or the mesh

uneven moon
#

the gun itself?

tired tree
#

yeah, are you rotating the gun towards the motion controller

#

or the mesh for the hand

uneven moon
#

ah

#

the hand

#

that makes sense haha

tired tree
#

thar be your problem

uneven moon
#

lemme try fix brb

#

Perfect

#

Thank you

mighty carbon
real needle
#

@mighty carbon Looks nothing like Battle Royale...

#

Last Man Standing Sure

mighty carbon
#

I simply posted the link.. It's not my game ๐Ÿ˜‰

real needle
#

And I simply commented on the link, not my game ๐Ÿ˜‰

#

I appreciate it though, you've found many new VR games I wouldn't have noticed

#

How's your game coming along?

mighty carbon
#

writing story/lore

#

pre-production will take some time

#

(and in between I poke at AI)

tired tree
#

Their example level is a straight rip of a CSGO gun game map

glossy agate
#

Was about to say, with a map this size its more like free for all deathmatch instead of BR

mighty carbon
#

what BR then? I thought it's just like LMS but with a lot more people

glossy agate
#

Couldn't find where it said how many players. Seems like people like BR for the stage gameplay ie early game strategy is a lot different from late game. Not sure how that would work in a map this size. Hopefully they can pull it off though beause BR in VR would be sweet.

#

Wonder if ARK park is going to make it out this year still. Looks like they are trying to push the graphics pretty far.

uneven moon
#

I've been thinking about a Rust-like battle royale in VR

#

Would be cool

#

Multiplayer pop is too low in VR still, I feel

glossy agate
#

Yeah. Anything longer than 5 min without drop in MP is too long I think. BR needs a full roster and no late joins though so I think it will be an uphill battle for now. Guess we will see.

tired tree
#

wow that loading lag

noble charm
#

That looks terrible

glossy agate
#

Out of everything they could have done with halo vr they chose that haha. Out of touch

#

I hope that is just a tutorial or something for it.

sturdy coral
#

lol, Halo: hogan's alley?

full junco
#

@mighty carbon ha, now you're saying mobile vr is "shit"? when someone said something like that a year ago you abundantly disagreed ๐Ÿ™ƒ

full junco
#

reviews for that samsung headset seem to be really good

#

"Should Oculus Rift and HTC Vive be worried? HTC Vive may be safe for now because of its industry-leading room-scale tracking, but Oculus Rift may want to watch its back. We find the Odyssey to be more comfortable to wear than Rift, and this editor didnโ€™t experience the nausea she usually does after just a few minutes in Riftโ€™s worlds."

#

quite interesting:

#

"By the way, don't expect these headsets to work with your shiny new Xbox. Microsoft won't talk about that at all: "Over time we'll absolutely have it in Xbox, but we have nothing to announce in that space today," said Microsoft's Alex Kipman."

uneven moon
#

If anyone has any references for PSVR move controller mapping in UE4

#

Please PM it to me

full junco
#

until you find someone who has any references for that you have already figured it out yourself by just printing out on every button press ๐Ÿ˜„

vagrant mantle
#

Anyway to toggle between Vr and Screen?

full junco
#

you would have to program that yourself

vagrant mantle
#

@full junco any commands, hmd enable, stereo on doesn't work

full junco
#

you can do stereo off and that will turn off vr

#

obviously not in the vr preview, only in a standalone game

vagrant mantle
#

@full junco for vr enable what command?

full junco
#

stereo on

eternal inlet
#

anyone know why i can't bind an event Hit?

#

mainly curious... i guess i can use OnComponentHit

#

hmm... next issue... i thought i could delegate a OnComponentHit down to another component i made myself?

#

apparently it does not allow me with a err msg about delegate (ref) is not compatible with delegate

#

sorry if this seem mega noob, but is it not possible to get an event on a component i make, but recieve an event from another component in the actor?

#

so if i have an actor with comp1 and comp2 and comp1 fires a OnComponentHit, can i not tell Comp2 without hooking up something in the Actor?

#

obviously i wanted to just be able to plugnplay a component, so that if a staticmeshcomponent fires oncomponenthit, my own component would play a sound

#

but this is ok:

#

wtf?

#

omg nevermind

#

missed the byref "Hit" ๐Ÿ˜‚

real needle
#

@vagrant mantle vr.enablestereo true

#

but as john said, it only works outside of editor

mighty carbon
#

@full junco mobile VR is shit because I am not making any money there :p

vagrant mantle
#

@real needle okay, i have to package it

real needle
#

@vagrant mantle What are you trying to do?

#

I just have a certain pawn to play when I hit "play in editor", and another when I hit "play in vr"

vagrant mantle
#

@real needle im trying tomorrow switch between screen and vr

#

*to

real needle
#

it helps debugging outside of stereo

vagrant mantle
#

@real needle can't get you

wild mauve
#

how's it going today my dudes

mighty carbon
#

Nvidia doesn't want to make and release VWorks using 4.17.2

#

"Do it yourself"

wild mauve
#

probably because 4.18 is out?

#

Do it yourself.

#

You aren't entitled to anything?

dusk vigil
#

@vagrant mantle Set it up like this : Have a PlayerStart actor in the scene. When game starts, if we launch in VR preview, spawn the VR pawn, if in normal editor mode, spawn an fps character

#

Depending on your game, you can get pretty far with this to speed up development iteration, skipping out on having to put headset on and off all the time. If your game is heavily dependent on hand interactions, you might need to build something around that. Depends

mighty carbon
#

I have better things to do @wild mauve than fck around with VRWorks code merging

wild mauve
#

I'm sure they do as well.

mighty carbon
#

I am sure they want people to use their stuff and since not too many people use it, they can do better job by offering out of the box solutions

wild mauve
#

I'm sure they're doing fine without the people that can't merge it themselves into a version that isn't even the latest version.

mighty carbon
#

I am sure you have a lot of attitude

wild mauve
#

Everytime I see you here it's always some other entitled bullshit comment. "Mobile VR isn't doing well because I'M not making money off of it".

dusk vigil
#

Geez guys chill out, it's a bummer to see bad vibes flaring up

wild mauve
#

Is it hard for you to just not be respectful to people?

mighty carbon
#

Everytime I see you here I am thinking what a dickhead

#

No, no respect for you

wild mauve
#

You're just mad that someone wont cater to your bullshit. Don't confuse that with being a dickhead.

#

And you're seriously not worth the time, consider the argument over.

vagrant mantle
#

@dusk vigil i want to switch between Vr and normal camera, if I press space it has to toggle

wild mauve
#

Just add a command to space to spawn in and possess the spawned in pawn whenever you press it

real needle
#

@vagrant mantle Destroy current pawn -> Spawn Actor from Class (your flat pawn if you're playing in VR) - > Possess new pawn

full junco
#

@mighty carbon let me comment on it too, there is really absolutely no need for nvidia to update their branch to some hotfix. merging the hotfix stuff it yourself is easy and usually doesn't break anything. their branches aren't meant for people that don't know c++

mighty carbon
#

There is no need for NVIDIA to release anything

#

Yet they do, half ass

full junco
#

if they would spend time updating to hotfixes and whatever they would also do stuff like make binaries available. but they expect people to compile it manually

mighty carbon
#

I can built it from source

#

But like I said, it's not some FOSS engine you toss out there and let people mess with (and we all know what happens after a while - those engines get abandoned)

#

So, it's just Nvidia being lazy

full junco
#

what you can and should criticise about the nvidia branches is how long they need to update them. its always like, when 4.X is almost released they only just updated their VRWorks for 4.X-1

mighty carbon
#

Do you use nvidia branch?

vagrant mantle
#

@real needle what you mean by flat pawn?

real needle
#

Not VR

vagrant mantle
#

Tried but that doesn't work

real needle
#

vr.enablestereo true/false

full junco
#

@mighty carbon I don't use it, because until recently I didn't have an nvidia GPU to test it on, and if I would use it for my game I would always have to test it with AMD too to make sure it still works there, and I consider that too much work at the moment

vagrant mantle
#

@real needle tried that also

real needle
#

@vagrant mantle In packaged/right click .uproject and hit launch game?

vagrant mantle
#

@real needle unpossesed and destroyed and spawned a new player,after those console commands, doesn't work

real needle
#

what doesn't work

#

you don't have to unpossess

#

possessing a new pawn automatically unpossesses the old

mighty carbon
#

@full junco so, if you don't use it, why do you even care about my whinings about the subject and "defend" lazy ass Nvidia ? ๐Ÿ˜‚ ๐Ÿ˜

full junco
#

@mighty carbon well I saw you talked with Supa about it and I pretty much agreed with what he said, and I just wanted to show you that he's not the only one telling you that ๐Ÿ˜›

mighty carbon
#

He/she is a dickhead

full junco
#

and in the end we can be happy that nvidia has any branches at all

mighty carbon
#

Plain and simple

full junco
#

I don't know him, but based on what he said to you, he's very reasonable ๐Ÿ˜›

mighty carbon
#

I didn't start this bs.

#

Not in my book

full junco
#

well I don't have to read your book

mighty carbon
#

When you release your game 1.0 and tell people to update it to 1.0.1 by merging changes themselves, I'd love to see if anyone will side with you on that decision.

full junco
#

๐Ÿค”

#

I don't see why I should expect my players to know c++. doesn't make any sense. the majority of gamers aren't programmers

mighty carbon
#

Nowadays a lot of game devs aren't C++ programmers either

#

By the same token Blueprint should be removed since everyone should know C++ by your standards

full junco
#

nvidia didn't invent blueprint. in nvidias "world view", game devs are still programmers

mighty carbon
#

Then they lag behind time

full junco
#

they seem to be quite successfull with it. doesn't hurt them financially

mighty carbon
#

They make no money off UE4 forks

full junco
#

in the end, they do

#

when a game runs 30% faster on nvidia GPUs, people who want to play that game are way more likely to buy a Nvidia GPU

mighty carbon
#

It's their side gig, which they don't really care about. Thus - being lazy about fully committing.

tired tree
#

NVIDIA doesn't cater the the low end indie game dev and VRWorks is an SDK that they are only implementing because it needs solid implementation references

#

historically they have never done everything for everyone with their research projects

mighty carbon
#

I haven't seen any UE4 game in retail that uses Nvidia branch.

full junco
#

ideally you don't even notice that they use it

#

player should just think "game runs well"

mighty carbon
#

๐Ÿ™„

tired tree
#

I mean, if you want to complain, complain about Unity having a fully valve created OpenVR integration and UE4 having a partial one

full junco
#

how is that related to the vrworks stuff? ๐Ÿ˜„

tired tree
#

its not

#

but if he wants to take affront about engine integration, that is far worse than VRworks

full junco
#

well he doesn't use steamvr so he doesn't care that much about it I guess

mighty carbon
#

Not yet at least

#

But I am not using Unity

full junco
#

when most headsets support eye tracking, engines definitely need to have their own "multi res shading" implemented

wicked oak
#

multires shading is not "fixed"

#

so you can do eye tracking with that

#

essentially its just "broadcasting" the geometry to multiple viewports at once

full junco
#

@wicked oak I mean't that once the benefit of it becomes even bigger due to eye tracking, engines will have to implement it themselve

wicked oak
#

yup

#

i could really use it on ps4

#

free performance

#

and i dont care about compatibility

vagrant mantle
#

@real needle worked, Thanks, i didn't include the level to package

real needle
#

@vagrant mantle Np! Make something awesome

mighty carbon
#

digging AI and wondering what drives anims.. I figured that animBP drives anims based on values of variables set by BT (in the blackboard). Is that correct ?

wicked oak
#

no

#

@mighty carbon the anim blueprint should only read from the state of the "character"

#

if you want special animations, you just use a montage

#

the way i do it when i need complex stuff, is that i have that stuff on the character

#

and the animBP reads from it

#

its best to have little logic on the animBP

#

let it be about animations

mighty carbon
#

right, but if I want AI to crouch, how do I kick in crouching anim ?

wicked oak
#

bool bIsCrouched in the character

#

the anim BP reads from that variable, and copies it

#

then you have some kind of state machine or "blend poses by bool"

mighty carbon
#

what I am wondering about is how do I set that var on the character from BT

#

AI seems bizarrely complex due to BT being comprised of services, decorators and tasks and with blackboard on top of that :/

wicked oak
#

dont use BT if you dont need it

mighty carbon
#

hmm.. why not ?

wicked oak
#

becouse it can be more complicated than not using it

mighty carbon
#

it seems that all issues people complain about with AI arise from the fact that engine is designed to be used certain way, but people try using BP instead

wicked oak
#

the behavior tree is the best when your AI does sequences of actions

mighty carbon
#

from what I understand, you can run task and not wait until it finishes and start next one in line

wicked oak
#

you can abort/cancel tasks

mighty carbon
#

well, that's why I am wondering why would you not use BT/blackboard

tired tree
#

generally you should

#

its the most flexible out of the box solution

#

he is saying for "running straight and fire" logic it might be overkill

mighty carbon
#

I see

#

ok, so say I am using BT/blackboard.. How does that connect to animBP and its states/variables ?

tired tree
#

you can set variables in tasks

#

or a State enum

mighty carbon
#

I see.. And those are the same variables as in animBP ? (or do I connect them via blackboard)

#
wicked oak
#

sad they say that there is screendor

#

seems that the pixel fill is low

#

why didnt they add a difussor then?

mighty carbon
#

interesting (related to discussion we had about root motion)

tired tree
#

yeah? what about it

#

I told you it was implemented, but that it was not fun

#

you are going to need animations for everything

#

and to control the animations instead of the character itself

glad plank
#

Root motion is awesome! Love making root motion controllers, but we have an animator... ๐Ÿ˜›

tired tree
#

yeah you pretty much need one for that

glad plank
#

Or spend a lot of money on asset store anims

mighty carbon
#

I assume using root motion eliminates all these issues with abrupt turns on the path nodes, etc.

#

also all mocap stuff has root motion by default

#

(I am thinking AI here, not player)

wicked oak
#

it doesnt

tired tree
#

it doesn't solve abrupt turns, you would have to deal with playing the turn animation and then the walking/running one again and going forward

wicked oak
#

root motion cant do pathfinding in fact

#

you would have to implement your own path following component

#

that essentially just sets some variables on the character

#

and then the anim tree runs the animations

#

like "movement Vector" or "rotation angle"

tired tree
#

yeah you would have to follow the navigation nodes yourself

#

though it would let you smooth out the turns, since you can manually transition between nodes then

mighty carbon
#

"Does it work with AI MoveTo?
No, MoveTo moves the actual pawn. But there is an alternative: just use the underlying path nodes to orient the agent's direction. Remember, as long as the pawn is facing the correct direction, root motion will drive it toward that location. So instead of using AI MoveTo, you can use FindPathToActor to generate a list of path points, and use them to orient the direction the agent is facing."

#

is it much harder than MoveTo ?

#

I guess there is no way around sinking time into R&D

#

(while it's not an issue with code, it might be an issue with making anims)

tired tree
#

AIMoveTo does all of that for you

#

you would basically be re-creating it yourself

#

but a little different, since all movement is done by playing animations

#

you would need to know the amount of movement in the animation

#

to apply it for the right time to go as far as you want

mighty carbon
#

I see

#

Hmm.. I guess I'll go route of least resistance for now ๐Ÿ˜›

glad plank
#

I would do it a bit differently

#

I Would do location checks

#

and play the animation until it got close enough, then blend to a stop state

#

So set a location, if angle greater than certain angle play turn animation, if distance to location greater than minimum play animation.

#

I assume you would have to do your own pathing node points or something

mighty carbon
#

well, I still need to figure out basics before I get into that stuff

#

meh, I e-mail the author and I didn't really like response I got. Something along the lines "no support"..

#

not willing to trow money on assets that will not have support provided

glad plank
#

He released a V2..IDK, for me its less about support and more about work I don't have to do myself

#

I never just us an asset

#

I completely convert it to how I need it. ๐Ÿ˜›

#

What was he not going to support? AI?

mighty carbon
#

that's cool.. I don't have such capacity (mostly because I don't have much time nowadays)

#

AI isn't part of the package, why would he support AI?

#

I was asking about how to integrate it into my project

glad plank
#

Ahh

mighty carbon
#

just watched v2 video.. looks like he made it more modular

#

probably worth messing with it even if it will be hard

#

is there documentation available for that package ?

#

looks like no docs available

glad plank
#

no clue

mighty carbon
#

4.18p3 is out!

sturdy coral
#

@mighty carbon raw data uses vr works, and so do some others. I have already told you it is almost always a clean merge. just checkout the vr works branch and type git merge 4.17.2-release. if it doesn't work you aren't out anything

#

it would be more complicated if nvidia did the merge themselves

#

because if they make new changes afterwards, you'd have more trouble pulling just ranges of their changes, etc.

mighty carbon
#

have you messed with it already ?

sturdy coral
#

I'm still on 4.16 for the moment

tired tree
#

yeah hot fixes aren't a pain

#

its manual engine upgrades if you can't wait for their version that is the pain....but they have been far more consistant with it

#

it used to take 2 versions or more to catch up to one so it was always multiple releases behind

mighty carbon
#

has anyone tried new cloth sim in UE4 in VR ?

alpine torrent
glad plank
#

Is there anyway to get the performance statistic screens centered in VR?(edited)
Whenever I try to use them half of the screen is out of my vision

tired tree
#

ouuuuuuuuuuuuch that video

wild mauve
#

@glad plank if you figure that out I would love to know as well.

glad plank
#

@tired tree any idea how to fix the performance statistic screens?

sturdy coral
#

@glad plank you can edit GetOrthographicProjection in the hmd cpp file

#

I think that is the name of it

glossy agate
#

Is roomscale supported on the windows HMDs?

#

I see inside out, but not sure how roomscale would actually work unless it uses the cameras or something

sturdy coral
#

@glossy agate it should, hololens does

alpine torrent
#

@glossy agate from 360 to world scale so as roomscale is after 360

eternal inlet
#

anyone know if it's possible to send sound out through the HMD and some out through a different audio output like the regular monitor or a different pair of headphones?

sturdy coral
#

@eternal inlet IHeadMountedDisplayModule has virtual FString GetAudioOutputDevice() that you can override

#

but it changes all audio

#

I think you would need source changes to do some audio in one place, some in another

eternal inlet
#

yeah ok

#

i had a feeling it wasn't supported

sturdy coral
#

the player controller has the listener position

#

for e.g. splitscreen, but they get downmixed onto the same device

eternal inlet
#

i see...

#

i was also wondering how they handled sound for splitscreen

#

where two people can be each their place

mighty carbon
lusty ledge
#

anyone get blueprint nativization to work with steamvr? i keep getting warnings that "NativizedAssets doesn't list plugin SteamVR as a dependency"

#

but NativizedAssets.uplugin is dynamically generated and I can't add things to that .uplugin file

mighty carbon
#

do you think 4.19 previews will come out before Christmas ?

granite jacinth
#

^ VR?

#

But possibly. Even though it would be strange given their one per quarter release trend lately.

mighty carbon
#

well, we already have 4.18p3, which I thought was a fast delivery

#

I am sure all those optimizations would also work in VR

wicked oak
#

yeah, 4.18 is about to drop

#

does oculus still have forward renderer special builds?

#

i might do one with that version for DWVR oculus store

mighty carbon
#

no, no special forward renderer builds

granite jacinth
#

Aye, but

#

I think Epic is taking a sebatical

#

with 4.19 until post holidays

#

They should anyway

mighty carbon
#

I just hope 4.18 comes out less buggy than 4.17

thorny niche
#

Hey everyone! Need some GearVR help. I have a problem with performance loss on rebuilding and relaunching project several times in a row. Looks like some sort of cache getting full or smth. UE4's 'Profiler' tool shows Stat Thread to jump sky high hence a lot of CPU stalls. Is anyone familiar with stuff like this? Right now to fix this I try deleting it from the phone and building project again -> reboot phone -> 'Optimization' function with RAM cleaning. It usually helps but result is not 100%. I wonder is there some sort of Task Manager app that can clean all the RAM and cache files time to time?

mighty carbon
#

what phone ?

thorny niche
#

And no, it's not about overheating, phone temp might be same, but build is dirty

#

S6

mighty carbon
#

and what UE4 ?

thorny niche
#

4.15 atm

mighty carbon
#

get 4.16

#

I had a lot of issues with 4.15

thorny niche
#

I'll try it out! Thanks!

mighty carbon
#

np

#

also developers builds have some issues when running them on Gear VR. However once you build shipping build, it should run fine.

#

Make sure you use Multiview and Multiview Direct

glossy agate
#

For testing VR mp is it worth packaging and sending around in a zip or should I just grab a steam app id now and do all the testing through that from the start?

wicked oak
#

grab steam app id

#

that way you can setup steamworks

#

and use sessions to get a server browser

#

thats what i did

sturdy coral
#

yeah, and it is way easier to get updates to people and not have to deal with firewall issues

tired tree
#

heh, those 4.19 character movement optimizations were something I did last month except a little more extreme

#

I was considering pull requesting them

sturdy coral
#

@tired tree which ones?

#

is it in trello or something or did you just notice on github?

copper stag
#

Help! I'm trying to create a movie in a VR scene that I created and when I try to create the movie out of the matinee, it starts recording the VR camera and not the cinematic camera that I've included in the matinee. What have I done wrong?

sturdy coral
#

ah I see lots of character movement bandwidth optimizations and stuff in master. fortnite battle royale related I guess?

trail shale
#

So I finally had the money to build a VR capable machine and I have my longtime Unreal Project ready to make into VR...anyone recommend which videos I should start learning?

glossy agate
#

The VR template

#

Like you already have a made game you are porting to VR?

trail shale
#

basically

#

there are a view tutorials from Unreal but nothing that seems from the basic

#

Do I basically need to make a new VR project and then migrate the BP's from it to my existing project?

glossy agate
#

You can dissect the vr template and add it or build it in your game. I don't know what kind of game you are making though

tired tree
#

@sturdy coral yeah, specifically where if some variables are empty it uses a different RPC to send saved movement

#

before every time it used the same RPC even if some of the variables had no data

#

I think I have 3 different RPC paths and a struct that netserializes only what has data for a few things though, i'm passing a ton more through for the VR suppor

trail shale
#

@glossy agate http://www.thelivesimulator.com VR training for First Responders

tired tree
#

thats a neat use case

glossy agate
#

Yeah, looks like you guys already have some stuff setup for VR?

#

in the pictures it looks like at least

dusk vigil
#

Hmm... the plane hitting a building, the worn metallic sidelines... I have to say it doesnt look so attractive from clinical business perspective

#

"Help Firefighter build VR Simulator"
-> "Help a Firefighter to build a VR Simulation of Emergency Events"

full junco
#

@mighty carbon I do think 4.19 preview will come out before Christmas, yeah

mighty carbon
#

would be cool ๐Ÿ˜ƒ

full junco
#

@mighty carbon but why would you care now about when 4.19 is available? I don't think you know about any cool feature in 4.19 yet?

mighty carbon
#

Besides all the optimizations from Fortnite BR, I don't

#

do you have some info about what's coming to 4.19 ?

full junco
#

not yet

#

though I just see there was a dev-rendering merge to master 7 hours ago! I love those

#

so thats 4.19 stuff

#

now I have to read through it

tired tree
#

just did

#

mostly a bunch of small bug fixes

full junco
#

well they finally made point lights use lumen for intensity

mighty carbon
#

I wonder if they improved performance with dynamic point lights/shadows in VR

full junco
#

not really exiting, but its something that was planned for quite a while, lumen for all intensities

#

@mighty carbon maybe don't wonder and just read through it yourself? ๐Ÿ˜„

mighty carbon
#

watching anim tutorials

granite jacinth
#

@full junco How do you know what in Master is in 4.19 though?

full junco
#

everything thats now in master is in 4.19

granite jacinth
#

Until when?

full junco
#

until they branch 4.19

granite jacinth
#

staging?

full junco
#

you see when master is set to 4.20

granite jacinth
#

Yeah but I don't even see it 4.19

hollow socket
#

evening!

full junco
#

@granite jacinth it is one commit that says something like "bump version number to 4.19". that commit was done like 1-2 months ago, when 4.18 was branched. from that time on, everthing that was added to master was 4.19 and not 4.18

granite jacinth
#

Man

#

So many commits

#

I wouldn't have found that unless someone told me ๐Ÿ˜‰

full junco
#

found what?

#

@tired tree just finished reading through the commit, its not just a bunch of small bug fixes

#

its a lot of vulkan and metal stuff, as usual

#

good vulkan stuff though, finally focused on improving performance

tired tree
#

i saw some vulkan rollbacks

#

don't care about the metal much tbh, though ti was most of the changes

full junco
#

and the disappointing part of this commit is that, just as in the last one, Daniel just worked on static lighting...

#

thats so frustrating to see

mighty carbon
#

how come?

#

static lighting needs love too

uneven moon
#

Can I make a sniper scope without using a scene capture? It's very costly

full junco
#

@mighty carbon for arch viz its nice, but games where everything has to be static are definitely not the future...

mighty carbon
#

that's what I was saying when I started with UE 4.7 or so..

#

๐Ÿ˜‰

#

and we are still in the realm of static lighting

full junco
#

with fortnite currently being epics main focus it would make sense for them to improve the dynamic side of the engine and not the static one

sturdy coral
#

@full junco PUBG is lots of static and probably makes them more money than fortnite for now

full junco
#

I don't think so

#

epic said 7 million players in fortnite

#

from pupg they only get 5%, from fortnite they get everything

sturdy coral
#

@full junco people are probably mostly playing fortnite battle royale, not the main fortnite, and it hasn't really been monetized yet

#

it is free

full junco
#

oh it is?

sturdy coral
#

yeah

full junco
#

I thought they wanted to make it free in the future

sturdy coral
#

eventually it will probably be a bigger money maker from them, assuming there isn't tons of competition from more and more battle royales

#

they want to make the main fortnite free in the future

#

but battle royale is free right now

full junco
#

ah

sturdy coral
#
SteamSpy - All the data about Steam games

Steam Spy automatically gathers data from Steam user profiles, analyzes it and presents in simple, yet beautiful, manner. Steam Spy is designed to be helpful for indie developers, journalists, students and all parties interested in PC gaming and its current state of affairs.

#

13 million owners at $30

full junco
#

19.5 million dollars for epic

sturdy coral
#

but yeah fortnite battle royale is all dynamic like regular fornite

#

and does time of day over the course of the battle

#

they did have some optimizations of the peak memory usage of the distance field stuff

mighty carbon
#

but that's directional light, which is as I recall works pretty well on Gear VR too

sturdy coral
#

but for consoles only I think

full junco
#

memory usage is pretty irrelevant

#

unless people have a gtx 970

sturdy coral
#

there are lots of weird console only optimizations they can make I guess since the memory between CPU and GPU is unified

full junco
#

well like half a year ago they made distance field shadows twice as fast, that was for fortnite

sturdy coral
#

yeah and fortnite has been in development for so long I think most distance field stuff was put in for it

#

even before kite demo

full junco
#

I wouldn't say that

sturdy coral
#

yeah I can't remember for sure, they just almost always use fortnite screenshots for DFAO and stuff

#

landscape GI was for kite but I don't think that really uses distance fields

#

and I'm not sure about daniel's work on distance field GI

#

what that was targeted at

full junco
#

landscape GI is height field GI, no distance fields

#

and it's broken

#

stopped working around 4.13 or so and was never fixed again

#

I think

sturdy coral
#

yeah the last time I got it working it had to have a ton of care to make it work

#

but I think that was before 4.13

#

dynamic wise what I would like to see is better shadow stuff like doom

#

not just caching when there are no movables, but caching even when there are and then composing movables over it

full junco
#

@sturdy coral yeah dynamic shadow performance is something they should really work on

#

but I think they are kinda going the "just just DF shadows" route

#

so they don't really care about CSM performance that much any more

sturdy coral
#

yeah, and df/raycast shadows don't work for landscapes :/

#

so you have to turn far shadows on for them and burn a shitload of draw calls

#

or have simplygon I guess and bake all the really far stuff into a few meshes

#

I haven't needed dynamic time of day, so I've just been baking landscape with low res and having cascaded up close (same as pubg)

full junco
#

yeah far shadows are crazy expensive...

#

I do use them in my game

#

I do other stuff to reduce draw calls, but even without that many draw calls, far shadows are still expensive

#

the issue with DF shadows is not only that they don't work for landscapes, but also not for any procedural geometry

#

always have to be calculated in the editor

#

so while DF shadows are quite nice regarding dynamic shadow performance, they are not as "dynamic" as CSM

sturdy coral
#

if you put the sun straight overhead you can get by with no landscape shadows, since a heightmap is oriented towards the sun, but then you aren't using time of day anyway

#

and sun directly overhead looks kind of bland

full junco
#

yeah thats looking bad

#

and I also really like dynamic time of day

#

one issue is UE4 is not really checking for what meshes are actually visible and if they need to receive shadows and if other meshes need to cast shadows

#

so when you are looking at the ground and only 1 floor mesh is visible, a lot of stuff around you casts shadows on objects that aren't in your view

#

if a mesh casts shadows only on meshes that you don't see, it doesn't need to be renderer in the shadow map, but UE4 still renders it because it doesn't check for that

sturdy coral
#

ah yeah, I haven't been sure how they handle that

#

it should do the union of your frustum and the orthographic/rectangular frustum from the sun

#

and then cull anything outside of that

full junco
#

I did my own optimization for that, I trace the bounding boxes of all visible meshes against all other meshes and if a mesh is never hit by the bounding box trace of any visible mesh, I disable shadow casting for it

#

I do that in a seperate thread on the CPU

#

takes 1-2 ms on the CPU but can improve GPU performance by 1-2 ms

sturdy coral
#

I saw some paper where they take large areas that are shadowed by a landscape

#

and turn off all shadows for any objects within that volume

#

since they are all guaranteed to already be shadowed by the landscape

#

I think they got huge gains from it especially at times of day like sunset

full junco
#

ok

#

sunset is hardest for shadows, yeah

mighty carbon
#

but not sunrise ?

full junco
#

that too

#

its same of course

mighty carbon
#

๐Ÿ˜ƒ

full junco
#

there my optimization also doesn't really do anything since when the light comes in horizontally the bounding boxes of meshes pretty much all hit each other

mighty carbon
#

so, basically there is no good shadow solution for large worlds right now ?

#

(that works with Landscape and in VR)

full junco
#

yes

mighty carbon
#

๐Ÿ˜ฆ

sturdy coral
#

baked

full junco
#

baking large worlds also isnt a good solution

#

motorsep doesnt have 128GB RAM ๐Ÿ˜„

sturdy coral
#

hah yeah I've gotten some crashes doing it

#

but, landscape wise baked at a really low res looks decent enough since the shadow maps are signed distance fields

#

and then still use cascaded shadow maps as you get close

full junco
#

well, for static worlds that works

mighty carbon
#

I am guessing you don't bake static shadows cast by trees/rocks/etc.

#

you only bake skylight and indirect lighting with AO

sturdy coral
#

no, you bake everything

#

but I haven't done it on maps with foliage, not because of the foliage baking to the landscape, but the shadows baking to the foliage

#

paragon uses that though, it has a lot of foliage but isn't a huge map

#

when you get close the cascaded shadow maps fade in and you get wind movement on the nearby foliage shadows

#

far away and it is baked with no motion

#

same setup as the infinity blade grasslands example

full junco
#

well for such games it works

sturdy coral
#

yeah, just time of day is the big limitation

#

sunset would be a challenge too, just because even the near cascades may get a ton of draw calls

#

vulkan and ryzen will save us

full junco
#

ha yeah, lets hope so

jade goblet
#

hey guys has any one done vr editor mode in daydream

#

i was wandering if it were possible because if so i am getting one like NOW

mighty carbon
#

no

jade goblet
#

darn

upbeat meteor
#

Hey, y'all. Can anyone point me to a VERY clear, VERY current configuration for UE4.17 to send a project to a Gear VR (Note 5, or Note 8). I've got most of a build happening, but I keep running into errors that are very vague. Thanks.

wicked oak
#

@muchcharles#2724 that for shadowing wont work

#

the easy reason. Imagine you are on the shadow of a building

#

and you look completely on the opposite direction

#

if shadows were related to frustrum, that means that the shadow from the huge building wont be there

real needle
#

Are y'all hyped? ๐Ÿ˜„

#

i sure am

#

(microsoft 'mixed reality' stuff i mean, and that sammy hmd)

river fractal
#

how would one detect if the hmd locked camera has collided and/or is on the other side of a staticmesh (wall)? I get how i would find out when the camera is overlapping with the mesh, but after it goes through to the other side im lost ๐Ÿ˜ฆ

full junco
#

@real needle whats there to be "hyped" about? sure, the resolution is very interesting, but the tracking isn't that great, so in the end the samsung hmd is on the same "awesomeness-level" like rift or vive

real needle
#

From what i understood the tracking was actually pretty good

#

and at the very least theres stuff happening in the 'vr market' so theres that to be hyped about

full junco
#

well it is good, just not as good as rift and vive and it doesn't track out of the FOV

#

so something like pulling a bow back out of your view won't work with that

real needle
#

the way i understood it its slightly better then the rift & vive experience wise, apart from that tracking outside fov (wich is bigger fov then the screens fov btw)

#

so pulling a bow back will work since your 1 hand will bscly be next to the headset so likelly still tracked

#

grabbing an arrow from your back is impossible tho

full junco
#

its not better experience-wise, its not just the tracking, the controllers are also mediocre compared to the touch controllers

real needle
#

so theyll definatlly need to improve on that for later versions (360' inside out tracking)

#

did hear that too, the controllers are not great

#

theyre not bad either tho

#

idk, i guess i see what your saying, theres still a bunch to be desired

#

but i like that theres [a] at least stuff happening in vr hardware [b] they start at alot more reasonably priced hmd's [c] that 'windows home'/cliffhouse actually has a use

full junco
#

I don't really think it has that much of a practical use

real needle
#

and im all for inside out tracking once they fix that issue with hands, so ill have to accept this in between step

#

why not?

full junco
#

not with this resolution, its still relatively low, you see every pixel

#

you won't do anything in VR that looks way nicer on a regular monitor

#

so its just a gimmick

real needle
#

disagreed ๐Ÿ˜›

#

vr is not a gimmick

#

but to truelly see that youll have to have some cool experiences inside it

#

the resolution is definatlly a downer

full junco
#

I didnt say VR is a gimmick, I said the "2d windows apps in VR" is a gimmick

real needle
#

but not a dealbreaker

#

hmm, i disagree with that aswell ๐Ÿ˜›

#

theres not enough '3d' (read; proper vr) apps yet, so having access to your 2d apps is worth something

full junco
#

for some reason you seem to believe in stuff the microsoft marketing wants you to think

#

but people won't use 2d apps in VR at the moment. just the keyboard alone is unbelievable slow to use in VR

real needle
#

depends on the app

#

like, photoshop doesnt need a keyboard

trail shale
#

@full junco John's right, it won't go mainstream until they have trackers on the keyboards and perhaps even on the hands, so you can actually see your keyboard

full junco
#

compared to a real keyboard any virtual keyboard you use with motion controllers is very slow

real needle
#

true true

#

but not all apps need keyboards

trail shale
#

The resolution on the headsets is too low as well

real needle
#

personally i dont see myself doing much in the microsoft home thing either

trail shale
#

I've tried working on photoshop and Unreal and the screendoor is just too much

real needle
#

but itll be nice as a overview

#

with some simple actions possible trough clicks

#

hmm

full junco
#

@trail shale using one of the windows headsets or with vive/rift?

real needle
#

oh well, 2 weeks from now we'll know more

trail shale
#

how much more is the resolution, it's not much I think, maaaybe 15% more?

full junco
#

it is 76% more

#

it is a very relevant increase in resolution, but it is still very low

trail shale
#

hm I would like to see it

real needle
#

also oled

trail shale
#

How do you get one?

real needle
#

wait 2 weeks

#

then buy it

full junco
#

@real needle rift and vive are also oled

#

so oled is nothing new

trail shale
#

How much is it?

full junco
#

and most of the windows headsets are not oled

#

only the samsung one is

#

the samsung one is $500

trail shale
#

ah

real needle
#

fair enough

#

vive n rift are amoled

#

and yea only that sammy one will have oled

full junco
#

the samsung one will likely also be amoled

#

samsung is king of amoled

real needle
#

still, for 500 bucks, im interested to see what will happen

#

tru

full junco
#

rift and vive also use samsung screens I think

real needle
#

prolly

full junco
#

it will be quite hard to sell at that price

real needle
#

at least the rift openly stated using their galaxy note <some number> screens at some point

trail shale
#

https://www.roadtovr.com/samsung-odyssey-windows-vr-mixed-reality-headset-hands-on-preview/
Granted, I was still able to see individual pixels, and the screen-door-effect (the space between the pixels) was more pronounced than I would have hoped given the resolution, quite possibly due to Samsungโ€™s unique PenTile subpixel arrangement.

full junco
#

when people can buy a rift for significantly less and get better tracking that isn't limited by FOV and from a brand that is way more established in VR, most people will take the rift

real needle
#

whats a rift these days? for the full room scale setup?

#

im actually not sure

full junco
#

$420 I think

#

or something similar

real needle
#

for me its around 600

#

euro's

#

vive 699

full junco
#

well it was significantly cheaper recently

#

(the rift)

#

499 euros

alpine torrent
#

yeah it was

#

like vive was cheaper

full junco
#

vive wasnt cheaper

alpine torrent
#

bundle was cheaper

full junco
#

what bundle was cheaper than what?

tired tree
#

think he means Vive was discounted, not cheaper than rift

real needle
#

either way, point is that samsung headset will be cheaper then rift & vive, and will have microsoft endorsing it, im guesstimating they will ship alot of them

full junco
#

the samsung headset is more expensive than the rift

real needle
#

especially if it works well with steam which it does

full junco
#

it "does" not. microsoft says it "will" before the end of the year

alpine torrent
#

how much rift is now?

real needle
#

look up the price for a rift + controllers + extra sensor so u can do roomscale

trail shale
#

599 i believe

full junco
#

@real needle controllers are always bundled with rift now

real needle
#

uhu, for the price blake mentions

full junco
#

and you dont really need the third sensor

real needle
#

u do for roomscale

full junco
#

no

trail shale
#

You're both right

real needle
#

k ๐Ÿ˜›

trail shale
#

They had more realibility with 3 sensors in the past

#

I'm not sure but I think they did fix some of the tracking issues so you can get it to work with 2

full junco
#

I do have a rift with 2 sensors

real needle
#

k, they must be slappin on some extra money for non muricans then

full junco
#

and set it up for roomscale

trail shale
#

Anyone here a guru with converting projects in Unreal to VR?

real needle
#

whats there to convert

#

well apart from some effects etc most likelly

trail shale
#

long story short I'm a Firefighter/Paramedic who was devleoping a VR simulator for First Responders http://www.thelivesimulator.com) - I made an alpha version before I had enough money to build VR rig

#

I just finished building VR rig yesterday

real needle
#

john thats good to hear i guess

trail shale
#

so now I have to start from scratch in terms of the VR stuff in Unreal

#

tutorials are over a year old so I wanted to know where to begin

alpine torrent
#

what windows mixed reality headsets are using tracking is light dots in controller plus bluetooth 4.0 in the controllers

trail shale
#

For example, do I have to play with VR template or can I access it from my project already?

full junco
#

@real needle I wouldn't recommend to do roomscale with 2 sensors though if you want to play a lot with it. its just fine for me since I don't need perfect tracking, I only bought the rift ro make sure my game is compatible with it, but I prefer the vive a lot for regular development

real needle
#

and thats sorta the issue ๐Ÿ˜› so u do need that 3rd sensor ^^

#

meaning youll end up at idk, 450?

#

not that far off from 499 for a newer system

#

i see how in your case the rift is cheaper but i still think alot of ppl who would buy a new headset would go for a newer one even if a little pricier

full junco
#

well the tracking with 2 sensors will probably still be better than the tracking you get with the windows headsets

real needle
#

@trail shale depending on how much you have built so far, you can either start a new first person template based project and move assets over, or tweak your current project to incorporate vr

full junco
#

its just that for vive level of tracking you need a third sensor

alpine torrent
#

@full junco have you study bluetooth 4.0 device to tracking where they are?

real needle
#

and for as far as i can tell at this point (based on early reports*) the tracking of that windows stuff matches vive, par the hands behind back issue

full junco
#

@alpine torrent as usual I find it very hard to understand what you say

#

@real needle eh no, absolutely not

#

vive tracking is 100% perfect

#

reviews for the windows headsets say its like 98% perfect

real needle
#

hmkay ๐Ÿ˜›

#

vive tracking isnt 100% btw

#

99 at best

full junco
#

the vive tracks with 1000 hz

real needle
#

do u have a vive?

full junco
#

yes

real needle
#

never had tracking issues?

full junco
#

yeah, its perfect

#

I had a broken lighthouse once, and with just one lighthouse I had tracking issues due to occlusion

real needle
#

your lucky or have a well prepared vr space

alpine torrent
#

PC specs for Windows Mixed Reality HMD need bluetooth 4.0 (motion controllers need this)

real needle
#

for most ppl theres small tracking issues because of reflections from the lighthouse beams/leds

#

myself included

full junco
#

@alpine torrent ah, thats actually a very good point! totally forgot about that

real needle
#

wouldnt that bluetooth be in the headset thats wired to the pc?

full junco
#

I also would hav expected that, but I also read somewhere else that microsoft expects the PCs to have BT

#

not sure if it applies for the samsung one too

real needle
#

wouldnt surprize me if it depends on the model too

#

yea

full junco
#

and I dont have any big reflective surfaces

real needle
#

i do ๐Ÿ˜›

#

but to lazy to cover em when i go vr

full junco
#

unless you have a mirror in that room where the vive is setup you shouldnt have issues

real needle
#

so yea, theres that to improve, before im fine with calling vive tracking 100%

#

theres a tv, shutters(?), windows, some monitors

#

the issues are small and very rare

#

but theyre there

full junco
#

and mirrors don't belong in VR-rooms, for the reason alone that you could break them in heavy gameplay ๐Ÿ˜„

real needle
#

๐Ÿ˜›

#

o rite

#

and i for some reason bought a computer case with a huge glass panel on the side

#

didnt even realize that untill now lol

#

might try moving it to see if it changes anything because atm one of the lighthouses is sorta aimed at the glass lol

full junco
#

well I guess if you have so many reflective things and don't want to cover them, a rift with 3 sensors is just a better solution for your specific room

real needle
#

hell no ๐Ÿ˜›

full junco
#

not every VR headset is perfect for every room

real needle
#

true but not gonna buy any new headsets soon bscly

#

i can deal with the screendoor effect of the vive for another year or so

#

bought it at full price so dont wanna discard it already ๐Ÿ˜›

full junco
#

in a year every headset will still have screendoor effect

real needle
#

it'll be less tho

#

and cheaper

full junco
#

yeah, but still existent

real needle
#

thats ok

#

its prolly gonne be there for another few years

#

i wasnt expecting it to go away soon

#

in the end theres only so much u can do about having a screen against your retina, your eyes are able to focus on that

#

theres gonna be a point where while its still there you really wont be able to tell anymore though

full junco
#

put stuff into your eyes that remove the ability to focus on anything near would work I guess

real needle
#

like, if your playin some fun vr experience now you hardly notice it, while its a pretty obvious thing on current gen

#

lol that could work xP

#

vr eyedrops

#

anyway ye, even that 8k headset on ks atm still has screendoor effect

full junco
#

yes

real needle
#

so im not expecting that to go any time soon

alpine torrent
#

VR Eyes

real needle
#

just hoping it reaches a point where its alot less noticeable then atm

full junco
#

its not 8k though, a real 8k headset might have significantly less screendoor effect

real needle
#

wich should be in like 2 years at the current rate

#

doesnt that 8k x actually have 8k screens? or did i misread that?

full junco
#

its their marketing

#

they use two 4k screens

real needle
#

ah ok

full junco
#

8k would be 4 4k screens

real needle
#

ye i just thought they had sourced (or maybe even developed) some small enough 8k screen somewhere

full junco
#

that would have been nice

real needle
#

especially because they where very vague about refresh rate n stuff in some 'Tested' video about it

#

wich wouldve made sense concidering the sheer amount of pixels

full junco
#

in the end I'm not sure how important resolution is compared to just "quality" of the screen regarding contrast and colors

#

so LCD vs OLED

#

sure you can use 4k LCD panels, but they can't be near OLED image quality

real needle
#

true, im curious about that aswell

full junco
#

like most reviews say, the samsung headset just looks significantly better than all the other windows headsets

#

not really because of resolution, but because of LCD vs OLED

real needle
#

didnt even think about that yet, the cheaper ones beeing lcd

#

thats sorta stupid actually

full junco
#

it makes the samsung one the only windows headset that's worth considering at the moment, at least in my opinion

hybrid plume
#

inside out tracking seems to be holding up in the Windows HMDs eh

real needle
#

from what i gather it does it pretty well, from what john_alcatraz gathers its not perfect

#

^^

#

and yea, obviously the no tracking behind your back is sort of an issue

#

but im all for inside out tracking so im ok with the current limitations, and hope later versions will have 360' cams to overcome that issue

#

or idk whatever else they can do to overcome it

#

bluetooth triangluation lol

full junco
#

๐Ÿ˜„

#

well I do think that inside out tracking has a future for the more casual headsets that will come

#

so the ones that you can buy for $199 or something like that in a few years

real needle
#

yea, setting up sensors is sooo mundane ๐Ÿ˜›

#

just wanna strap in and go

#

wireless ofc

full junco
#

but for enthusiast VR, lighthouse is the perfect solution that will stay

real needle
#

itll stay but not for long

#

i see inside out tracking getting good enough to replace it with 1, maybe 2 iterations

full junco
#

it can never be as perfect as lighthouse tracking

real needle
#

why not?

full junco
#

especially when you think about something like a gun controller thats way bigger than a regular controller and that will easily occlude the other controller behind it

real needle
#

the only reason lighthouse tracking is so perfect now is because doing it with just image data is kinda to much processing for todays hardware

#

that will change as chips keep getting faster still

#

also controllers are prolly only temporary

tired tree
#

John is talking about visual occlusion and how inside out headsets have to estimate positions out of the field of view....

full junco
#

well if one day we have 1000 hz 8k cameras, then yes, image based tracking will be as perfect as lighthouse. but below that spec, it can't really be

real needle
#

then again, they may stick around cuz ppl like holding something i guess

#

@tired tree i got that, but with some logic you can map that, theres plenty of work done on it over the past decade

#

it really took off when ppl could buy kinects

tired tree
#

@full junco did you see that from the spec sheet the Microsoft hmds are using B&W cameras?

#

you can't map that perfectly

#

you would need far more tracking points

real needle
#

theres plenty of software now that can do good tracking, slap better hardware on it and you could do great tracking

tired tree
#

you can estimate.....but its not going to be 1:1

full junco
#

@tired tree ah, no, didn't look at that, so didn't know it

#

no reason to have color I guess

tired tree
#

there goes any hope of actual AR out of the first gen

full junco
#

oh, yeah

real needle
#

yea it wont do ar

#

at all

#

they call it mixed reality

#

but this gen is just vr

full junco
#

waiting for someone to sue them because the reality in those headsets isnt mixed

real needle
#

lols

#

waiting for them to trademark mixed reality

full junco
#

I hope they can't

real needle
#

they prolly did already

full junco
#

don't think they can

#

the term mixed reality was pretty established already

tired tree
#

anyway banana, they can't fix out of FOV tracking with post estimation without more reference points

#

they can get a slight aproximation

real needle
#

yea fully aware of that issue

#

but for now its what we get to deal with i guess

tired tree
#

then how do you expect it to be as good as lighthouse?

real needle
#

like i said, they totally need to fix that later gens

tired tree
#

and how will they fix that?

full junco
#

@tired tree the approximation from pure IMU tracking should work for the short whiles a controller is not in the FOV

real needle
#

well apart from out of fov tracking their tracking is said to be pretty great

full junco
#

not sure why they don't do IMU tracking currently

tired tree
#

imu drift going to be nasty for that

real needle
#

theres several ways to fix that im sure

tired tree
#

there is no correction aside from pose estimation

full junco
#

well the drift would be corrected every time it enters the FOV

real needle
#

one could be simply adding more camera's

tired tree
#

more cameras don't solve it...

real needle
#

another could be some sort of basic tracking for when its out of view

tired tree
#

they would be better off with magnetometer like that one setup

full junco
#

actually the solution is pretty simple: the controllers need to have their own cameras

real needle
#

if the headset has cameras all around it can see the controller anywhere u hold it, so more camera's would solve it

#

yea that could actually work john

#

not a bad idea lol

tired tree
#

whats the name of those controllers that have been non delivierable for forever? I forget the name

#

that would be a better solution

real needle
#

oculus touch?

#

๐Ÿ˜›

full junco
#

haha

tired tree
#

sixense

real needle
#

orite

#

and neh i disagree

#

standard gaming controllers != vr controllers

tired tree
#

sixense could be used with inside out

#

for the controllers

real needle
#

unless im missing something those arent inside out

#

u mean the playstation controllers rite

tired tree
#

wut...

real needle
#

they have basic 'g sense' and then on the latest iteration theres tracking with a cam

tired tree
#

no

#

the Stem system

#

its a magnetic field tracking system

real needle
#

oh lol your talking about something totally different xP

#

k

full junco
#

I think the perfect tracking solution for the near future is a mix of all 3 current tracking solutions. You need the lighthouse for perfect headset and controller tracking, theres no alternative to that. then you need the inside out tracking from the headset for being able to view a virtual representation of geometry close to you in a good resolution, like a keyboard in front of you. and then you need external cameras, like the rift has, for image based whole body recognition so that you can see your legs and feet and in general your whole body correctly in VR

#

and that all works with current hardware

real needle
#

i dont see why all of that cant be done with 1 tracking solution

full junco
#

it can't

#

every one of the 3 things I mentioned can only be done with one of them

tired tree
#

technically with additional tracked points and two cameras on a vive you could, but vive trackers are bulky, would need a smaller setup

#

also occlusion

full junco
#

more trackers are expensive and annoying to setup and attach to your body

tired tree
#

yes

#

you know, lighthouses with cameras could do all of that

full junco
#

and for whole body stuff you only need low precision

tired tree
#

if they acted as webcams

full junco
#

lighthouse would just need to integrate a camera additionally to their lasers, yeah

tired tree
#

which would be amazing

full junco
#

but for close up stuff you still want the inside out camera from the headset

tired tree
#

full rear and frontal camera tracking for pose estimation

#

you could still get a dot matrix for local geometry

#

there would be gaps but it would be functional

full junco
#

I think about a 5x5 meter room

tired tree
#

add in a single camera from the HMD filling in missing parts

#

you know the HMDs direction, you can fill in missing geometry from it

#

hell if the camera was better you could use the HMD as is as a 3d scanner....

real needle
#

hololens does a 3d scan

#

and very likelly all this new windows stuff does the exact same

tired tree
#

yes we know

full junco
#

with 2 cameras inside the headset you can very nicely do a 3d scan of objects in front of you with very high res

#

so lasers and cameras in a lighthouse unit and the additional 2 cameras inside of the headset are the ideal solution

#

all cameras in this setup only need limited res and refresh rate

#

so its cheap

tired tree
#

light house cameras if they aren't building area geometry could be simple cheap black and white cameras

#

but you would either need passive light or to add those LEDs back

real needle
#

i feel like the best solution is tracking cams on the headset and then some additional tracking logic in the headset to follow the controllers when out of view of the tracking cams, so its 1 contained system

#

then using software you can do your body in vr and whatnot just fine

tired tree
#

that is not the best solution

real needle
#

why not

tired tree
#

because you are estimating the controllers when out of view AND the body pose

full junco
#

the cameras in the lighthouse unit should be infrared (different wavelength than the laser would be good) and have a bright infrared LED to light up the room

tired tree
#

its all terrible when compared to actual poses

real needle
#

ah your talking full body tracking

#

hmm

tired tree
#

yeah thats what I meant, they would need the infrared LEDs back

#

I think Gen2 removes those? might be wrong

real needle
#

how about just wearing wristbands and stuff

tired tree
#

how do wrist bands d oanything?

real needle
#

theyd be tracked the same way as the controllers out of view

tired tree
#

if they go out of view...

real needle
#

i mean

tired tree
#

and how is that?

real needle
#

and then for waist ankles etc

full junco
#

@real needle the best solution can never be tracking cams on the headset, like I said you would need 1000 hz 8k cameras for that

real needle
#

idk, i joked bluetooth triangulation earlier ๐Ÿ˜›

#

it can be some better thing tho ๐Ÿ˜‰

full junco
#

nothing beats lighthouse tech for tracking the 3 super important parts, the headset and the 3 controllers

tired tree
#

people have talked about bluetooth triangulation before, its not that accurate and can be interfeered with

real needle
#

thats why its a joke

#

but im sure something similar to it might be doable

tired tree
#

.....

real needle
#

im not here to invent that tho

full junco
#

@tired tree why did the old lighthouses have LEDs?

tired tree
#

the sync

#

data

#

now its on beam

full junco
#

ah, the sync

real needle
#

@full junco idk if the cams really have to be that good tbh, especially the 1000hz part seems a bit of an overkill

full junco
#

@real needle lighthouse runs on 1000 hz

#

they had a reason to not use 500 hz I guess

tired tree
#

yeah, i looked it up

#

LEDs are gone from gen2

full junco
#

the biggest issue I have with lighthouse is the noise

tired tree
#

which is gone in gen2

full junco
#

completely?

#

they still have a motor

tired tree
#

its a single axis rotor

real needle
#

noise as in audio?

tired tree
#

it doesn't get the vibration

full junco
#

@real needle yes, audio

tired tree
#

its very slight

full junco
#

well it is extremely annoying

tired tree
#

though, more noticable when they aren't mounted perfectly

real needle
#

im either losing my hearing or have well mounted sensors then ๐Ÿ˜›

#

never heard a thing from the lighthouses

full junco
#

I can't work while they are running without having music so that I don't have to hear the bright noise from the lighthouses

tired tree
#

its high freqeuncy, doesn't take much hearing loss to not hear it

real needle
#

prolly that then ๐Ÿ˜›

full junco
#

high frequency sounds are the most annoying ones though

#

if you hear them

real needle
#

oh wait how old are u john

#

cuz with age u hear less and less high frequency

#

they use that to remove teens from 'hang spots' here lol

full junco
#

I'm probably younger than you I guess if I hear better

real needle
#

im 32

#

that might be it, depending on how much difference there is

full junco
#

at 32 you should still be able to hear high frequencies pretty well I think

#

I do sit directly below where one lighthouse is mounted

real needle
#

yea i feel like i do, i hear other things, even still the teens disperse audio boxes

#

but not the lighthouses ๐Ÿ˜›

#

hmm

#

oh well

#

maybe try remounting them

#

idk

full junco
#

I did remount it, I even installed a different one because it broke ๐Ÿ˜„

real needle
#

hehe

#

ive yet to break anything ๐Ÿ˜ฎ *knocks wood

#

i did instantly replace the cords of the vive controllers tho, that prolly helped

full junco
#

I didnt break it, it broke itself

real needle
#

ah ๐Ÿ˜›

full junco
#

HTC quality

#

just like the controller also have issues all the time

real needle
#

thats actually fine here lol, i just have the occasional minor tracking issue but never had issues with touchpad or w/e

full junco
#

first one of my touchpads didn't click correctly any more, thats a super usual issue, easy to fix with taking the whole controller apart, now a few months later the vibration on the controller is broken

real needle
#

that sucks

full junco
#

and thats not fixable, so bought a new one

real needle
#

why didnt u try get it replaced tho

full junco
#

but I can't wait for non-HTC headsets

real needle
#

there should be warranty

full junco
#

I have a vive pre, no warranty

real needle
#

ah ok, thats a shame

#

that may be part of whats causing it tho, i have the retail one and yea, never had issues with the controllers or lighthouses making sounds

full junco
#

the lighthouses make sounds for all ones

#

and the controller issues are also happening for retail ones

#

both issues I had I found multiple threads on reddit about

#

HTC is known for bad quality

#

the external quality of the controllers and headset is very nice, just the internals like to break

real needle
#

yea idk

#

i heard that too

#

but never had issues with anything from htc lol