#virtual-reality

1 messages · Page 156 of 1

full junco
#

@mighty carbon very interesting to see the Microsoft headsets already be at 199

#

seems that prediction from the German journalist was absolutely correct

glossy agate
#

^Just looked on the microsoft store and some are even getting sold out with the discount. Should bring in a bigger player base for next year hopefully.

pale burrow
#

hi again john 😛

full junco
#

@pale burrow if you count all the individual things together, do you get to anything like 20?

#

base pass 4 ms, PP 4 ms, and some smaller stuff

#

well, you definitely can disable the light shaft bloom, you have to

#

and screen space reflections

#

and some other PP stuff

pale burrow
#

i cleaned up the ssr from the last time we profiled

#

looks like this

#

so ssr is no big worry anymore

#

somethign is very wron with my sscene... im not using any specific pp settings.... why is usage so high?

full junco
#

don't always open that base pass up, it doesnt say anything interesting

#

just eats space and makes it hard to read

#

youre just on a way too high res

#

2000x2384 per eye is too much for that scene on your GPU

pale burrow
#

wait what

#

im not supersampling...?

full junco
#

you are on some quite high screen percentage

#

don't know the console commands for that in newer versions

#

@glossy agate I wish they would ship to germany 😄

#

but they don't, and in the german microsoft store its still $530

pale burrow
#

that was it @full junco thanks for the fix!!!

full junco
#

ha, great 😄

pale burrow
#

not sure what is happening tbh... alot of my setting keep getting overriden

sturdy coral
#

@full junco yeah $199 on those already is pretty awesome (or bad if they are clearing stock and canceling production lol)

#

already out of stock =/, the lenovo one is still $199 though

#

it's tempting to grab one

sturdy coral
#

yeah I'm going to grab it.. trying to decide between it and the HP

#

I've heard the HP is really comfortable

#

but all I want is the lightest weight one possible

#

both flip up which is nice for dev

full junco
#

@sturdy coral wanna buy two and send me one? haha 😄

sturdy coral
#

@full junco Picked up the lenovo. Forgot to get yours 😖

full junco
#

@sturdy coral too bad! well, they will likely soon be discounted in germany too

sturdy coral
#

yeah, since they just launched I didn't think we would see any discount for christmas, it is pretty awesome that they did it

full junco
#

just like the german journalist predicted 😄 I mentioned what he said on launch of the headsets

sturdy coral
#

$200 with the controllers is pretty insane.. PCs just need to get cheaper now

full junco
#

the hardware is so cheap that they likely still make a profit at $199

glossy agate
#

So I think the build is failing due to namespace failure. "error CS0101" in my full body plugin. Anyone know of an easy way to fix it? Log is attached.

tired tree
#

still surprising that it dropped that fast

#

and that they didn't choose to do it for cyber monday

glossy agate
#

Now its like the same price as that star wars AR thing, but way better

sturdy coral
#

@glossy agate it's saying 'FullBodyIKPluginEditor' was defined twice in the same namespace, can you post that build.cs?

#

@tired tree I'm not sure Valve is getting the price down fast enough to be competitive with this stuff, lighthouse 2.0 is $60 per base station in bulk to implementors which with two is over half the retail price of these on sale

#

though of course you'll get much better hand tracking with lighthouse

tired tree
#

60 bulk to manufacturers? that seems way too much

glossy agate
sturdy coral
#

maybe thats more the price targeted at people that are doing testing

tired tree
#

yeah i would assume

#

@glossy agate you didn't install BIK at the engine level did you?

glossy agate
#

No. Project level

full junco
#

@sturdy coral yeah, at $120 for 2 lighthouses it's pretty impossible for valve to compete

sturdy coral
#

MS just needs to release a standalone camera that you plug into the PC or that communcates wirelessly that tracks your hands when they are behind your head

full junco
#

why not just put a camera into every controller

#

well yeah, bandwidth

sturdy coral
#

yeah but it seems like it would just need to communicate features in the environment to get a shared lock with the HMD

#

then only transmit the location/orientation data

#

one camera on the controller might not be enough

#

it would possibly get blacked by the other hand

sly elk
#

i really like the idea of an easy to set up $200 headset. that doesnt include motion controllers, does it?

sturdy coral
#

but if you could also mix in the same LEDs and HMD's camera maybe not

#

@sly elk it does

#

that's what is crazy about it

sly elk
#

thats awesome

#

that opens up the audience for all of us. Its much easier pitch telling someone they need 400+ in hardware and they need to mount stuff on their walls

sturdy coral
#

definitely worth picking one up, especially after we saw that mention of it working with xbox in the future in one of the git commits

sly elk
#

*than telling someone

full junco
#

@sly elk they still need a $1000+ PC

sly elk
#

I thought MS decided not to do xbox support?

full junco
#

they said they will do xbox support

#

but not this year

sturdy coral
#

my guess is it will only work with the one x

full junco
#

yeah that would be good

sly elk
#

i haven't worked on any xbox game with this iteration. Whats the compute power like compared to PS4?

full junco
#

the xbox one x has a significantly better gpu than the ps4 pro

#

the xbox one without x is worse than the ps4 without pro

#

so xbox one without x really couldn't do good VR

sturdy coral
#

yeah unless they maybe let it run at 75hz or something

full junco
#

@sturdy coral just testing discord stuff, could you please tag me once? 😄

sturdy coral
#

@full junco

full junco
#

thanks!

#

even 75 hz is too much

#

xbox one is really a 30 fps console at 900p or less

granite jacinth
#

oh

#

No wonder Rifts w/ Touch are $400

#

They don't include xbox controller or that other controller

#

or that rockband thingie

#

btw, PSVR for $200 on dell

fresh laurel
#

oh nice

pale burrow
#

haptic help! haptics only working on xbox controller, not vr controllers... and reason?

wicked oak
#

Scorpio would be a great VR machine, with the GPU it has

#

base xbox would not be able to do anything

#

it cant even do 1080p at 30 fps

#

look at all games being 700p or similar

full junco
#

you saw that?

sturdy coral
#

@full junco lol no, but that's the same area of code I was just working on to fix this thing

#

I'll test his PR out

#

I think they fixed it in master too, building that to test

#

@full junco you should definitely look at master, the temporal super sampling thing looks really cool

#

and their dynamic resolution algorithm has a lot of interesting stuff for rejecting outlier frames and stuff

sturdy coral
#

(playercontroller ticks when you open it in blueprint editor and crashes if you are doing anything with the game instance, world, etc.)

full junco
#

@sturdy coral oh, I had such an issue like 2.5 years ago 😄

#

and damn, I don't want to have a reason to update my engine

wicked oak
#

@full junco you could have a look at the new resolution scaling TAA code

#

but is such a shame commits are superhuge so you cant just follow the specific features...

full junco
#

@wicked oak the amount of renderer changes between 4.15 and 4.19 are way too huge by now

wicked oak
#

i still have DWVR on 4.16

#

i wont update it unless i can port it to xbox and i need the latest version XD

full junco
#

well your game also is quite fine as it is

#

I plan to work on my game for a long time, so I will definitely update it at some point

#

in 4.15 not even BP nativization worked in a good way

#

and stuff like DF shadows were made twice as fast in 4.16

#

would be stupid if I would never get all the improvements

#

but I'll have to take a clean engine and cherry pick every single one of my >100 changes

#

and then fix all the conflicts that gives

sturdy coral
#

@full junco with the superresolution stuff you can run at like half screen percentage in 2D but get some ghosting

#
        // Conclusions:
        //      Good antighosting at sp=50%.
        //      Excellent antighosting at sp=83%.
        //      Excellent anti aliasing at sp=50% and 83%.
        //      Pixel shader animation are OK
        //
        // Room for improvement.
        #define AA_UPSAMPLE_CLAMPING_METHOD 3```
#

I tried it out it looks really good in the viewport, slight win in VR

full junco
#

so way more usable in 2d than in VR?

sturdy coral
#

I think just because the view stays still a lot more

#

I haven't tested it much moving around in 2d

#

it says it has ghosting issues until around 83% screen percentage

full junco
#

and it depends on TAA, right? so with forward+MSAA you can't use it?

sturdy coral
#

right

full junco
#

can you configure it on some way?

sturdy coral
#

all I've figured out is

full junco
#

like with the TAA current frame weight? do those TAA settings affect it?

sturdy coral
#

you set vr.pixeldensity 1.2 or so so the main buffer is bigger than the "ideal" buffer for VR

#

then set r.temporalaaupsampling to 1 to turn it on

#

then r.screenpercentage to lower the resolution you render at and it gets upsampled to the higher thing you set with pixeldensity

#

when you get too extreme with it it flickers even on static stuff, the old temporal filter was all at the subpixel and only had to searched neighboring pixels for change in a little cross hatch pattern

#

this one has to test against more surrounding pixels beyond the immediate neighbors and that seems to be the main thing they are tuning

#

would probably benefit from that ps4 pro id buffer stuff we were discussing

full junco
#

@sturdy coral interesting

#

and how expensive is it?

#

so much much time is spent on the TAA upsampling?

sturdy coral
full junco
#

hm, 0.8 ms per eye?

#

thats really quite expensive

sturdy coral
#

that was upscaling to 1.3PD

#

PD is a multiplier on recommended screen size

#

which is already 1.33X the native screen res on rift

#

total it would be like running at r.screenpercentage 176

#

in the old terms

#

what I had it was vr.pixeldensity 1.3, r.screenpercentage 70

full junco
#

well, but an additional 1.7 ms is not that much faster than adding 30% more real pixels?

sturdy coral
#

@full junco well it is in both axises

#

it puts the resolution it is going from and to in the profiler text

#
Out[3]: 0.49027777777777776```
#

so about 2x the pixels for a 1.7ms cost

#

I'm going to have to test it on a real scene with smooth locomotion and stuff to see how low you can take the screen percentage comfortably

#

I don't think it adds more cost to the TAA shader to go lower, but it might if it widens its neighborhood search or something

full junco
#

ok

#

does it actually make sense to make it do 2x the pixels?

#

in VR?

#

how does it scale regarding performance, so if you only go for 1.3x the pixels is it significantly less than 0.8 ms per eye?

sturdy coral
#

it is a good improvement in what I was looking at, but it was just the vr template startup map

#

just looking at the text in it

wicked oak
#

so you testing Master branch @sturdy coral ?

sturdy coral
#

@wicked oak yeah, I built it to see if it fixed this rift via steamvr issue

wicked oak
#

sweet

sturdy coral
#

it seems to

wicked oak
#

they do say that the automatic res stuff doesnt work well on vr

#

seems it does?

sturdy coral
#

@wicked oak auto res is different

full junco
#

@wicked oak and where do they say that?

sturdy coral
#

I haven't tested that

wicked oak
#

ah, ok

#

so the new TAA is definitely better?

sturdy coral
#

so far I think so

wicked oak
#

becouse im going to use it for the Aquila project in ps4

full junco
#

its not about TAA, its about upsampling I think?

sturdy coral
#

I need to test it on some real scenes

full junco
#

the AA is same like always?

wicked oak
#

PS4 doesnt support msaa well (actualling there is msaa stuff in Master), but i use Forwrad +TAA

sturdy coral
#

it is "TAA upsampling"

#

instead of jittering the camera a half pixel

#

it is jittering it all over the place

wicked oak
#

and it looks ok in vr?

full junco
#

@sturdy coral hm wait, does it "replace" the regular TAA? I thought its an additional thing?

sturdy coral
#

at 50% it is jittering a whole pixel, and then somehow it is sparsely sampling the results I guess

#

so that it doesn't blur over everything

#

it is just an option on normal TAA

#

I think the dynamic res stuff is related

full junco
#

so you don't have any other TAA time in the profiler apart from the 0.8 ms per eye? no normal TAA time?

sturdy coral
#

in that in 2d they are basically going to have the final output buffer always screen res

wicked oak
#

seems to definitely be for Switch

sturdy coral
#

and dynamically scale the intermediate one

wicked oak
#

and ps4pro/scorpio

#

wich allways use less resolution than the native panel display

sturdy coral
#

it will probably work really well with that type of scaling in VR once they implement it

#

should make the brief resolution changes less noticable

wicked oak
#

very interesting

sturdy coral
#

I need to put out a demo on reddit to capitalize on all the backlash about Fallout's Temporal AA 😛

wicked oak
#

on ps4 i use Forward + TAA

#

it actually works super well

#

becouse the ps4 runs hella fast in forward

#

on my current prototype forward gives me 4 ms back

#

normally the ps4 has some bandwith limitations, so the GBuffer uses far too many resources

full junco
#

@sturdy coral once they implement what?

sturdy coral
#

@full junco the dynamic res part

#

they have it implemented in 2d but one commit message said it wasn't implemented in vr

full junco
#

ah

sturdy coral
#

but there are a lot of commits afterwards

wicked oak
#

dynamic res is pretty much needed for VR. @full junco you definitely know about that XD

sturdy coral
#

yeah but the combination of the two with this thing is going to be a lot better I think

full junco
#

yeah, I just thought it would already work in VR @wicked oak

sturdy coral
#

text won't get blurry if it briefly drops and picks back up

wicked oak
#

seems i wont be using phyre engine or unity lol

#

this stuff is hella sexy for my needs

sturdy coral
#

as long as your head is pretty stable and the scene isn't moving it basically jut builds a much higher resolution view

full junco
#

yeah thats very good

sturdy coral
#

as you move and stuff I'm sure there will be limitations

#

foliage and stuff who knows

wicked oak
#

current TAA has some serious ghosting on my enemies

sturdy coral
#

but basically lots of new stuff to tune

wicked oak
#

i need to tweak it a bit to not have them ghost that much

full junco
#

@sturdy coral can you show me a screenshot of the full PP stuff in the gpu profiler with the upsampling enabled?

sturdy coral
#

@full junco nope, I'm back building 4.18 now to test that pull request

full junco
#

ah

#

but was there still a "normal" TAA time?

#

or only the upsampling one?

sturdy coral
#

just that one I think

full junco
#

hm, then the 0.8 ms is actually really not bad

#

normal TAA is also like 0.5 ms per eye at least

wicked oak
#

normal TAA is 0.5 per eye

#

bit more on ps4 cus its trash gpu

#

so this high end taa being 0.8 seems good

#

is it per-eye, or in general @sturdy coral ?

full junco
#

its 0.8 per eye

wicked oak
#

becouse if its 1 ms per-eye, 2 ms total for the TAA is kind of ridiculous

#

welp

#

thats one fifth of my budget

sturdy coral
#

@wicked oak yeah that was upsampling to something really high though

full junco
#

2 ms is just twice of 1 ms

#

and 2 ms is 1 ms more than 1 ms

#

I got to that conclusion without a calculator

wicked oak
#

im targetting 120 fps on ps4pro, at full res on the psvr

sturdy coral
#

equiv to what would have in the past been r.screenpercentage 176

wicked oak
#

ill definitely need the automatic resolution stuff

sturdy coral
#

I'm on a 980ti

wicked oak
#

so res drops when there are tons of FX and explosions around

full junco
#

@sturdy coral ah, so then it might be even quicker on a 1080

sturdy coral
#

ah I had it overclocked though

full junco
#

ah

sturdy coral
#

had it overclocked from oculus medium

full junco
#

oculus medium settings, not oculus high settings?

#

😄

sturdy coral
#

hah

#

I also had like a billion browser windows open so that might have interfered

full junco
#

no, thats normal

#

I also always have a billion of them opened

#

I still have tabs from 4 months ago opened

sturdy coral
#

yeah just don't leave them open on shadertoy

wicked oak
#

somehow gunheart runs at 20 fps for me

#

in vr

sturdy coral
full junco
#

ha, well, that guy is doing PRs about a lot of stuff

#

I see his name all the time

dusky moon
#

Any1 knows about Viveport's revenue stream compared to other platforms like steam and oculus ?!

#

wish there was steamspy for that as well

wicked oak
#

@dusky moon last time i looked at, its dreadful

#

and HTC takes 40% of the profit

dusky moon
#

@wicked oak oops! I thought maybe the chinese market could be as equal as at least oculus store

granite jacinth
#

If you want the chinese market, contact Tencent

glossy agate
#

Last I looked it was 30% on Viveport, but they had a winter promotion where devs got 100% for a bit. For me though 100% of 0 doesn't help

#

Its bloatware, and I deleted it off my computer. Don't personally know of anyone using it as their main sales platform

full junco
#

100% of 0? wow 😄

glossy agate
#

Yeah I only promoted the steam link haha

#

And I can’t write Chinese to reach the other market. Vive port did a free translation of the description though.

pearl tangle
#

they have 0% cut for the remainder of the year I think. Which started a few months back. Other than that I think they sit at 30% same as steam

raven halo
#

seems like someone at Oculus has been testing the multires feature with Daedalus! I'm happy! 😄

wicked oak
#

doesnt seem to have the improvements the pc version of multires has

#

but i guess thats becouse its less agressive

#

still, free perf!

mental chasm
#

Hmm

#

Has anyone here experimented a lot with asymmetrical setups?

#

like desktop/VR in local play I mean

#

like Keep Talking and Nobody Explodes

#

Is there any way to make UE run both a desktop viewport and a VR set at the same time? Looking around in the engine it doesn't really seem like an option

#

the preferred way to do it seems to be by launching multiple clients and a network game

#

but has anyone here had any success using an alternate path?

wicked oak
#

@mental chasm you can use render to texture with a camera, and put that texture as the screen

#

it IS possible to do asimetrical stuff in unreal, perfectly possible

#

its just a HUUUUUUUUUUUUUUUUGE performance drop

mental chasm
#

hehe, at that point it might be easier just to do the network solution, although it is an intreaguing solution for sure

tired tree
#

nah its pretty easy now to mirror to the screen

#

its just the perf drop and difficulty of menus that are the problems

mental chasm
#

hmm, well it's not about the hardware mirror

#

but coop using desktop as a different player viewport 😃

#

but the points are still valid I suppose

tired tree
#

you can do it with a camera and by seperating keyboard / mouse input to player 2

#

but like I said, it doesn't use a seperate viewport, so you can't just add widgets

#

they have it on their todo list to make mirroring be able to be an entirely seperate viewport

sturdy coral
#

@raven halo nice. I wonder why it doesn't help much with bandwidth limited stuff? Maybe it still reading textures at the mip level they would be at without it?

wicked oak
#

@muchcharles#2724 most likely, becouse it if it tried to sample one lower mipmap, it could cause extra artifacts

#

given they are doing this at the driver level, so it works backwards compat, its probably the absolute safest implementation you can do

sly elk
#

I'

wicked oak
#

some lightbake shanengigans

#

outside of ue4

#

im baking the procedural geometry in houdini and importing the textures, then building a mat

#

i blame ue4 for not letting me lightmap from the editor into a random object

#

those are light bake, cavity, AO

#

all generated on any object with 1 click 😃

sly elk
#

what are you baking in?

#

oh

#

houdini

#

I co-own handplane. You can bake from command line, might be useful for you

wicked oak
#

i can bake from comandline with houdini too

#

eventually ill have a "BAKE ALL" button

#

right now my bake button is a single button in the editor, that bakes the current object

#

ok now floor is also lightmap-d

wicked oak
#

and now i know how the fuck to use the renderer

#

seems to be kind of noise. but given the way the islands are done, i might be able to run a blur filter on it

glossy agate
#

May just be the map sizes if its pixelated AO. There is no actual lightmass bake right?

wicked oak
#

no

#

its pixellated becouse this is fully raytraced

#

its not like lightmass wich uses photon mapping

#

its an actuall full movie style raytracing

#

that just bakes the colors into the pixels

#

then i modulate it to be browner with the material

#

and well, not be brown on the floor

#

this is how my bake looks right nwo

#

the black zones are becouse they are on the other side (the wall has outside side)

glossy agate
#

Ah got it. Similar to V-ray and Keyshot then where it has a little static look to it

eager pine
#

i guess this is the vr place? xD

glossy agate
#

Nah its all Hentai here man

chilly thicket
#

Just wondering, could anyone direct me to a good place for learning about how to make a bow for VR?

#

The only real thing I can seem to find is the Live Training on making it, which isn't very helpful.

real needle
#

@chilly thicket Is it not? I think they provided a project as well

chilly thicket
#

Yeah, I found that project, I'm sitll having a bit of trouble grasping how it works though.

sturdy coral
#

@chilly thicket you can download the VR funhouse source too

eager pine
#

Okay, tried making my own character for the vr, when I start it's like 1/3 to high or low for the height, and i can't make the motion controllers track

sly elk
#

I have been dealing with system hangs about 1-2 times a day and im going to reinstall windows today to deal with it. Since im already doing a thorough back up, how many people here are on win10 for VR dev? I'm thinking now is a good time to switch to win10

eager pine
#

@sly elk i am

#

but having problem with ue4 and vr really

#

cant seem to get the template right

#

and the motion controllers dont update location

#

so weird

#

why is this happening

#

this is the htc vive on the floor but its aboce the floor

sly elk
#

If you load up a clean template file and hit play, it doesn't work correctly

#

?

raven halo
eager pine
#

Yes that works, but i want to setup my own character, now i get the head to work, but the restriction area and head is over the ground when the headset is on ground, and the character vr spawns the controllers but they arent tracked to current position, they are stuck to the ground

eager pine
#

Im using the vr expansion plugin character vr

granite jacinth
#

Top Co-op VR games?

tired tree
#

@eager pine did you offset a scene component like in the Epic tutorial? you don't want to do that with my character

glossy agate
#

@granite jacinth for me serious Sam games are so fun coop

#

8 person coop.

sturdy coral
#

got that mixed reality headset in, I like it so far, it is definitely lighter weight

#

it is clearer in the center than vive or rift, but blurrier on the edges

real needle
#

@sturdy coral didn't you discover the same issue w steamVR and windows Mr, as with Oculus and steamVR?

#

Other users reported the same problem

sturdy coral
#

@real needle I haven't tested it yet, it will be interesting to see if it has the same aspect ratio mismatch thing as rift

#

or if I was wrong about that being the reason

#

4.19/master seems to fix things partly by storing the coordinates from steam for the ideal buffer size in a rect though which is similar to the fix I implemented and it seems to work fine with Rift

uneven moon
#

I'm trying to get my flashlight to reach further

#

I can really only see things that are somewhat close to me with it

#

It's a movable light, no shadows, attenuation radius is set to high (50k)

#

Intensity of 20,000

#

Am I missing something?

#

p.s. I have exp fog in my level

real needle
#

@uneven moon There's a max distance to attenuation. Just place a spotlight in the scene and see if its reaching it's max attenuation limit

eager pine
#

@tired tree no, i only made a vr character, changed no default values really

spiral zephyr
eager pine
#

really weird @tired tree i just downloaded 18.2, started a whole new project, migrated your plugin, placed the vr character, and still above ground even tough the headset is on the ground and the standing height is like 1/2 to high

#

it seems as it crashes with the vapsule component?

eager pine
#

D:\Marius\VR\Prosjekt filer\VrTrainingSimulator\Plugins

#

ops

eager pine
#

it seems as the headset is offsett as well

#

hmm

#

the printstring appears to far left

eager pine
#

anyway

#

used this method, but it seems everytime i grab it twiches

tired tree
#

@eager pine bad room calibration can cause a vertical offset, but also characters float a couple of millimeters above the ground anyway]

#

1/2 inch is more likely to be stale room calibration though

#

print strings will also always appear far left, that is how their print strings show in VR, they don't account for the HMD eyes with them

eager pine
#

But the room calibration is good, its perfect anywhere else and! With the standard template from ue4 vr, works perfectly on the ground, but with the vr expansion its above ground

#

Yes thr print strings i agree, just had them in the middle before like in 4.16

#

Its just weird as i see other youtubers have their character on the ground with the vr expansion

sturdy coral
#

@eager pine adjust const float UCharacterMovementComponent::MIN_FLOOR_DIST = 1.9f; const float UCharacterMovementComponent::MAX_FLOOR_DIST = 2.4f; in Engine/Source/Runtime/Engine/Private/Components/CharacterMovementComponent.cpp

#

but they are probably there for a reason, depenetration efficiency, etc.

eager pine
#

But will this affect the vr expansion character vr?

sturdy coral
#

@eager pine it should, test and see

eager pine
#

But why is the vr expansion character above ground, but not the normal template vr from epic?

tired tree
#

because the normal template character doesn't use a movement component

#

and yes, its for depenetration, it can be adjusted but by how much before you start getting things like falling through the floor on peaks I don't know for sure

#

you can move the NetSmoother component down a little bit on the character to adjust

#

that min floor distance of 1.9 down on Z would bring it to 0-0.5 units from the floor

#

come to think fo it, i should probably do that by default

tired tree
#

@eager pine there...added the offset by default

#

never thought to do that before

eager pine
#

wait, so the offset is now default on the download plugin?

#

😃

#

@tired tree

#

Forgot to tag sorry

tired tree
#

the repository yes

sly elk
#

just got a 4th oculus sensor

#

much better tracking

eager pine
#

Thanks morden! Ill check it out!

#

but im still long off the floor

#

do i have to replace the character? make a new one?

#

thats the offsett right

#

accordingly its off by -95 @tired tree

tired tree
#

1.9

#

if yours is off by 96, then you are adding an offset

#

you aren't using the BASEcharacter are you? had someone do that once

#

need to flag that as not blueprintable

#

need to use the VRCharacter

#

you can check out how it is set up in the template project

#

to save some time

#

off by 96 is caused by using epics pawn method

#

my plugin char doesn't require the -96 offset on the root

#

so you have something setup wrong

eager pine
#

Im using the vrcharacter, blank template, using vr expansion plugin

#

O hold on

#

Not the base character?

#

Not that one?

tired tree
#

PM me further so we don't spam here, but yeah, not VRBaseCharacter, you want VRCharacter as the parent

wicked oak
#

this small application lets you do easy screenshots to post on places

#

its called snipping tool

iron talon
#

ive started following that climbing tutorial, are there any major pros and cons to using that vs somehow migrating the openvr plugin to a new project?

eager pine
#

Use lightshot @wicked oak much better

wicked birch
#

@tired tree Which build would you recommend if I'm on 4.18.2?

abstract panther
#

Any recommendation what VR to buy if you gonna get into VR game dev in unreal ?

glossy agate
#

@abstract panther Either of the main 2 will work fine

eager pine
#

@iron talon what do you mean?

#

is there a way to check if the player in vr is already in climbing mode?

abstract panther
#

@glossy agate Thanks!

tired tree
#

@wicked birch 4.18.2 is what i would recommend

wicked birch
#

Just to be certain, is that the one you posted on the forum thread?

tired tree
#

the plugin itself is from 4.12 - 4.18 different builds, the template is always on latest so 4.18.2 currently

wicked birch
#

Alright, yeah I figured the template was on the latest UE4 build

#

Just wasn't sure if I could download the latest build version and compile it myself

#

But I think you had some 4.19 stuff going on

tired tree
#

nah, i don't start porting to a new version (usually) until the first preview is out, they tend to revert commits sometimes

#

its best to not use the pre-compiled binaries, they are there for people that need them for w/e reason

wicked birch
#

Ah yes I recall you saying that to me before.

#

I'll just grab your master branch and build that

glossy flame
#

Anyone know of a good way to rotate VR player character yaw standard r-analog style, like how it's done in Alien Isolation? It works pretty good there. I've spent a bit of time trying to see how it's done, but I get strange results trying to rotate the Actor, or using Rotate VR character. most of the threads I could find on the topic were using the first version of the Epic VR template and they don't seem to apply to contemporary VR player character setups.

tired tree
#

You have to offset by the HMD location as a pivot

#

or, since you are using my plugin, you can use the new AddRotationOffsetVR node and rotate by the delatime * speed

glossy flame
#

Thanks a lot, MordenTral!

full junco
#

last dev-rendering merge was surprisingly boring

novel tree
#

@eager pine @wicked oak for screenshots there is only one good app and that is ShareX, try it, powerful stuff :3

balmy surge
#

Has anyone done anything with movement with Leapmotion in VR

fresh laurel
tired tree
#

@full junco well considering how much they have been packing into 4.19 so far....might be good to slow down and stabalize?

wicked oak
#

if the dynamic res and new TAA make it for the next update it will be great

#

becouse i need it lol

#

im going to target 90 fps on base ps4 and 120 on pro. So im going to need as much performance as i can get

#

a high speed shooter like what im going to do will double-image a lot with the 60 fps to 120 thing

#

went to have a look at the commits

#

WTF is the Ocean project?

#

the next Epic game?

granite jacinth
#

Could be

#

BossKey dude left for secret project

#

Or could just be a codename for an existing project

tired tree
#

mmm, they are updating the launcher today

chilly thicket
#

Random simple question.

#

I have a VR Pawn blueprint, and I'm trying to get the forward vector of a mesh that's part of it, from another actor blueprint.

#

I'm trying to cast, but I can't figure out why it isn't working.

granite jacinth
#

Doesnt' seem like a VR question

#

But, you should read what I've been telling the other guy in there

#

Basically watch that video about Blueprint Communcation

wicked oak
#

tfw you are debugging shit on materials and do some wrong shit yet you create art

#

lol

tired tree
#

What is this? VBlanco games don't have colors!

wicked oak
#

made a high res screenshot

#

or this

spiral zephyr
#

these are def my colors

wicked oak
#

this is the material

granite jacinth
#

@wicked oak You definitely want to make peopel sick

wicked oak
#

plug this into emmisive

granite jacinth
#

My brain is already saying, no

#

But I am on the extreme end of mot/sim-sickness folk

#

So yeah, if anyone wants me to playtest their game, I am definitely the best subject around

wicked oak
#

the actual material is this

#

im trying to get a look i like while bypassing unreal shaders completely

#

this is just plugged into emmisive, its on an "unlit" material

#

(the floor isnt,yet)

#

the blue light is lightmap

tired tree
#

why are you doing fauceted normals

wicked oak
#

trying things

#

i dont want a fully realistic look

tired tree
#

ddx, ddy isn't the fastest thing ever

granite jacinth
#

what is ddx,y anyway

wicked oak
#

my stats say 5 instructions @tired tree

#

for the pixel shader

tired tree
#

derivatives

granite jacinth
#

They give the approximate screenspace derivative of the value you request. The most common use is to compute your own mipmap level, which can be passed into the texture sample node.

#

shader nonsense

tired tree
#

they are very useful

granite jacinth
#

I'm sure

#

But still no clue what they do

wicked oak
#

arent they calculated by dedicated hardware, and on a primitive basis?

granite jacinth
#

what's a screenspace derivative?

tired tree
#

crossing them gets a worldspace normal

granite jacinth
#

Screen-space partial derivatives are awesome. They let you compute the approximate partial derivative along x and y for any value, which comes in handy in many different situations. ... In GLSL, the partial derivative is dFdx(value) for x and dFdy(value) for y.

wicked oak
#

i was going to use them on everything i wanted to be faceted

granite jacinth
#

More freaking nonsense

wicked oak
#

before actually doing the faceting myself

#

as normal faceting will duplicate the vertices

#

and increase the vertex shader cost due to that

granite jacinth
#

I guess I should read into that one day (never)

granite jacinth
#

What's the cheapest way to outline in VR nowadays?

#

this still it?

tired tree
#

depends on how much you are outlining

#

that would be cheaper if you are outlining a lot

#

but you can turn off custom depth generally if you aren't

#

and use a mesh mirror or material outline like RR uses

wicked oak
#

material outline is the best pretty much allways

#

but of course, double triangles

#

so it only really works for specific objects

#

postprocess outline is kind of absolute garbage, doesnt work very well (and its expensive). but it doesnt double triangles/drawcalls

granite jacinth
#

meh, only using it every once in awhile and I do turn it off

wicked oak
#

then use geo outlines

#

there is a marketplace pack that has them for like 2 dollars

real needle
#

I have a huge problem with pp outlines and my scope. Anything that doesn't do depth test is a problem with the second perspective from the scope

#

I tried disabling the "no depth test" objects in the scope, but seeing it outside of scope and not inside is very confusing

nimble junco
#

hey guys, is there a way to register the motion controllers from vive? i need a way to ask them if they are being turned on/off or if they are being tracked/static

#

but that doesnt really work, or i dontcorrectly understand how it works

mighty carbon
wicked oak
#

i have the lower res version of that one

#

it has been useful for prototypes

#

that one looks good

#

jesus christ is a fucking huge pack

#

with varied biomes

mighty carbon
#

yeah

wicked oak
#

you can make a whole game out of it no problem

#

and a decent one

granite jacinth
#

Interesting

#

With all the spam that comes into this channel

#

I am super surprised no one's linked this here yet

tired tree
#

finally

wicked oak
#

finally indeed

#

they already had it for oculus

#

vr.oculus.pd

#

much better to be like that, makes it easy

#

now i can have like base ps4 on pd 1, and ps4pro on 1.2

#

or some other stuff like that

sturdy coral
#

@real needle ^

#

Also fixes the Oculus and windows MR via steam vr issues we are seeing on 4.18

#

For example, if you wanted to reduce the number of pixels to render of your VR project by about 20% you would set vr.PixelDensity=1-0.2. If you wanted to increase the number of pixels to render of your VR project by about 20%, you would input vr.PixelDensity=1+0.2. that's ... wrong, total needs to be sqrted...

sturdy coral
real needle
#

So when is the adaptive resolution implemented? I'm assuming the pixel density work is one of the things they wanted to do before to generalize it

wicked oak
#

most likely for 4.20

#

becouse its on the blazing new "master" branch, they probably already locked the 4.19 branch?

#

if its included for 4.19 ill be super happy

#

this is so next level man

#

too bad its exclusive to ps4

#

its like oculus medium but on steroids

sturdy coral
#

@real needle It is already in master for non-vr, the initial commit of it didn't have vr support but later ones might have

#

Yeah Dreams looks really great

#

It is a shame it is stuck as a PS4 exclusive, they could have had a shot at creating the metaverse

#

I will definitely pick up psvr for it, it isn't launching with vr support though but it is supposed to be coming in 2018

wicked oak
#

Dreams on PC for VR PC controllers could be a killer app for any artist

#

oculus medium is neat but this is on a whole other level

#

it lets you do actual "concept" art and kitbash pieces randomly in 3d really well

wicked oak
#

does that mean the stuff before it will be 4.19?

raven halo
#

I think usually that's been the case yeah

sturdy coral
#

medium has one big advantage that probably won't be possible with Dreams' approach

#

the move tool

#

but dreams doesn't suffer from the same kind of voxel resolution limitations as medium

#

it definitely seems better for kit bashing, which is where VR seems to have the biggest advantage over traditional sculpting

#

just positioning and orienting things is really intuitive and fast

#

and then dreams has a whole bunch of other stuff, a whole game engine and audio, and just way better look overall

#

Dreams keeps things as an edit list of signed distance field primitives and can blend them together nices, and then it continuously generates splats for it. medium keeps everything all in one signed distance field grid per layer and continuously turns it into a mesh, so it can't soft-blend primitives together in the same way and it runs into performance problems pretty quick

#

but it uses a geometry shader on its mesh representation to implement the move tool, and then reconverts the result to the SDF grid, which is really really nice

tired tree
#

oh, so 4.19 should be due out any time now then

sage gulch
#

yeah that seemed to work nicely without a headset.. cute interface

#

If I make a test level and run it for a while, eventually I start getting duplicated hand meshes that are just barely offset from the initial hand meshes, as well as a copy that is stuck at the player spawn but still responds to translation and rotation of the motion controller

#

anyone else seen this? known issue or fix? is this possibly related to my level being 'persistant' and if so how can I resolve this?

#

thanks

#

I don't really see how persistant would be the problem because it doesn't crop up until I've been using the level for a while and adding things into it.. making a new map and the problem is gone, doesn't seem like it's in the character necessarily,.. when the player dies it's still a bit messy, but if persistant was the issue, it seems like it should have a duplicated mesh after the first time I test the map

tired tree
#

you are spawning extra pawns

sage gulch
#

yeah, like crazy

#

the character has a whole bunch nested, and the MotionControllerPawn BP is one of them

#

it's placed as a child actor in the component tree

#

then there are a bunch of other, relatively unrelated child actors, as well

#

that's not a bad practice, is it?

sturdy coral
#

@tired tree there's a 4.19 branch now

weary flame
#

Hi guys....I want to try the Nvidia VRWorks branch to use all the very promising features (like the VR SLI) but all I find of it is this page
https://developer.nvidia.com/vrworks

I can't find any deeper info, do u know if there is a guide that explains better the usage?
I already compiled the branch and opened my project with it. The only thing I find about to enable my features are some checkboxes in the project settings.
I enabled them but it doesn't seem to do big changes in performances.
Anybody who already used that branch knows something more?
Thanks

wicked oak
#

there are some console commands

#

there are no documentation, gotta find them in the source XD

quartz acorn
#

Hello guys

#

We are working on a UE4 Vive project and want to use Depth of Image to render Target, but facing issues for same

#

Here is some more details of what we working, if anyone can take a look and help us finding the issue, that will be nice

tired tree
#

@sage gulch well its why you are seeing multiple controllers, you have multiple pawns not getting destroyed on your map....

sage gulch
#

shouldn't I see it after the first time I run a map then?

tired tree
#

also...why are you nesting multiple pawns (I could see 1 inside a character)

#

and child actors are still bug ridden

sage gulch
#

is it a bad practice? I have independant pawn sensing on some, for example

#

so they can each acquire targets.. and well this is third person and first person simultaneously

#

it all works reasonalby well until I try and do something with the motion controllers but I'm a bit of a n00b

tired tree
#

should be able to run both third and first with a single pawn is the thing....

#

sounds like one big ball of yarn and overly complicated

sage gulch
#

yeah I saw an example that made the hands inside the pawn, but the pack that comes with the engine chooses to spawn them and attach them , so.. I dunno

#

yeah it is getting there 😉

#

you can attach 16-Bits.. to your head

#

point and shoot with extreme vengence :p

#

I think I could possibly lower the number of sensors in a few places

#

there's like... at least 3 dozen pawns

#

heh

tired tree
#

what......

sage gulch
#

fake it till ya make it, make it till ya break it, roll on kickstarter

#

suffice it to say I've rolled back from 4.18

#

it's mostly a bunch of attached turrets basically, plus a couple characters that can swap a few interaction modes

trail shale
#

If I see the steam VR splash screen when I am using level streaming, am I doing something wrong? I've tried not blocking on load but it still throws you the steam VR and breaks immersion..every tutorial I've seen is non vr

mighty carbon
#

and

wicked oak
#

man they are spider goggles

mighty carbon
#

something tells me it will be on par with ARkit and not Lightfields

tired tree
#

a six second "meet" with a photoshopped goggle

#

wow impressive

#

the actual page doesn't go much over that either...

spiral zephyr
#

The sigur ros-demo doesnt seem to show anything that ARcore/kit can't do, except hand tracking i guess. Hand tracking is very possible incorporating existing ML stuff or similar, too.

#

im 87% confident googles efforts will evolve faster, and "win" sometime late 2018

wicked oak
#

@tired tree they have been beyond secretive

#

@spiral zephyr not the same thing, this has lightfields and lazors and shit

#

i honestly did not expect magic leap to do anything this soon

#

they truly looked like vapoware at least for 2-3 more years

#

guess its time to start building my own 3d engine 😛

#

unreal or others are not a good fit at all for AR

spiral zephyr
#

how so? Integration with stuff?

#

Or iteration speed? (im noob, but having a real good time making arcore stuff with unreal(except when I wanted bloom and had to sacrifice almost everything to run mobile hdr..))

wicked oak
#

iteration speed + control of shaders + more than anything, performance

#

for ar you arent rendering a world, mostly just a few 3d objects lit by a cubemap

#

so you really do not need ue4 fancy shit at all, and that only does overhead

tired tree
#

don't overlook PBR for AR....

wicked oak
#

PBR is just a few math formulas, you can use that shader on your stuff without trouble

#

making a 3d pbr viewer is not that hard, you can even follow a basic tutorial that does it

#

what is hard is use it for full levels

#

for a single object you just use image based lighting and light it from a cubemap

spiral zephyr
#

Well if your future lightweight (AR)engine has blueprints i'll come!

wicked oak
#

i wouldnt add blueprints

#

just my own scripting system or just in C++

#

i wonder how Godot would do

#

given that is very, very lightweight engine

#

and while it doesnt have culling or things like that, that stuff is not needed for AR

spiral zephyr
#

as long as its highly visual, i can deal with any script interface.. text code always was a struggle, patterns don't stand out.

#

stupid brain, no imagination.

spiral zephyr
#

haha they have "soundfield audio" too, which I assume is the regular spatialization/attenuation/effects/HRTF combo of stuff. Taking pages from Apples book of twisting terms for hype, if its the case.

mighty carbon
#

I wonder if 4.19Pre1 drops this week.. I doubt that, but I hope it will land before Christmas

tired tree
#

it should

#

branch is out and master is locked to 4.20

#

i'm already starting diffs to port

spiral zephyr
#

Nice! The mechanics are impressive and interesting, but I cant get over how good that wrench looks...

#

(unrelated: 4.20 calls for special celebration 👌 )

#

Torque feedback seems intuitiive

mighty carbon
#

@tired tree ohh.. are there any cool features / updates coming in 4.20 ?

sage gulch
#

that alec moody guy is nuts

#

teehee

tired tree
#

depends on what you consider cool

#

dynamic res, VR pixel density, VR model loading natively built in, the networking efficiency changes, those are all pretty cool to me.

#

but I assume you were hoping for dynamic lighting changes :p

sly elk
#

4.2, are we getting any kind of foviated rendering?

glossy agate
#

Isn't that only useful when we have good eye tracking?

#

Or is there another use for it?

sly elk
#

Just more pxiel density at the center where the lenses resolve better

#

we had it on our unreal build for starblood arena and I have wondered why it never made it to PC platform use

glossy agate
#

Nice! What was the center set at compared to the edges? Like SS at 3 in the center, 0.8 at the edges or something?

#

Also is Starblood on Steam? If you have it working I would like to check it out

sly elk
#

yeah, something like taht

#

starblood arena is a ps4 game

mighty carbon
#

@tired tree I wouldn't mind faster dynamic shadows in VR for sure 😃 I don't hold my breath for VR friendly dynamic GI as it's just not feasible any time soon.

glossy agate
#

Starblood arena looks really cool! Also good progress on Wrench. If you auto spawn the correct socket size how will you account for how the 9/16 is usually always missing?

wicked oak
#

also GLTF starting to happen

#

wich, for blender users, is a fucking godsend

#

like, the best thing to ever happen to blender->ue4 interoperation

#

if it works as it should

#

fuck autodesk fbx shanenigans

granite jacinth
#

@wicked oak Hope UE4 gets support asap for it

#

Lightwear that is

wicked oak
#

@sly elk we do. but for gearvr/oculus go

granite jacinth
#

But wonder how much that bitch is going to cost

wicked oak
#

gearvr/go will have multires native in the opengl driver

#

they have already reported 20% perf improvement in nearly all GPU bound games

#

i dont care how much it costs im getting one

granite jacinth
wicked oak
#

unless its completely fucking retarded like 5k dollars

granite jacinth
#

But they have a crapton more money than Meta

#

All those investors will want their money back

#

But they need to sell in bulk if they want this to be as consumer-friendly as they are making it out to be

#

I have a feeling it'll run around $1k at first, alongside Meta2

#

if not lower to cut out any real competition

#

But yeah @wicked oak

#

There definitely needs a dedicated VR/AR game engine

sly elk
#

I really don't get the AR thing right now. All the cool ideas I have for AR fall apart without really good machine vision that can recognize real objects

granite jacinth
#

How don't you?

#

Imagine if Pokemon Go were with Magic Leap

#

Instead of your phone

wicked oak
#

literally pokemon world lmao

granite jacinth
#

Yeah

#

Dude, that shit would be so good

sly elk
#

yeah that doesn't appeal to me. Seems like a VR game that happens to have a real world background

granite jacinth
#

then you are just silly

wicked oak
#

@sly elk but we do have that stuff. Remember magic leap is under googlle

granite jacinth
#

And have no sense of adventure

wicked oak
#

and see all the fancy shit google has

granite jacinth
#

of gamedev

wicked oak
#

at the very least, this thing actually 3d maps your environment

#

it creates a low res 3d mesh of the environment

granite jacinth
#

Yeah, I mean, the potential for AR once it becomes mainstream is going to destroy VR

wicked oak
#

you can then use for collisions/pathfinding/whatever

granite jacinth
#

Unless VR somehows steps up to the plate

#

But, AR is wirless

#

which is already a win

sly elk
#

So to me, what makes the idea of AR compelling is that the real world can interact with the virtual

granite jacinth
#

Well most anyway

sly elk
#

but right now that interaction is limited to being a gameboard

granite jacinth
#

?

#

Not with any real AR google

#

goggle

#

The issue has been price with AR

#

Until Meta2

#

But Meta2 never really got any attention

#

Magic Leap's secretive shit

#

has always kept them in the spotlight

#

It was great marketing

#

As soon as they released this tease today, the whole world went to shit

#

It affects every industry

wicked oak
#

@sly elk but you could do your game on AR

#

and be the exact same thing

#

just on your room

#

instead of in a VR garage

sly elk
#

yeah

wicked oak
#

you could give your 3d helpers to an actual mechanic

sly elk
#

so to me, the idea of what would make AR cool is if it could recognize the parts on your actual car

#

and help you

wicked oak
#

but it could

sly elk
#

but right now, it has no idea what it is seeing. It can't tell you "hey that fastener needs to come out"

wicked oak
#

and even if it cant, you can just show a 3d image of the part, and tell the user to try to locate it

#

once you find that part, you can then use computer vision to "attach" your 3d virtual version to the real version, and be able to add overlays

#

or just tell the user to "calibrate it"

spiral zephyr
#

ARkit doesnt do this, but arcore does remember the whole playsessions registered surfaces. This means, if you tell player to "scan" whole apartment first, then you can set up virtual walls on outside edges of planes, and have a custom fit virtual world. Not a nice autosolution, but there is cool stuff we can do.

wicked oak
#

you can tell the user to identify some very specific bolts, for example. And then use it to align the virtual car with the real car

mighty carbon
#

@wicked oak I thought you were anti-mobile VR 😛

wicked oak
#

but im not

#

im doing stuff with mobile VR at this exact momoent

#

preparing the launch of a game, making sure it runs at 60 fps perfect performance so oculus doesnt whine

#

game has GREAT graphics for a mobile vr game

granite jacinth
#

No it doesn't

#

Your game stinks 😉

mighty carbon
#

sure, but back in the days you were not so cheery 😉

granite jacinth
#

Rahahah

wicked oak
#

this is for the iphone port

granite jacinth
#

Show us the money!

#

That's yours?

wicked oak
#

wich is cardboard

#

no, but i got hired as consultant for this

granite jacinth
#

Oh

wicked oak
#

make sure it works perfect and launches on Gear (wich has stupid high requirements)

granite jacinth
#

Does it?

wicked oak
#

yes

granite jacinth
#

Nice

wicked oak
#

this runs at the quality shown in the pics, at full res with MSAA x4, at 60 fps

#

on gearvr

granite jacinth
#

I think once I get this PSVR port done

#

I'll try my hand at Mobile again

mighty carbon
#

shooting dinos is bizarre 😦 Better have them as pets.

waxen musk
#

What was the hardest part of maintaining a stable 60fps?

wicked oak
#

dealing with fucking unity

granite jacinth
#

Art

wicked oak
#

the issue is that unity is shit

#

but ue4 is huge

granite jacinth
#

Unity is better at VR/AR though

wicked oak
#

unity is a black box

#

and does not abstract VR

spiral zephyr
#

unreal lite, coming 20?? pls

wicked oak
#

this is a HUGE FUCKING PAIN

#

making a project that runs on every single vr headset in unreal is trivial

#

on unity, its a mayor level cluterfuck

granite jacinth
#

Meh, I would love to stick with Unreal for the rest of my life

wicked oak
#

becouse the "native" vr, its just the camera movement

#

literally just that

#

no way of doing settings, tracking zones, controllers, input, anything

#

just a camera that rotates

granite jacinth
#

But, I have a feeling, Epic might get sent packing in the next 5 years if they don't unfuck themselves

wicked oak
#

the rest is with each specific VR plugin

granite jacinth
#

They need to refocus

wicked oak
#

and each of them has their own components and prefabs you gotta use

spiral zephyr
#

vblanco you are describing my nightmares

granite jacinth
#

Other open-sourced / AAA engines are outpacing it in the last year

wicked oak
#

it makes multiplat VR in unity a fucking nightmare

#

and im not exagerating

#

it is a fucking nightmare

#

to port from old cardboard to new cardboard i had to change ALL the player cameras in the game

#

and write a wrapper

#

and to add gear vr, i then had to do it AGAIN

#

and it CONFLICTED

#

if you wanted to add psvr and vive to the mix...

#

literally impossible

spiral zephyr
#

so how many strands of hair you have left now?

wicked oak
#

but i got 3000 dollars from the contract

#

so im ok

spiral zephyr
#

how many hours?

#

of work

wicked oak
#

was a good deal for me anyway, took 1 month

#

im finishing touches

spiral zephyr
#

worth it for 3k

waxen musk
#

I see. Guess I’m glad we picked UE4 for our current situation. I need to maintain a stable 50fps to sync with broadcasting systems (1080i 25fps) for augmented reality with UE4, so I’ve been scratching the surface of performance optimization and especially mipmapping.

wicked oak
#

mipmapping doesnt affect performance

#

it affects a bit of bandwith, but not that much. Its mostly about graphics quality

#

not using mipmaps will alias horribly

mighty carbon
#

UE4 is fine for Gear VR if you stick with ES2 renderer.

waxen musk
#

Especially on interlaced screens 😵😵

wicked oak
#

for VR, you need to setup your mipmap settings properly

#

if you do them right, you get great antialiasing on the textures

waxen musk
#

The horror

#

Good to know that mipmapping isn’t that much of a performance optimizer

granite jacinth
#

After 4.20, we'll see the direction that Epic is going to take UE4

#

But I don't see VR being in the spotlight again for awhile

#

It's a shame they still haven't fixed motion controller replication

wicked oak
#

until apple hires them to do AR stuff

mighty carbon
#

what's there to focus on?

wicked oak
#

i dont think motion controllers should be replicated automatically

#

you do lots of random stuff with them, it would be problematic

#

its not like implementing your own motrion controller replication isnt trivial

#

just sending the transform and lerp-ing to it has worked flawless for me

mighty carbon
#

dynamic GI, docs, AI, performance optimizations, support for non-game related stuff..

wicked oak
#

lul dynamic GI

#

there is absolutely nothing about it on the master branch, they arent even looking at dynamic gi

#

AI is already top tier, anything more is game specific

#

docs need to get stuff done

mighty carbon
#

well, since VR isn't going to be a thing according to @granite jacinth , why not ?

wicked oak
#

performance on what? personally i think they should open the graphics pipeline a bit more

#

so you can easily disable passes and other things

granite jacinth
#

@mighty carbon ?

wicked oak
#

but VR is still a focus on the latest comits. Personally i want FULL VULKAN

granite jacinth
#

@wicked oak ?

#

What focus?

#

one or two commits

#

is not a focus

#

If you think about how big it was focused on earlier thisyear

#

It

#

s been on a steady decline

spiral zephyr
#

Audio revolution has been going on this year, new audio engine soon the default. Its existense is mostly essential bc VR/AR.

granite jacinth
#

Aye

#

I think Audio is definitely going to be a refocus finally

#

Finishing up that Audio Engine

#

And getting VOIP reworked

#

and a few other audio oddities

#

Niagara

#

maybe one day

#

They jsut have a a lot of unfinished crap in there

tired tree
#

?

#

they did MASSIVE VR changes in 4.19

#

they have been abstracting more and more of the back end

granite jacinth
#

Meh, not that big

tired tree
#

just because it isn't "front end featured" doesn't mean they aren't working on it

granite jacinth
#

They did changes because they had to

spiral zephyr
#

All the reworking of media-framework, audio-engine is because VR/AR

granite jacinth
#

It's more like refactoring

#

for XR

tired tree
#

every single plugin and platform got touched heavily in .18 and .19

granite jacinth
#

It's nothing that the end consumer (us) is going to see

tired tree
#

....

granite jacinth
#

Only a few things

tired tree
#

its a damn good thing for the engine

#

and for future VR work

granite jacinth
#

?

#

Not saying it's not

tired tree
#

having an abstract platform

granite jacinth
#

Just saying it wasn't for us

spiral zephyr
#

unification of arkit/core going well too

tired tree
#

wanting them to throw features on top of a system where every HMD is implented differently is terrible

#

they are working towards correcting the baseline, its the proper approach

granite jacinth
#

Oh, I don't disagree at all

tired tree
#

also these release are significantly faster than other engines, not sure what you expect them to achieve in a couple of months between versions

granite jacinth
#

?

#

No it's not

#

Not even close

spiral zephyr
#

i barely have time to read the changelist between versions, its so fast :p

wicked oak
#

they are preparing for openXR

granite jacinth
#

Aye, I stated that earlier

wicked oak
#

making sure the VR subsystems are completely plug and play with every platform

#

unlike unity, wich cant multiplatform vr for shit

granite jacinth
#

Which is good

wicked oak
#

unreal vr is getting more and more unified

tired tree
#

what engine is iterating faster?

#

def not unity

granite jacinth
#

Yeah, because it had to

wicked oak
#

UE4 not even a contest

tired tree
#

godot maybe but they are half done on everything

granite jacinth
#

Unity was rekting them in VR/AR space

#

UE4 is playing catchup

wicked oak
#

well, true, godot has blazing fast in dev time

#

but thats becouse its faster to do code on a small engine

granite jacinth
#

Well, it's getting bigger and biger

tired tree
#

unity was already set up for the renderer

granite jacinth
#

Tons of UE4 devs going over to it lately too

wicked oak
#

its definitely looking very serious for a vr engine, given its very flexible render pipeline and forward focused renderer

tired tree
#

90% of what unity was doing was community supported

wicked oak
#

unity didnt do shit

spiral zephyr
#

i thought tons of unity devs were coming to UE4 lately

granite jacinth
#

I doubt I'll jump onto the godot bandwagon

wicked oak
#

and thats actually real. Unity Oculus support was due to oculus making plugins

granite jacinth
#

But I can see why people are

wicked oak
#

same thing with steamvr

tired tree
#

valve directly makes the plugin for unity...

#

hell, ue4 is going to beat them to the dynamic res too

granite jacinth
#

@spiral zephyr TBH, that's a myth

tired tree
#

because the steam branch is out of date for it

wicked oak
#

meanwhile unreal was a pure deferred engine

granite jacinth
#

Most Unity devs, stick with Unity

wicked oak
#

wich is a bit overkill for vr and no antialias

granite jacinth
#

It's rare that Unity-> UE4

#

If anything, it's the opposite

wicked oak
#

they stick with unity coust its what they know

#

all i know is one thing

#

ive seen local game schools changing engine

#

from unity into unreal

#

and the results

granite jacinth
#

Right

wicked oak
#

are spectacular

#

night and day difference

tired tree
#

early into VR unity was the obvious choice, jsut due to the renderer

granite jacinth
#

Except my school focused on UE4

#

we have classes in Unity though

tired tree
#

thats been upended

granite jacinth
#

So I was able to do handson with both luckily

spiral zephyr
#

yea, companies with devs AND artists are tempted by UE, seeing how much better artist/progrtammer cooperation is with BPs

granite jacinth
#

hmm, I don't know.

#

BP is such a trap

tired tree
#

no its not

granite jacinth
#

Eveyone thinks it's super "artist-friendly" and easy and you don't need to know programming

#

But..that's not the reality

#

It is a trap

tired tree
#

set it up correctly for artists and they don't

granite jacinth
#

I am stating if they don't have an actual programmer

#

To build the game systems pipeline

tired tree
#

thats obvious, you can't make a game without logic

#

but, go do archvise

#

or scene layouts

granite jacinth
#

ArchVis is surprisingly a fuckton of money though

#

I've gotten a few ArchViz contracts and I am amazed at what they paid me

#

which means they are making 5x as much as me at least

tired tree
#

BP is totally artist friendly, you don't expect your artist to make an AI system, and even if you do for some reason, its far easier in blueprints for them to learn

sly elk
#

Unity is easier to get an artist who has no game engine experience up and running (just bringing in assets). My experience with unity is that when you start pushing towards building something complicated you find features that aren't implemented correctly or just aren't implemented, then you need to go to third party tools and it becomes a maintenance nightmare.

tired tree
#

unity def has plug and play advantage

spiral zephyr
#

Im a visual learner/worker, and I can do complete wonders in BPs. I've also learned a lot about algorithm complexity and data structures, and I can read articles on almost any language, and use the lessons in blueprints. I doubt i'd get this far without BPs. I can even read cpp now. And text code reallly scared me before! Thanks blueprints.

tired tree
#

but thats a result of its core logic running c#

#

and its a lot harder to modify for artists than a BP package

sly elk
#

Also thank to you guys pointing out that its crazy I wasn't running version control. I now have a perforce server

tired tree
#

really free's you up being able to screw up totally and just revert eh?

sly elk
#

Yeah. i have had VC set up on every other game I have ever worked on but its nice peace of mind. Also VC is way easier to deal with when nobody else is checking shit out

wicked oak
#

@sly elk thats why i use VC religiously

#

even on solo projects

#

its so.. freeing

#

to have all the backups in case you fuck up

#

and yeah, unity is a serious maintenance nightmare

#

unreal with blueprints is bad, but unity everything is worse