#virtual-reality
1 messages Β· Page 142 of 1
yeah
Do I still need a player controller or a motion controll? I've been able to get Vr to work in the project by changing the default pawn to the VR Pawn, and it tracks and has a hand, but none of the movement controls work
so it default isn't added to the VR pawn? the camera is
correct
you may wanna look at the first person template, or the virtual reality template
they both show you bscly all aspects of vr
I am but I'm not sure what comes over when I migrated
ah
so I'm clicking on the pawn in the example
and? π
Yeah the HMD is working fine
Epic spawns the motion controllers as seperate objects
and I have a hand (although the performance is terrible on the hand, but everywhere else in the level is great)
because they have all of those attached teleportation components with them and I guess they wanted to support no controller setups?
ye
what is the motion controller attached to
afaik too the pawn
talking about his
opening ue4 myself to check now
no epics, I am aware of how epic oes it
ah
I keep my controllers as part of my base character, if something doesn't have controllers it is easy enough to deactivate them or destroy them
That's just it, I can't see even in the example, WHERE the controllers are
it works fine when I run it
but if I wanted to change the mesh, I wouldn't know how
so, to make my project VR ready, (I have already migrated all assets) I need to add a Motion Controller Vr Pawn (settting it as default in the game mode) .... annnnnnd???
should work from there, you'll want to make sure it is actually spawning the controllers
I am just using a simple pawn
oh it's an input problem
event references are unknown because I never made the input keys? I thought that was native or whatever
Or maybe it's outdated?
could be outdated
orite, if u moved your project over into a new one you need to copy that too
defaultinput.ini
no I did not copy the config over, that must be it
I don't want to overwrite any binds, where is the ini files for both located and I will just copy and paste the bits I need
Maybe that's why the hand was so jittery, it was throwing errors every single time an Update occured
Hm... still not working
is it just jittery?
if so you might be falling below frame rate
since your scene is 2d normally
no, it's 3D and I'm running a 1080 TI with 7700k
Also, everything BUT the hand works super smooth, so I must not have hooked everything up
the inputs are there and the errors on the motion controller are gone
but i'm not getting all the same effects
nope, I should
because the hands are where you will notice the frame drop
ugh, I'm stuck in Vr mode, escape doesn't work neither does alt+V
that was weird
how do I turn on stat graph again ?
stat unit is prob what you want
what do the numbers mean? they say m/s, not FPS
so it says like 30 m/s for Frame
that's at engine scalability from low to epic, no change, so I know this isn't performance based
you need 11ms
in VR
to hit 90FPS
1000 ms is a second, 1000 / 90 = 11.11 ms
you WANT to hit 90fps in vr
yeah, I'm hitting like 20 ms
And it is on a BIG, unoptimized Map
22 ms? that would be 45 fps, would mean it is reprojecting if using steamVR
and would explain why the controllers are blurred / jerky
it seems smooth except for the hands
they are being tracked at half speed
its not smooth, it is doubling frames for you
so you don't get sick
Optimize and hit 11 ms
the controls aren't working though, I didn't migrate something or set it up
Here is exactly what I did...my project before had ZERO VR in it
So I migrated all folders from a New VR BP Project with all the content
then I added MotionControllerPawn to map, and changed the default pawn to that
then I imported the inputs from the VR template
that's it...
But it doesn't teleport around like the VR template
You must be missing some things then
That's what I'm asking for
Do I need to do anything to BP_MotionController or is it already hooked into the Motion Controller Pawn?
its already there or you wouldn't see the controller
And I don't need to change the PlayerControllerClass in the game mode correct?
I don't remember if they have a custom player controller or not, I haven't use their template for anything, but you shouldn't
I will build it from scratch just to learn but I wanted to try it out because it's been a 5 month wait to be able to try this in VR... also I wanted to see what else needs to be done, from the looks of it, this gigantic map is going to have to have some serious optimizing
Yes, a blue rod comes out
That's it, when I try to teleport
Do I need to do anything with Navmesh bounds?
that was it... I needed NavMesh bounds
Are there any guidelines for NavMeshes for VR? Or is simply define where the player can teleport (and it must intersect with floor
navmeshes in their setup are used just for checking valid teleport locations
You can use them for actual navigation of the player as well of course
Ah okay, thanks, I've almost got it working
Holy shit this is wild
Here's a question - Before I was teleporting my player around using the controller as a reference before transforming it via a new vector and a seperate new rotation.....do I just need to change the controller reference to motion controller?
the teleport function WORKS but I need to adjust the position
https://www.dropbox.com/s/n5n53setwq8nnjq/Screenshot 2017-10-05 12.05.33.png?dl=0 do I need to change playerpawn or controller?
Can someone point me to the best optimization tips for Unreal & Vive (I've read the best practices one) but now I need to optimize this map big time
Did they fix arktika so it doesn't require pre determined teleport points? Thought I saw some character movement in this video
that I don't know
looks like the moments character moves are actually on-the-rails
What's the best approach for capturing 360 vr video?
I want to create VR cinematics
There is a built in streopanoramic plugin for it.
You can pre-order your Windows Mixed Reality headsets at Microsoft Store and Microsoft.com.
well the launch is in 17th
Does ue4 support these yet?
someone in the today stream asked about it. i didnt hear if @magic mango asked the question
https://forums.oculus.com/community/discussion/58069/rift-1-19-release-notes/p1 << less and less new features from release to release.. I wonder if it means anything π€
@glossy agate well the platform have Steam VR support in preview
Ah ok. For some reason I thought it wasn't going to be supported until later. Guessing these will run on OVR input settings?
it's not going to be supported until December
@alpine torrent which question?
I think he is asking if the mixed reality headsets work in unreal right now
also my mind is a little blown that you can make a camera obscura inside unreal now.. Need to play with all the new lighting features
What's the max ms we can push for before dipping under 90 fps on the low-end VR pc?
I failed the Oculus performance test so I'm trying to optimize
On the new low end since asw released it is 22ms
Those are currently not supported, no.
@full junco @tired tree noise in gen 2 lighthouse is supposed to be reduced because it only has one rotor like you said, but also I think they said it will rotate at half the speed of the current rotors which should cut it down too
Multistreaming with https://restream.io/
static lighting in 4.18 is amazing
static lighting in ue4 was already amazing enough before...
Hey folks! Quick question on player boundaries (Vive). I'm teleporting the user to new locations, and I need to force-set not only the MotionControllerPawn, but the users location itself as well (aka the camera) to prevent teleporting inside of geometry. Does anyone know, inside the blueprint, exactly which actor I need to teleport?
Is it GetPlayerCameraManager?
or a child of GetPlayerPawn? <-- if I understand, this is like casting To MotionControllerPawn, correct? I've tried setting transforms of the Camera from this Cast but no dice =l
Well the nav mesh should stop that right? You can use blocking volume on the mesh.
Are you the real velveeta company? What kind of weird cheesy shit are you trying to do in VR?
lol naw just my name I swear :p
and we're not using NavMeshes in this, the floor is a custom per-poly collision mesh, also this teleporting is not referring to the controller teleport but a separate warp feature to warp to key locations
Ok so apparenty I can't force set the camera's location...
tried both SetWorldLocation and SetRelativeLocation on MotionControllerPawn's Camera component, and neither of them change the camera's position
I can if I disable LockToHMD on the camera component..
well i figured it out π inside MotionControllerPawn, I'm moving the SceneRootComponent such that the user's head ends up where I want it
and now i realize that that's the only way to do it, as moving the head itself would totally mess up the user's boundaries lol...
So, somebody was asking bout the Windows ME controllers. They feel very similar in hand to the Vive axehandle grip
Right at the back, there's the Hypereal Pano controller
All 3 feel about the same in hand, but I haven't had time to methodically compare them in engine
Good thread on tracking accuracy : https://www.reddit.com/r/Vive/comments/74k7su/quivr_windows_mr_and_archery/
@glossy agate & @alpine torrent me and another guy asked several times if they could give any details, no response, maybe 4.19?
maybe
@dusk vigil nice little review, ty for the share
@full junco you should read it too, considering our discussion yesterday, turns out its really not as bad a picture as u painted it to be π (which im sure you'd like to know too, as in, theres still hope for wmr lol)
Not really, appears to be just about as expected?
He is working around its downsides to get it functional, John was arguing for near perfect tracking not being possible. For many uses (especially business) the MR headsets are likely great as is and without workarounds.
you can make it 200% accurate tracking
yeah that sounds plausible..
no its not
@real needle ok
so yea, hands behind back is definatlly an issue, but good to read its not as big an issue as we concluded yesterday, because appearantlly the gyro tracking is surprizingly good
just thought youd like that little info too ^^
well I actually saw that post on reddit and stopped reading after the first half because, well, what he says is just obvious and nothing new
accelerometer data from the controllers and headset is simply fantastic. You can look at the controller shadow behind you and rotating your hand around is spot on. More impressively, you can cover the cameras on the headset, and while it does lock your position, the rotational data is perfectly accurate at 90fps and I canβt tell the difference.
that just shows that he has no clue how the stuff actually works
rotation is always perfect, you dont need any cameras for that
your phone can also perfectly tell you the rotation it has
and every quadcopter needs perfect >200 hz gyro too, so thats absolutely nothing special, but he makes it sound like microsoft would have achieved anything impressive here
Yes and no, you definatlly have a point, but i think your overestimating the quality of sensors in phones and quadcopters, they tend to go for cheap ones (ie; chips for smartphones) and then just do some proper guestimating with software
wich obviously can be the case for M$ too
idk, proper reviews will tell
it shouldnt get confused when walking around near magnetic fields, unlike phones
"overestimating the quality of sensors in phones and quadcopters"? in quadcopers good gyro data is way more important than in VR headsets
and phone gyros will also not be confused by magnetic fields
gyros are perfect for many years now
i pmuch disagree π
i hobby in robotics alot
a 5 bucks gyro will fly a quadcopter and balance a robot
but a 150 bucks one will give u so much more detailed info about translation
i guess im sorta confusing two things tbh
gyros and imu's
imu's are packages of gyro & accelorometer
same story tho, they come for cheap and for premium prices
and really, the cheap ones are just that, cheap
the guy on reddit said the "accelerometer data" would be fantastic, while he just talked about the gyro
hmm lol didnt notice that
he has no clue how the stuff works
what more detailed info about "translation" will a $150 IMU give you compared to a $30 one?
alot less noise for one
meaning alot less guestimating needed via software
and u can usually poll em faster, but ive never really needed that
translation isn't tracked at all by the IMU for the windows controllers, translation is 100% dependent on the camera
its mostly just that what you get is actually whats its sensing
yea tru
dang lol
well fine
i will try get you stoked about wmr again in the future, consider yourself warned π
even cheap IMUs are 1000 hz
and yeah, feel free to try to exite me about windows mr again π
well you absolutely don't need those for rotational tracking of a controller in VR
there even 200 hz is pretty much perfect, the speed you can turn your hand with is very low
problem is your hand is attached to a bunch more joints/your whole body
so add movement of other joints and u can turn your hand quicker π
well yeah, thats true
sienceβ’
windows MR is fine for what it targets, its just not ideal for every situation
weird....Qivr guy is saying there is no chaparone / proximity detection with walls
I would have assumed with that type of headset that a feature like that would be priority #1
maybe it will rely on Steam VR's chaperone ?
that's also the #1 thing some review I read criticised about it. or maybe it was the tested review? @tired tree
it's definitely something that's missing.
@mighty carbon it starts its coordinate system from where is it turned on
@mighty carbon well it will take a while before steamvr is even supported on those
thats not a steam thing
I wish Samsung would release their headset, just with diodes instead of cameras. so compatible with steamvr lighthouse tracking
I see
i'm all for inside out being worked on, its going to be great outside of gaming and for limited gaming
in its current state
it would be easy for Samsung to release the same headset just with different controllers bundled and with sensors around the case instead of cameras
to target the exact rift and vive market
they aren't
but they don't want to? they are targeting microsofts market, they want big business
I don't think "big business" is relevant for windows MR yet
by far the most people that buy it are gamers
the headsets are straight designed to bring VR into the boardroom
thats their gamble, that microsoft can tie into its office partners and sell VR as a corporate asset
well, I say if they count on that with the current gen, they will consider it a fail
its not like the VR gamer base is exactly huge compared to the numbers they normally work with anyway...
currently VR is almost just about gaming market though
when windows MR won't just be VR any more then stuff might be different, but currently its just VR
?
have you SEEN the architecture interest in VR? or the medical? or the engineering?
its massive
of course not, their systems are unweildy for that purpose
thus microsofts MR
they are trying to corner that market
there are a lot of people already working on applications for those.....MR is their perfect platform
it is better, due to no setup required, but the amount of applications that can only work without having to setup some area for it isn't that big I think
most of that architecture or medical or whatever stuff you can do on vive or rift just same like on windows MR
for many applications the client can come to you
even then
any office can be used without cameras and setup
I don't know why you are so down on it, its applications are obvious
and there is interest
when the client comes to you, then you have a room for that anyways, and then it also doesnt hurt you to have cameras or lighthouses installed there
well I don't think its terrible for gaming
"good gaming"
I think the main market for windows MR will be xbox, because thats what its absolutely perfect for
so, how do I get a free MR HMD from Samsung ? π
its way better than PSVR regarding the tracking, and PSVR is already very successfull
@alpine torrent any insight on how to get a free Samsung MR HMD? π
@mighty carbon If I know I would tell you
π¦
and with xbox scorpio MS has the perfect console for VR
how?
they could have dropped the price and lowered the quality if they wanted to target just consoles
but its multiuse
thats fine
it is multiuse, but xbox is the perfect market for it where most of the headsets will be sold for
in 5 years maybe the professional applications will be more important than games, but in the next few years, gaming will absolutely dominate the VR market
I bet it going there
have you spoken with anyone from say China about the VR interest there John?
no
I have been messaged by 12 dev teams out of there
all working on industrial applications for VR
well its probably easy to get investor money for that currently
not to mention outside of china, us interest and eu
and what did you tell the 12 dev teams?
I sent them contact information for some people and sent them on their way....a few I sent to Unity
as much as that hurt to do
didn't they want to pay you to do stuff?
yes
do you not want to work on any industrial applications because its boring, or are you just too busy with other stuff?
I have 2 (soon 3) kids, a full time job, and a life, VR is my hobby not my business
no real interest in working remotely through a language barrier
ah, well, makes sense then
I like the application concepts actually
one was an MRI scanner program where you could slice and interact with the 3D point cloud
kind of like that one thing in the lab?
yeah except actual point data, much more detailed, zoom and different view modes ect
I agree that such stuff is interesting, but it will be a long time until millions of headsets are sold for these usecases
@tired tree so you could use data in a database and AI where the data is so it could help
you mean stored positions and machine learning? that doesn't fix total tracking loss, it helps with drift to a point
eventually data starvation can't be accounted for
yeah machine learning where you store the data from MRI
oh you are talking about the MRI program
does anyone here have some experience with the stereoscopic 360 rendering in UE?
the kite and lighting plugin?
@tired tree well the concept like that AI where deeplearning and machine learning do the work where data is stored so then application itself can do stuff like slicing
thats fairly unnecessary...
arcore handles short term complete tracking loss, but no idea how many "memories" are stored before old ones are destroyed. Havent pushed any limits yet, but its been impressive for indoor use. Plain old cut wood is close to the ideal surface to track, but posters on the wall do the job too.
been using the arcore template project on googles 4.17.1 engine source.
short term as in 10 seconds of wildly moving/escaping room/phone in pocket..then return to room and BOP everything where it should be. let me know if i should try something else.
Robots patrolling streets in #scifi #indiedev #gamedev Planetrism #VR Made with @UnrealEngine and @Google #ARCore @Allegorithmic #SubstanceDesigner #AR #ARKit #UE4 #3D #animation https://t.co/N6lcz5NI6a
238
404
outdoor use of ARCore
wish i could do modelling/animation like that
heres a cool video showing just plain tracking over 3 floors(notice the 3d stair view when on 3rd floor).
heres the link too https://www.youtube.com/watch?v=ZmjMq5Hxg4Y
Google ARCore Tracking Test with Pixel XL - 3rd Floor Building Step Climbing 1. ARCore Overview ARCore is a platform for AR apps on Android. ARCore uses 3 ke...
with spatial sound recently possible with unreal/android, ar + headphones is pretty darn immersive.
Knew the improvments where coming in future updates! This is why I like UE4.π―
@glad plank since allways, epic only develops the features THEY need
can backfire for people that do things that epic isnt interested in
you only need to see people complaining about multiplayer performance since the UDK days, and now epic adding mayor upgrades to the server
that just means it is time to convince Palmer to hire Epic to make an open world fully dynamic lighted GI VR game
oh with playable 2D game on tablets so paper gets upgrades too
lol
pretty much
look at oculus bankrolling Epic for VR
thats the only reason the forward renderer exists
but seems they have now stopped continuing it as robo recall is done π¦
thats sad becouse that forward renderer could have better performance than the deferred renderer
and more customizable shaders
it already does though
just gets canceled out with MSAA
granted some of that is due to missing features..
but msaa gives great image
with a bit of tweaking, the unreal forward renderer could be like the Doom one
essentially also rendering to a gbuffer for postprocess, but a thinner one
didn't they add gbuffer features in a limited sense already do it?
i think not
they use it for specific things
oh
nvm
I was thinking reverse
they do forward rendering with specific features with the deferred now
yeah
when you get right down to it, fully deferred was top of the line when UE4 was started
but clustered forward kind of killed that title
the new forward renderer is a higher potential path
but they would really have to commit to it
another thing with that forward renderer is that it scales quite well
you can turn the gbuffer completely and make it a "pure" deferred "easily"
for higher speed
mobile gpus dont like deferred
the mobile renderer is forward
and in fact, they have ported the desktop forward renderer to IOS
oh did they? its not on the simple forward anymore? or is it optional?
didn't see that
its on 4.18 preview
clustered forward works SPECIALLY well on mobile gpus
they are tiled after all
so if you make your clustered forward use the same tile size, your memory access and branching is homogenous
and they mention metal. In vulkan you can call command buffers per tile
so you could actually just send the light data there
per tile
instead of saving it in a texture
so, forward renderer is now poop in UE4 ?
like, should I use deferred for VR nowadays ?
I've seen some research somewhere about having a small gbuffer per tile and doing all the passes there
So low memory overhead
Some stuff doesn't seem like it would work with that alone though, like ssao
@mighty carbon I would say it depends on your content
Some stuff is better suited to deferred still, though most of that stuff could work fine with forward + taa too
Not just content but final look
Forward doesn't support everything
generally forward is best for VR
its not "poop"
and its better than before...not worse
so when should I use deferred ? What would be the practical use case ?
Yeah nothing has regressed, they were just saying you can use desktop forward in addition to simple forward on mobile now
At least on ios
Ssao, light functions
I prefer deffered for lighting but we need better GPUs to brute force it.
Certain shadows aren't supported
my 980ti doesn't like deferred VR. We might be using it in our game for shadow reasons
light functions don't work with forward? bizarre :/
But our game wont be out ulta way after new gen cards make VR Deferred much more viable
Cloud shadows will only show up in the near cadcades and not the distance
But they work with deferred
The light limitations for forward are kind of unfortunate
I was thinking about using light function to fake clouds casting shadows on my landscape @sturdy coral
I mean, there is a reason almost all modern games us deferred right?
Try every feature you are going to want to use early
Don't make all your content optimized for forward and only then test if other stuff you want will work
But they are improving it pretty often
also could use post processing ink outlines (they don't work with forward), since they look x1000 better than the one I use now
they DO work with forward
I told you that they don't work with MSAA
and that the issue is only when you are doing partial occlusion
I contacted author of https://www.unrealengine.com/marketplace/post-process-shader-pack-vol and he said the stuff works only with deferred
faster
yeah, but AA is a must in VR
you can run FXAA with forward
(Oculus won't let a game in the store without AA)
but yeah, MSAA is obviously the best choice
so what's the downside of using deferred in VR ?
TAA blur, performance, and gbuffer memory usage
oh, performance.. that sucks :/
performance depends on what you are doing
Isn't forward like 20% savings or something? Or I guess dependent on the features used. Im just going with deferred.
performance difference depends a lot on your geometry too if you are using mssa
Yeah my foliage pack didn't work in forward so I went back. Made the palm trees all transparent even the trunks of trees.
ah weird
one of the robo recall mod maps from unreal tournament has some foliage and that works ok
ok, so let me put questions this way - if you are doing vast open landscape with some indoor stuff over it and under it, day/night cycle and everything has anime'ish ink outline, what renderer should I use ?
either and not be VR?
you are going to be in for one hell of a time optimizing that for vr
you can wait on vulkan and even then g/l
nah, that's what I was told about my Gear VR app and it passed tech review with flying colors
I am going to have limited draw distance for objects
it's not like you can see 10 km ahead and see every detail
but, I guess I can only say if it would work when I test it
maybe it will work fine but look like p00p
@mighty carbon if you want to do stylized, forward
very low poly would help
my main concern right now is ink outlines..
becouse you can do fully custom shaders in forward, you actually get more flexibility on the materials
wich could let you do a proper toon shader
yeah, modeling character right now and polycount will be a way lower than on Mannequin
@mighty carbon have you seen the guilty gear XRD talk?
they talk about outlines in there
saw that, hell no...
both external and internal
for internal they were painted on the texture
for external they did the oldest trick in the book. Render the mesh 2 times, the outline pass with the normals flipped
that's what I use now (from marketplace) and it works fine on Mannequin, but doesn't look nearly as good on lower poly models
in guilty gear (tho they use ridiculously high poly models) they modulated it with vertex color
to control how thick it is
I might just get in touch with the author and see if they can fix it.. If I make ink hull manually, it should look fine.. I did it for Steel Storm.
that way the outline is smaller or nearly nothing on some smaller details
which marketplace pack uses an inverted model?
what does flipping the normals do? was it to get a contrasting color and not just black?
most I know of were tracing outlines in the post process
@sturdy coral so only back faces render
the problem is that it increases polycount and while it's not a big deal for a fighting game, it is a big deal for game with several characters in the view, level geometry and stuff
there is a outline plugin that uses the duplicate model trick
i guess motorsep is talking of that one
and then they expand it out with a geometry shader (amount modulated from vertex color)?
All Platform Outline Pack @mighty carbon ?
i bought it on sale like for 2 dollars just for the hell of it. I already had my own outlines, but i wanted to see how they implemented them
ah, k, yeah that would explain its name too
as expected, such an "old" trick works literally anywhere
some of the fancy stuff in the marketplace pack is that they had "screenspace" outlines
wich just modulated the thickness with the world distance
@wicked oak yep, that one
so they stay at a "constant" thickness
that is a problem with him wanting a fully outlined world though
fully outlined world = postprocess
yeah
or dont outline the world. You can keep the characters outlined and the world "painterly"
grass and foliage not outlined but objects and buildings are?
but characters, landscape, rocks, etc. should be
if you want to do like borderlands, (outline EVERYTHING) then its a postpro
....
going to really have to stlylize that then...
and you won't be able to use instances
What is "outlines" ? Like anime? Cell shading?
Or literally just object highlight outline?
that likely would end up MORE performance taxing than the post process, easily
Borderlands didn't have inner parts inked.. My old Phaeton game did. So preferably I'd like to get that look, but if not, oh well.
Man, I want a super optimized cell shader for VR.
If anyone knows of one, hit me up.
so I guess i have to use post process ink outlines and deferred renderer ?
or forward and FXAA
(FXAA looks like crap in forward)
there are some work around methods for the MSAA custom depth precision issues, but none that I was happy with
there is bound to be a good enough solution
for vr it would prob be better to have the flipped normal versions of the meshes baked into the meshes themselves
I'm a bit annoyed that most cell shader are only PP solutions
polycount will be unbearable most likely (and drawcalls will be through the roof?) with inverted mesh ink outlines
@mighty carbon probably yes if you want them on everything, not that bad if it is only characters and some other objects
motorsep you can enable specific objects to be outlined in a PP with custom depth btw
Or you can go in there and hand paint the style.
A lot of the boarderlands effect is just in the texture map. Looks cool, but it will take some time
it won't look good as texture will be aliased and I don't think I want to do what Guilty Gear did
plus you still need procedural outline for the silhouette
@tired tree aye, I'll have to look into it
if I could use pp outlines with forward renderer, it would be sweet
@mighty carbon are you sure you cant?
becouse the forward renderer still does a depth pass
its because of MSAA
the scene and custom depth values don't match up, there are some imprecision issues, likely from MSAA downscaling
so when comparing them you get artifacts
if he was doing FULL scene outlines it shouldn't matter
doesn't MSAA work based on actual geometry and thus all pp stuff is not being accounted for ?
what happens to pp stuff when using forward + MSAA? How does it get AA'ed?
π
I'm having issues just getting the thing to work!
does it work like a normal pluging ie just copying the plugin into the plugings folder of the project?
You have to build it unless you downloaded the prebuilt binaries I think. I haven't tried pre built.
MordenTral I believe made it
Yeah he did.
I downloaded the prebuilt binary for 4.17
Is it all C++ or does it have BP?
DId you manage to the get the uproject for the template to open?
It has installation instructions in the wiki
I know
how should I build it in VS?
Double click the little vs thing above uproject then build when VS opens. Make sure you rename the folder to something shorter.
this is my project directory
nothing is above the uproject
and the project files don't even generate when I right click "generate visual studio files"
You need to follow the steps to add a new c++ class to your project
right now it doesn't have any source and won't compile
orrrr..
did you not install visual studio
good to see you here @tired tree
Appreciate your work and would love to get to grips with it
i'll start PMs, don't need to clutter this channel
ok people, back to fighting
fight the good fight brother
Guys, got a Rift earlier this week
I'm running into some massive camera jerking when I'm using the VR editor with motion controller
Lots of times when I move the controllers, the camera spazzes around
It's not the tracking, I assume, since it only does this in UE4
Any ideas?
(Tag me if you respond please <3)
Actually, after a little bit of testing
The issue seems to only/mostly appear when hovering over the foliage-mode panel with the controllers
Landscape etc are fine
Only foliage causes jerking
Using the forward renderer makes all my actor viewports black, any way to revert this back to lighter?
VR compositor refused to start for some reason.
@boreal walrus Rift tracking degrades when all cpus are used heavily
It happened only after I attached my 5th screen. π¦
5th screen PLUS an HMD, or is HMD sceen 6?
The HMD is the 5th screen.
I need more screens. Only working on 1 screen at my home office like a casual
I have 3 screens at work.. Working at home on 1 screen feels like a massive downgrade π¦
One screen is fine.. as long as it is a big 4k tv
Website: http://media.ee.ntu.edu.tw/research/DodecaPen/ Publication: DodecaPen: Accurate 6DoF Tracking of a Passive Stylus, ACM UIST 2017. Authors: Po-Chen W...
cool
That's cool for being cheap
These are more interesting as an alternative for sculpting, because they have haptic feedback: http://www.geomagic.com/files/cache/2d56ff2514bc2f23491707eb6c984d28.JPG
Much less range of movement though
yes
Trying to create a bolt sliding mechanic but I'm not sure how to restrict its movement once it's grabbed to only back and forth so that the player can't just grip it and move it around in space
It has to stay inside the gun and move down a predetermined path but the player has control with their motion controller / hand
How do I restrict its possibility space?
@uneven moon I think several of the vr templates/plugins have examples
(at a minimum the slider ui in mitch's vr template, levers in mordentral's)
Awesome thanks
@real badge what have you tried. It depends how you are handling movement as well
Turns out the player doesnt actually move normally
It just teleports and lerps every frame
for locomotion
Nothing @uneven moon
The character doesnt use normal movement
It teleports to the location
Oh so your locomotion isn't working to begin with
No?
It works
It just doesnt use a movement component
and just teleports to the location
So you point+click and it moves there or
Can you set it up to use the movement component but still have that functionality?
Im going to have to add a movement component because this method is pretty bad
Yeah, you should be using one that way you'll have access to all of that information
Yeah, I didnt make this VR template
Getting the velocity now depends on exactly how you set it up
How is the movement handled?
oh jeez
Yeah, not a good way to do locomotion even if it's seamless
You have no velocity
What you could do maybe is calculate the time taken from point A to B
And use that
Im just going to use the movement component if I can
probably going to break everything and hate myself
Make a duplicate project and try it there
If it works out, keep that
Shouldn't be hard
Always a good idea to start VR projects in a blank project, setting up the basic VR pawn is very straight-forward
Then, if needed, use the templates as references for certain mechanics if you don't know how to produce them yourself
hmm, I can't really understand what Mitch did in his lever / slider examples to replicate them into a bolt-action rifle (pulling back the bolt on the X axis, clamping the possibility distance)
Anyone looked at it before or has done something similar?
just saw this bug report, anyone seeing something similar in 4.18? https://answers.unrealengine.com/questions/712897/hmddevice-isnt-recognizing-the-vive-1.html
I haven't yet tested any VR stuff with 4.18
hmm nope but I haven't used vr in 4.18 either
@sturdy coral I'll check if it's my CPU... But why would that suddenly go to shit when hovering over foliage types in the menu?
It also does the lagging in the content browser#
Basically, I guess anywhere with the content-browser preview images
Whenever I hover over those it freaks out
And it's not CPU usage, I can happily paint tons of foliage
But when I hover over things in the menu, my tracking freaks out
well slate is sometimes doing weird stuff, you just have to accept it I guess
@uneven moon Mitches examples, while well made for what they are, don't cover all of the bases.
you won't get a decent (attached) slider copying the logic
and it doesn't work entirely in relative space
so putting it onto a gun will bug out
its a great start though, you'll just have to convert the levers and buttons and slider to be purely in relative space
@real badge unless you are using the floating movement component you are going to have to do a lot of work
the character movement component only works with characters, so you would need to custom make one or use one of the already available plugins with one
Ive got the movement component from the character class to work with VR with nearly no problems in the past
with a character?
because you will definitely have issues if you aren't having it follow or offsetting the collision
unless all you care about is the movement itself, at which point..floating movement should be fine
even then though, all movement components do it store a saved velocity, its not like you can't manually calculate that yourself
Does anyone else sometimes have the vive teleporter teleport them to the origin, instead of to the selected location?
I did a check on all places that return or take the position vector, but nowhere in that chain does a (0,0,0) result occur
and yet, every so often I'm put there
if the navigation mesh check fails it returns an empty vector
so you should filter the teleport logic to fail if the return location vector is 0,0,0
the node isn't implemented correctly, it only returns false if the navigation system doesn't initialize, it still returns true if it just couldn't find a valid location
did you see this discussion with palmer luckey? http://www.pcgameshardware.de/screenshots/original/2017/09/Palmer-Luckey-sprach-in-einem-Bikini-ueber-VR-Pornos-und-Roboter-Arme-pcgh_b2article_artwork.jpg
π
its funny how crazy he became after he left oculus
haha. You have a link to the talk? Couldn't find it on YT
I'll check it out, Morden, thanks. I'm just using your (and in turn the Epic) teleportation BP though, I reckoned it should behave normally
found it.
hey did they remove arcore for unreal?
https://github.com/google-ar-unreal/UnrealEngine/tree/4.17-arcore-sdk-preview
its 404 not found
Are you linked and signed in?
@frigid kite I had to fix that very thing a couple of weeks ago in blueprints
you may be on an older version than that
I didn't know thier node was bugged
I had fixed it on the car exit but forgot to check their teleport setup, don't use teleportation much at all
@glossy agate nevermind
performancewise, is it ok to have a splinemesh in your game or is it much better to position your splinemesh, and merge actor it?
i mean, for a splinemesh there will possibly be some overhead of having a tick (which i suppose i can turn off)
but which option is best?
aha, alright
@tired tree I'm still fairly new to UE4 so I can only understand about half of what he did. Someone else suggested doing this via a timeline somehow, as in having the motion controller, once gripping the bolt, adjust the float when travelling from A to B on the X axis but I don't imagine how that would work. Someone else said to clamp the relative location which sounds like what I need
@uneven moon What are you doing for your regular bolts on other guns?
I have my bolts working, but havn't figured out the 2 part bolt action yet ie. have to lift before you can slide back. I just moved onto other stuff for now but if you figure it out let me know
@uneven moon clamping is really really easy, record initial hand position and get (currentHandPos - InitialHandPos) each frame and set the slide to that delta value from its original location, but Clamp(Minlocation, Maxlocation) the result
you can dynamically change min/max location to handle things like changing allowable directons
rotation is another matter though
I'm trying to optimize for the first time, using the GPU profiler after watching the Epic's livestream on it...I found an entry in lighting (lighting was 11 m/s total!) I have StandardDeferredSimpleLights is 9.7, how do I go to fix that other than just deleting random lights? Even if I put the headset down it says that (so it doesn't matter where I look)
https://1drv.ms/i/s!An9AJkcK_v3PiyY4dR9Z88cZBzCa hmm...shadow depth
You can lower shadow resolution
yeah there were about 20 m/s in lightning and mipmaps so there is probably a lot I can do pretty easily
this was all when I was brand new to Unreal and just threw things in there that looked good and then waited until I had a really good machine
How do I get a component's location relative to this red line I've drawn? (There is a spline there)
Motion controllers need to be able to move from point A to point B but getting relative locations doesn't account for room-scale
It does. It's just relative space of your gun BP
@glossy agate this is inside my VR pawn BP, the gun is attached to my hand, it's not a separate component that I pick up
gonna be cleaner to put the logic in a separate gun BP. Will save you some headache.
but even then, I would need to calculate the motion controllers vector locations
so it doesn't change much from how far in I am now
if I strafe left and right, the motion controller's "locations" change
both in world and relative
even if the controlls are still in my hand, not moving
seems impossible
It does. When you grip through a BPI you just set off a tick event to match the world location of the controller along just the one axis and use it to set your relative location of the slider in gun BP.
@uneven moon you need linear math
i actually have that kind of thing
a line is done from 2 sockets, or from a socket + an offset
once you have a line, you can use math to find the closest point to that line (in fact ue4 has that function), then you snap the position of the hand there
if there is a lot of offset, then you detach the hand from that
think he is more confused about relative space than anything else
hi, is there an existing way of getting a loading screen with a dynamic element? I was thinking of streaming in a loading screen level, and showing this while loading the next level, but if something already exist, then I'd rather not.
@uneven moon ifyou are holding the item and strafing left and right, both the held object and your other hand get the same offset, you still have the same relative space coordinates
You need to inverse transform the motion controllers location by the guns transform
either both in world space, or both in relative space to the actor
that provides a location vector of the second hand relative to the gun
@wicked oak splines also offer closest transform to point as well, so you can literally draw out paths to follow
including built in rotation
@wicked oak @tired tree I tried this using a spline and using point 0 and 1 as my linear limits and it does work but it's not 1-to-1 with the motion controller, even after I did an inverse transform and I tried both world and relative space for everything. The bolt moves back and forth but a lot faster than my hand even though from point A to B the timeline moves from 0 to 1 and the bolt is moved from A to B depending on that value
Couldn't figure it out.
@full junco nice review there
Still a gamble
It will be good for VR as a whole
Next gen Rift/Vive might take notice
@uneven moon of course it won't.....the distance would be different between the two... You want to sync the hand motion up to it not a timeline curve....
From the creators of the Metro series comes a new vision of the apocalypse. ARKTIKA.1 is a highly immersive, action-packed, first-person shooter designed exc...
no free movement (using RR-like teleportation nor sliding locomotion)
its the same exact locomotion i had in a prototype
i decided against it becouse it needs a lot of developer time to put all the nodes properly
but it offers very good "flow", as the player controls it directly
kind of like time crisis
its suuuuuuuuuuuper high end, graphics and FX
and sound
the gameplay we will have to see
but looks quite cool
yep, looks cool.. Production value is really high
Wilsons heart had the same sort of setup. I think that one was pretty popular.
No steamspy on it though to check.
Hello, am running into some trouble. Am using the VR template in Unreal and added a socket to the bone of the VR hand which comes in the template.
Copied over the cube pickup BP and replaced it with a pistol mesh. Tried attaching the pistol to hand on pickup and snap to target but issue is the pistol always ends up in a dfferent location as compared to where the socket is at.
@pale whale is it different depending on how you pickup the pistol
@noble charm no, its always in the same position regardless of how i pick it up
Most likely due to the centre of the pistol mesh being off
Have you looked at that in the blueprint
I've tried attaching it to the mannequinHand_Right skeleton mesh in the level blueprint and it snaps correctly to the socket in the level.
cant get it to attach properly on my hands in VR
Is there a difference between the two methods you are using to attach
Level bp vs motion controller
in the level BP i'm using attach to actor while in the pickup pistol BP i'm using attach to component.
my pistol pickup BP, bascially the same as the cube just that i add the socket name and have it snap to target.
as a ref my level BP
What's passed to attach to
a scene component based on the PickupActor interface.
Is it the root component
Because thats the main difference
What if you use attach to component in the level bp
In the blueprint editor?
hmm, the hands are spawned and attached to actor
@pale whale Did you solve your problem?
I ran into an issue like this once because the pivot point for my mesh was not centered. Import your gun mesh into another software and make sure it is at 0, 0, 0 then export it as an FBX and re-import into UE4, then try to attach it.
@uneven moon I managed to solve it by passing over the skeleton mesh component as part of the interface component being called. sorry for the late reply.
Ah okay, cool. Np
anyone noticed that if you have two or more materials on a static mesh, then the physical material returned from ie. a trace does always return the phys mat from index 0?
The console command r.ScreenPercentage is crashing in VR for some of my users, has anyone else experienced this?
not that i've heard of, is it possible to get their log from them?
often there's some info about what happened prior to the crash in the log
how high are they setting it? might be using too much VRAM
Absolutely nothing in the logs except it ends after the r.screenpercentage command @eternal inlet, and setting it to anything crashes, even lower values than the default @tired tree
what branch are you on
Vanilla 4.17.2
Turns out the crashes are resolved if they use a .pak unpacker, which is for modding / ripping game assets
Very strange bug
Packaging a build without a .pak files resolves the issue. Has anyone else com across anything like this? Where the .pak files cause issue?
Okay correction, using compressed .pak file seems to be the issue, not the .pak files themselves
Anyone experience black BP viewports and thumbnails when using forward rendering?
@tired tree Remember when we talked about teleport sometimes still going to origin, and you having fixed it by checking the projected nav mesh result because the implementation by Epic is incorrect?
yes
Turns out, you can still teleport to the origin despite that check, because of the delay between the initial action and the actual teleport
during those frames, the position is still being updated if you move the controller slightly
but the bool for valid location is no longer checked once the delay is started
aye
Someone should just release a teleport MP asset
pref in plugin form
with editable stats
$$$$
@frigid kite I fixed it in my template, obviously if someone is basing off the Epic one it would still have the issue.
@pine nimbus you don't want to use compressed pak files anyway.
I think it interferes with steam's delta compression
unless you are trying to squeeze on mobile below a certain install size I would avoid it, steam will compress the download on the fly anyway
and delta compress updates
Interesting, thanks for the info. I found if I archive the uncompressed packaged build using winrar, it ends up smaller than when I archive the compressed packaged build, so I guess it's related to what you said about Steam's compression. If it can't compress it as effectively because it's already been compressed by UE4. Space taken up on harddrive is a non issue for my game so it's good to know π
@sturdy coral
yeah steam will compress the download stream, and if you make a small update to something within the .pak users will only have to download a small delta
but if the .pak is compressed I'm pretty sure they have to redownload the whole .pak
yeah, I first had compressed pak files enabled but steam really doesn't like it
I just bust my nuts buying a single 1080ti. I don't think I can afford an 8k headset nevermind another 3 1080ti's to run it :p
does it matter? either it would come and fix more bugs, or 4.18 would release?
more previews isn't always better
granted, 4.18 looks like a stability patch, which is good
4.17 was preeeeety unstable
dunno.. Just reporting that it's coming π
do i need to build from source to make ARcore work?
4.18 just has it, I believe
so, not if you use that
I played with the 4.17 custom engine build for ARcore (well, I built the Google sample), and I must say I was rather disappointed
tracking is quite unstable, the surface detection is far from robust
Vuforia's extended tracking works at least as good, if not better, and that's been out for several years
Google has been working on Tango for years
ARcore is basically tango without the depth capability
In the 4.18 features preview vid ARkit seemed to be more stable, but I don't have any iOS devices to test with
(or just any at all)
yeah
it would be more stable
it came out first
it's google it will work π @frigid kite
I'm hoping it will, aye
Is there no way to 'zoom in' with a 2d scene capture? Trying to create a scope but it shows things the same size as they are when seen with a naked eye (it's a sniper scope)
I think you just the FOV
Naw that doesn't affect zoom
I read that this can be done in VR by modifying the render target
Trying to 'scale up' the texture
I have a working scope. I'll check in a bit.
But yeah it's 2d render target. Gotta figure out how to do the parallax effect still though so it feels like a real scope instead of a little screen.
FOV was just for the capture to fake a zoom and you probably have to scale it up in the mat
FOV totally affects zoom
I thought a scope should optimally be an independent custom camera render
Therefore with all the bells and whistles ie custom fov, dof, whatever
Tip for scope : don't let the eye see it unless it is right over it, like a real scope, it should just be black unless lined up
It's not about parallax or whatever. It's about not being able to see through it unless lined up very precisely
You can sort of get that effect automatically with parallax though
If you look at a scope from say 10 cm away at 45 degree angle, you see nothing period. black
If you show anything, you break realism
Yeah it won't be as extreme, but if you look off angle with parallax mapping, you will also see black
unless it's a scifi scope monitor
If you want to go for realism, just fade to black if eye is not lined up. Easy and works
Fade isn't enough, it will look flat even if only to one eye
Yes. sniper scopes look flat, because you look at them through one eye only
You need it to move when the angle of the gun to the eye changes, which parallax gets you
I dont understand why you should need to add extra cheats, if you are modelling the scope correctly
It isn't quite the exact effect of a real scope, but fade to black definitely isnt
- question of what the scope looks like lined up in use
- question of what the scope looks like when you are not looking through it
which one are we talking about?
The transition between the two
Easy and cheap is the fade to black, I dont see why you would want to make it more complicated than that. A fellow team made this system, and it was very convincing without any more extra work
They also had a fully physical rifle with tracker puck
Very nice. The weight of the gun really made it feel good
The scope was kept black until pretty much perfectly lined up, much as real scopes are
Yeah, real ones transition pretty suddenly and have a pretty narrow sweet spot
Yes indeed.
On low power ones you can definitely see the keyhole kind of move into place with a sort of parallax though
Why bother spending too much extra effort on such niceties, when one should focus on actually surviving in this world
You are not going to make extra sales by spending a day or two on that
It's basically just a parallax material node and a sphere mask node
Prioritize your battles. Pick low hanging fruit. Make and publish games
I made the mistake of polishing a turd too many times in my life ; p
Already had to learn the parallax stuff to get a good dot sight anyway
Fair enough : ) Send me a key when you are on steam beta
simrak...its pretty common to use parallax in VR for scopes
Will be keep to try it out. Yeah, I was not arguing that, more the amount of time spending on polish of view outside of scope -> to targeting
?
its a core component of a game usually if they have scopes
why the hell not polish it
I feel like for example mirrors in project cars
Would look 20x better with parallax to fake the 3d
Instead of looking like screens
I have not investigated parallax effects in this area. Will look into it. Still feels like overengineering. Mirrors are mirrors, scopes are scopes. Is the added parallax effect doing something to compensate for what the camera cannot do?
It is under engineering, instead of trying to get a perfect stereo mirror with realistic distortion or whatever
It would be faking some stereo with parallax
To look better than what they have: flat and looking like a screen instead of a mirror
I can understand stereo necessity in mirrors, but since scope is fundamentally a one-eye thing, why on earth go for stereo? I dont understand that?
It is a one eye thing, but you are moving it with your hand independent of your eye and if you don't see some effect from that it can feel flat
No, admittedly I am not a shooter fan as such
I though the camera of the scope would be configured for whatever is needed. Time to download Horseshoes and really look at it close
Still a question to make things clear : you only look through the scope with one eye, yes or no?
even in VR
straight texture render lacks depth
and high powered real world scopes have parallax in them
to the point where you have to learn to account for it
Ok, will buy Hotdogs and check it out tomorrow to see what the difference is
Demonstrating 'parallax' effect in a Simmons 3-9x40 at 50 yards. The opening in the copper plate is ~1.5", note how far the crosshairs wander just by moving ...
it gets you that type of effect
distance and angle based parallax to line up
The main difference I think you will see is if you move your eye closer to one that is just a scene capture painted on the surface with blackout, you see the same fov as if you move your eye farther back
With a real scope as you move your eye back farther, the image doesn't really grow or shrink much, and just gets cropped in more by the sides of the scope
I get what you are going for, but still think it is wasted effort unless you are making a pure sniper game
its often mentioned in every vr game with scopes at all
Yeah that video shows a lot of it well
and isn't very hard...
It is definitely the first thing I'll notice in a gun game if it just looks like a screen
No worries : I am just trying to be pragmatic, focus on giving players lots of content instead of getting tied up in details. Admittedly the devil is often in the details : )
If you don't have polish you better have a lot of innovation imo
there isn't a pre-built package for parallax scopes in ue4 yet is there? I know unity has several
I'm not sure
These friends of mine with the tracker puck rifle tried to sell their rifle range prototype to a gun manufacturer. For them it wasnt really a dealbreaker of realism, it was more that they were just so used to firing real guns they could not see the point of a simulation in the first place, at least the old school dudes. Apparently some of the younger ones appreciated it
there are a lot of tracker puck rifles out there
pretty sure the AIM controller is going to be more popular than any of them ever will be too, due to first to real market and not requiring additional unavailable hardware
Yeah, at first I was dazzled by the idea of a puck that could serve many purposes, but in the end it seems like it may be a pipe dream. Though it might still work out, who knows
It just needs to be like $20-30
the tracker modules themselves have a real use, built into hardware itself
but the vive trackers are likely DOA, considering any company can make the same but cheaper and smaller with the LH 2.0 tracker kits
Yeah I hope 2.0 really brings down cost
Someone said the photodiodes plus the little driver board on each are over $1 even in bulk
Though I don't know how they knew the bulk price, just saw it on reddit
I think there is a public price on batches of 1000 or so
yeah, but mass production gets deals on that
Just so I get that parallax thing straight, sorry if I feel obtuse : we are talking about that black space, the scope body, not the view through the scope? ( Sorry to get back to that, don't really want to spend too much time on it but could not help investigating a bit more )
I think I understood it right but wanted to double check
Aaanyway. What would you do with a thousand pucks?
no....its the line up of the two lenses on the front and the back of the scope
Ahhhh
the lens will show some of the inside of the scope when misaligned
Ok, so that's where that little ultrarealism kicks in
Guys, I'm trying to get D-Pad Inputs of GearVR controller. and according to documentation its "Facebutton 2/3/4/6" but I can't get those inputs ... touchpad is working fine though
I could never get controller mapping to work, so I only used touchpad press and trigger in my project
@dusky moon ^^
4.18pre4 is out
damn, that's sad news
310 employees
for a 60k dollar camera
that is in an incredibly niche market
it had absolutely NO chance of ever succeeding
like, even if VR went to smartphone level right now, it STILL would be a bad idea
assuming a 20k profit per camera, wich is really high end, and the employees getting 2k dollars per month (wich is very low end) then they would need to sell 31 of those cameras a month
a more logical cost would be 10k profit per camera, and 4k dollars per month on employee, wich means more than a hundred of those cameras a month
how the fuck do you expect to sell so many of such a niche expensive as fuck product
they should have made consumer grade camera
even like that, spamming 300 employees to put a few cameras into a ball is absolutely retarded
you dont need that many
right
@mighty carbon oh man I finally found it! Unreal's Documentation is quite outdated for that.
to get D-Pad Inputs you have to use "Motioncontroller (R) Thumbstick Up/Down/Right/Left"
yep, so I did.. My issue was that when I'd press d-pad, I get 3 events happening simultaneously - touchpad touch, touchpad press and random D-pad press (I used keyboard input node that returns any key/button press). The main problem was that when I used Motioncontroller Thumbstick, I could never guess what is being pressed. So if I pressed exactly on directions, it would work as expected. However, no one will do that and user would press Down for example, but a bit closer to Left or Right (physical location of the finger). And UE4 would return 3 values - Down/Right/Up..
that was in 4.16.2 though
Has anyone tried Terrain Foliage for VR? Meaning using static lights for terrain foliage?
@mighty carbon That's True ... as far as I tested it's still a thing in 4.17... but generally I feel its better to avoid D-pad and stick to swipe gestures as you said. d-pad can be a bit confusing
I'm looking for some more in-depth guides for GearVR dev ... just generally, like The Cheat sheet in documentation. If any1 knows some other sources... would be helpful
has anyone used SendKeyChar with the WidgetInteractionComponent
I really wonder how Vanishing of Ethan Carter did all that foliage :/
cause this just look ugly and I can't get it fixed for some reason
@dusky moon I used trigger and touchpad press for my project. Tried making UX as simple as possible. If I wanted something more complex, I'd require gamepad for my project.
@tulip surge Maybe try turning off the self shadowing. Kind of looks like you have it on. At least for me it seems to make foliage with baked lighting look like shit.
yeah baked light is shit I know now, I'd have to make the forest a dark place if I use baked but I dont want that , especialy not for VR
Well turn off your self shadowing on the foliage and try cranking up your skylight some to see if that helps.
there's probably a way to get it look good with some settings, but I don't know anything about baked lighting
That should help avoid getting the black leaves
several reviews praise ARKTIKA.1
6 hrs campaign
also, a little too late, but https://developer.oculus.com/blog/open-source-release-of-rift-dk2/
Hello, somebody knows how I could enable collisions on motion controller? I would like the hand meshes to sweep when encounter a collision
would that be Cast Shadow ? The self shadowing cause self shadowing itself doesnt seem to be possible to disable
@timid anchor motion controllers can't collide with stuff, they are always exactly where your hand is
@full junco yes, but the mesh attach to it can theorically collide, I tried to just set the location and rotation of the hand to the motion controller's ones with sweep enable, but It just takes the root bone of the skeletal mesh to do the sweep
if you make the mesh collide, it won't feel good because it will be at least 2 frames behind the motion controller component
you can do it with physics handles, but you shouldn't do it for the motion controller mesh
he can manually sweep it from last pos to current after the motion controller ticks
it would only be about half a frame behind visually
which is still a lot, but all of the game logic is half a frame behind or more visually
hm? a sweep happens on the game thread, the GT is 1 frame behind the render thread, and the render thread is usually 1 frame behind what you get with the late update, so it would be 2 frames behind
or 1.5
render thread is not 1 frame behind late update...
late update is applied late in the frame on the render thread.....
based on game threads reported pos
late update is at the end of the render thread, and I think you could say that end of render thread is 1 frame later than start of render thread. or half a frame