#virtual-reality
1 messages ยท Page 37 of 1
it's why I'm taking perception and neuroscience classes as a graduate computer graphics student
Nice.
I took cognitive science in uni. Never really helped my career directly...
until now!
But even so, I've had to keep my own knowledge up to date - the stuff I learnt in uni only got me started.
Anyway, take it easy. cya later
oh I've seen Walkabout
that's what I thought you were talking about
or rather how I thought your thing worked
Yeah, I can do walk about with my system too.
But the difference is I've slaved the turning to my hand
which makes it much more flexible
I see
The analogy is that you can move a big curtain around before your face without inducing vection.
Ciao
add me as a friend
Done
So... how many of you have set chaperone to developer mode?
i.e. no chaperone, just floor line
tried it
forgot to turn it back on and let my friend use it for a day
he had a harrowing experience
I turned off the floor line too
this is just after you and I tried making our own cages ๐
i have chaperone gone completely
well at work anyway. Normally just set it to standing mode only and you just have the circle but then go into the config and set the alpha to zero on it
I do roomscale so it's pretty necessary
even with the cage it's easy to go out of bounds
yeah depends what you are building
when you are working on 1 for ages you kinda start to learn when you are going to nut yourself on the corner of the desk hah
also im doing event based things so we build physical props to match the virtual space
Ah cool
@ the event base thing with physical props
I've still got my custom chaperone bounds... and got them working beautifully.
Found that they actually, counter-intuitively, enhance presence.
Because when they turn up, and you've got a virtual object inside the bounds
it actually feel like that thing is in the room with you.
My custom bounds are occludable.
Unlike SteamVR's normal chaperone, which ends up screwing with your depth perception
Funnily enough, works ok with the actual chaperone bounds too. It's like a double boundary
still, I wish it worked in packaged builds
I made one that appears faintly when you get near then turns FUCK OFF RED when you get too close
yeah that was the main problem showing the circle became really annoying so was handy to turn it right off. definitely some better ways to do the chaperone with dynamic 3D objects rather than the steam bounds
I made a failover with the chaperone so it just defaults to 100 x 100 and lets you set the actual scale in menu
So now I can have my custom chaperone when packaged!
Albeit with an extra bit of work.
A lot of the benefits I implemented are still functioning even when the boundary scale is wrong
Hopefully epic sees it and realizes they need to fix that shit.
The failure is in the SteamVRChaperone component
Failing to init properly when packaged
Yeah that's quite a problem, there should be an option to include in some way.
Question: How do you handle Output (like Print Screen) in VR Preview? The Stuff is always out of FOV and i just can find it in the Output Log. Another stupid thing is, that the Editor minimize if you start VR Preview... Using still 4.13.1
@digital vault click on filter and type in Print String (?)
Otherwise have it spam per tick
and you'll see it come down as it ticks.
Otherwise if it's a trace, turn on the debug
Ye the tick isn't an good option.... I don't like the "flow". Turn VR Preview On, Switch back to Editor, find Output Log, VR Action,... Maybe i find an option to move the print string (text render isn't an real solution for me).
Well, you're eliminating all the viable work arounds!
So, if you do that, you're gonna be stuck!
๐
hm...
What's wrong with text render?
Maybe try reporting it to a widget that you put in front of your face?
Ye that could be an option... I often use different Prints in different colors to see a specific flow... this works fine (and fast) on 2D... but sadly not in VR.... Oh maybe i could get a second camera? As 2D camera? i'll play around
Text render is slow in VR?
What's the vertical size of the active volume for Constellation ?
It's a 90 degree FOV camera
and it gets jittery about 5 meters out
4 to 5 meters
Is that what you meant?
Like, if I raise my hands all that way up, will tracking be lost?
That depends entirely on how you've positioned your camera
Uh... If you go and download one of the sexy Rift demos... Vieviev, there's a function in there that lets you see the cone of tracking.
If it has 90 degree fov and only 4 meter out, I don't think it's possible to cover space from the floor all the way up to where hands end (when raised) :(
I don't have Rift yet. Just something I wanted to learn ahead of time
Now with Vive support :)
yep
Too bad they haven't expanded their lineup
When do you folks think MS will reveal specs for its VR hardware?
noot sure. bout the same time real ppl get their hands on it
So it's not even ready yet?
I see.. So it could be just a PR move and actual HMD might be crappy
...WOW asynchronous reprojection in SteamVR
how does it compare against Oculus?
i dont have time to test, im doing the build for tomorrow
it's basically the same
it seems to be exactly the same algorithm
it's obviously not ASW
but that's a different tech
I could complain about how late it was as a feature but it's just so nice to finally have it for those little times when your computer poops out
becouse my game hovers around 95 fps, and steamVR just put it back to 45
why 95?
unlocked fps
roight
its more or less where it is
custom engine features? why did you unlock it?
with a few enemies around and FX, it can go to around 70-80
default
but 70-80 for those times is fine in oculus
it just timewarps it
little change from 90
people dont notice
people start to notice lagging when it goes to sub 50
steamVR just put you at 45
where your controllers have noticeable judder
yeah, but I thought the SteamVR plugin handled negotiations of starting rendering of next frame
in a few days ill plug my Vive back and test
does it really just run out of phase?!
right now im fixing some late bugs for show, and testing Nvidia 4.13 build
because I've noticed the judder of controllers in UE4
Yeah my controllers judder too
becouse that nvidia 4.13 build has:
multires
single pass stereo
lens matched shading
VR SLI
the SLI part is useless on me, but the rest of things are very nice
i want to at least show a really technically solid game at the expo
check this
top one is the new interface
bottom was the old one
noice
Played battle dome again after someone mentioned it last night.
much better UI
big buttons made for blind people are good on VR
ahaha
It's umm... different. Got bots now.
are made with blender and rendered
yeah, forward MSAA helps so much, can't wait for stable 14.0
becouse i dont know how to do graphics properly in 2d
But still much the same. Their locomotion options all suck though ๐
hell yes, 4.14 forward MSAA is glorious
โ
i made them on blender, rendered them with Cycles
and then imported the .png to the engine
to use a sprite
ohhhh nice
Have there been any improvements in 4.13 or 4.14 for LOD distances being more consistant between VR and non-VR?
that's a great way to generate shapes for sprites
not the first time i do things like that, the Achivements of Deathwave are like that
all of them are the same blender file, just switching the mesh for the icon
nice, efficient
becouse i cant draw lol
you can really automate that
thats the whole idea
the blender file has the scene set up
thats why you can see the reflections be exactly the same in each image
if you can't draw, an alternative to making models is using a vector graphics program like Inkscape
also easily parameterizable
definitely should get that
I adore Inkscape
it's also OS
hehehe
I know that feeling
knowing a toolset so well that you use it for everything
soon I'll be writing my dissertation in Maya
๐
Hey man. I bet I could model architectural way faster than you ๐
mos def
Mainly because I have an interior design background.
at least it's not Rhino, and it actually supports some OK vertex generation
I've worked with so many architects that insist on using Rhino for VR model generation
Haha. D:
eventually I just stonewalled them into learning vector modeling with Maya
Vector modelling?
like, NO your fucking toaster SHOULD NOT REDUCE ME TO 20FPS
Oh, Rhino is all nurbs?
now that its for indies
yes
heh
(looks at the nearly useless blender nurbs)
and "easier"
ackackack
Because he bought it before he hired me.
u know what, i love that the ue4 nvidia branch has all this fancy stuff
Like... they don't do actual product manufacturing work, just design concepting that then gets farmed out overseas to manufacture.
yet no documentation file that says what each thing does
I wish they would just include it in the damn engine already
at least a commandline list or something
some kind of card-agnostic API wrapper
I don't want to do NV-only for my game
don't want to be that guy
Seriously Nvidia.
even though I love NV and I worked there, and I bleed green like all the others
If you guys think Rhino and sketchup produces bad meshes when polygonized... I dare you to get a 85k object large step file ready for Real time ๐
my demo is on a nvidia laptop
It just lacks critical features like... UV mapping.
We once optimized an entire jet engine... All parts
so im alright doing nvidia specific optimization for the show
its just that amd has nothing like that
nvidia has their own fancy tech branch
and amd doesnt
they just have that SDK that no one uses
dice is super friends with amd
but amd could try to do like nvidia, a vr branch of ue4
Well someone needs to be.
If you're doing anything but simple objects, you really cant get around doing proper vertex modeling, smoothing groups, uvs and triangulation ๐
with both branches doing similar things, maybe Epic can go and do an abstraction
over both vr optimization sdks
right now, they cant do shit against nvidia for VR
see the trend, almost all the UE4 VR games on steam are implementing that nvidia fancybranch
Raw Data, and Pool Nation both use that branch
Is it hard to implement?
no
Can you do it at the end?
?
Hans't AMD been out of the game for developers for a long time?
the only thing you have to do, its to download the VRWORks branch from github
compile it
and package your game with it
Ah... what about when UE4 updates?
4.14 has the forward rendering though
Still compatible or have to wait for update?
Surely that's a better option than the Nvidia branch?
depends
Depends on what you've done with your game
Isn't foward more performant?
in my game, Forward cant be done yet, too many drawcalls
depends
if your game has several dynamic lights, and lots of drawcalls
nope
if your game is using almost everything on baked light, and not that many drawcalls
forward will be faster
Meshes and materials if I'm not mistaken
So every unique mesh and material is a draw call?
as in, meshes with multiple materials are also multiple draw calls
As far as my basic understanding goes, yes
but I'm sure somebody will come along and prove me wrong ๐
meshes, materials, and a few extra stuffs
What if I reuse materials across multiple different meshes?
nope, still several drawcalls
Per mesh?
yeah, still a new mesh
each static mesh is 1 drawcall per material
even if its the same mesh, but repeated
if you want to improve that, you gotta merge meshes
Instanced meshes as well?
or go HLOD
instanced meshes is 1 drawcall per material
the number of meshes is irrelevant
100 or 500
1 drawcalls
Mmm
that is the whole point of instanced meshes
Nice
and no, you cant put every mesh of the map in instanced
believe me ive done it FOR SCIENCE
Haha
What if you have a bunch of cubes and just voxelize a game? Except all the cubes are the same instance of the cube?
well, you would still get to use different materials/colors
Less a trick question and more a - I wonder if this will apply to the game I've been thinking of.
Fair call.
But that would still not support using different materials would it?
Parameterize them?
the texture is a map with all the textures
like minecraft
so then you just use the UV
Ah
Nice
ohh, so that wouldnt cost you another draw call?
yeah
but Zaptruder.... that would require you do some UV work ๐
yeah, and custom geometry
Well, I've been learning Maya
i did it for a shitty custom engine
making a minecraft clone in opengl is a GREAT learning excersise
just opengl plus SDL for windowing
not even physics engine
But a couple months of Maya experience doesn't beat a few years of Sketchup experience D:
if you're going into the rabidhole that is game art, a real polygon modeller is neccesarry
No doubt.
I havent done any high end game art in years, and there are so many problems/issues you need to know when doing your modelling
such as how smoothing groups behave when baking normals etc.
it's a goddamn circus I tell you
Yeah, I'm just going to stay away from doing high end game art.
Use my design skills on strong art fundamentals!
have fun
i dont know how to enable the extra stuff yet
but the multires is hella good
+30% performance for free
pixel shader performance
so you can use it to keep the same perf, but increase the resolution
hmm... I have never done any of this compiling, is there an tutorial on how to get started?
Hmm... associating my epic user with my github... Cant seem to find the place where I enter my github account
nwm.... found it
Gonna try and do a build and see how my project runs in 4.13 VRworks
Having some troubles hitting smooth 90 fps on a GTX1070
remember, you gotta enable the fancy shit
to enable multires
is vr.MultiResRendering 2
there is also a 3, but that one is too aggresive
the 2 is basically invisible
are you pixel bound or geometry bound?
if you do stat rhi, what are your polygons and drawcalls?
So what extra shit is there to add?
Also, is the new shit nvidia card exclusive? :/
Arghhhhh
That's unfortunate ๐ฆ
thats the stat rhi on my own pc, no VR and a GTX970
yes
Not possible im afraid. We're doing some industrial demo for a client using their crazy ass step files
im at 1.5 million polys, but 1300 drawcalls
you have more polygons that pixesl
you can lod that effectively
use instalod or simplygon
Optimizing the model is the next step if the client likes what they're seeing ๐
wich is automatic
So what are people doing to get around the fact it's Nvidia card exclusive?
nothing
Trust me on this one, the mesh of the models are so bad that if you try to do any automatic optimization, it will explode
becouse 70-80% of the market is nvidia
alright
this stuff is nvidia only
but it doesnt harm amd
it just doesnt work
if you do a multires command on amd
it does nothing
Oh alright, it's no so bad then
Ohh well, we get 90 fps using the same project with the forward renderer in 4.14, which should be out by the time this project starts up for real ๐
So what are the secret commands we can use? :/
yeah, forward is very good with low drawcalls
not for me ๐ฆ
1300 drawcalls too extreme
luckily our projects are so small scale that I dont have to do so much worrying about doing crazy optimizations
but then i have maps like this
that is a absolute shit ton of meshes
becouse that
is infinity blade assets
ohh, I was about to say you could do some merging inside a DCCP app and save draw calls
im on it
im merging inside unreal at the moment
but its still slow
there is a crapload of meshes there
im talking about 7 thousand
well, thats after optimization
it was around 10 thousand
before some basic optimization, a general vista could go to 5 million polygons and 3k drawcalls
and no, i cant use hierarchical lod
it crashes
Why not use Simplygon's mesh combination feature or the feature that's built into the engine?
too much for it
im already using that
but manually
getting groups of meshes and merging them
that improved it
ill have it on the showfloor
but it has a WIP sticker on it
and a warning
becouse it runs at around 60-70 fps
good that i have spacewarp
i dont do art well, but i do tech well
at least i can show a game rock solid there
lots of indies doing VR games tend to forget about good tech and performance
and end up lagging
Hmm, my impression was the other way around. Its usually the tech guys doing the demos but forgetting the showmanship part
when a sony executive of something tech related tried it, he said it was the best tech of all the Vr games in barcelona
from all the indie demos
he was quite happy with how it worked
Palmer would approve !
but its a simple game with non consistent art
becouse its made from ifinity blade meshes and other packs
Marketplace art!
You know what UE needs?
It needs design specs for marketplace art.
assets
tends to happen when i got no artist in the team
I guess I'm just damaged from my line of work, where the client cant tell you anything unless you're doing final renders..
Make stuff for marketplace, and it should conform to technical specs.
in the general artist sense, concept and stuff
isnt there some of that already?
marketplace stuff has to be aproved
Ah yeah
It's not really easy setting up technical specs for 3d assets
Makes sense. Wouldn't want crappy assets that end up making their way into consumer games
associated with engine.
yeah, definitely not easy
into many consumer games
its not the same for a first person game than a 3rd person one
because the requirements could be very different depending on how you're gonna use the asset
at least modular kits tend to be consistent
True that.
VR actually sets up even bigger requirements for objects
But maybe then it should have something in the asset store that says - optimized for;
optimized for first person, small scale gameplay.
because they have to be optimized, but also have to hold up to close scrutiny
yup
gotta LOD that
becouse VR needs low drawcalls
and anyway, you cant see shit from afar
hierarchical LOd
you could texture atlas stuff
the "Top Down Scifi " pack is one of the best value packs tere are
What does hierarchical LOD mean?
its not too high res, but its very low poly, looks good, and its all in 1 material
its a engine feature
it does hierarchical lods
XD
๐
basically, it makes it so from far enough
it merges the meshes
imagine a whole building
Ah
amde from modular pieces
Does it do this on the fly?
from close, the wall will be all the individual pieces
but if you get far, it will eventually make the whole building 1 mesh
its a engine feature
works with sympligon best, but also with instalod
its slow
very, VERY slow
Oh, so it's like baking light.
slower
Gotta preprocess it
yeah, it does preprocess
it generates mesh groups
and then it generates more mesh grous that include the ones before
in kind of a tree
Yeah, I can see that.
when you are far enough, it stops rendering the individual meshes/chunk, and only renders the big chunk wich is everything together
it also uses absolutely massive amoutn of memory
in one of my maps
its 3 gb
Bloody hell. It's like the dark arts of modern game development are opening to me ๐
Dayum
Well, makes sense really...
becouse it has to bake light for every LOD level
yeah
its merging the textures
into an atlas
for the group of meshes
Yep.
obviously, it works FAR better if you already have them atlassed
All so it can cut down draw calls.
its the whole point of it
I thought DX12 was supposed to be draw call magic
Of course this is only what I heard back before I got into game dev
yes
it is
the whole point of DX12/vulkan is to make the driver much lighter
if well used, it will make drawcalls far cheaper
Give us all the good stuff, for free and no work! ๐
I'm happiest when the 'work' required is clicking check boxes! ๐
Like multi-res
Want multi-res yes/no? Yes.
well, it does require you to build the source your self ๐
true that. I am by no means a developer, but lots of tools that I come by is like "Yeah... Just compile this obscure thingy and then import it and you're good to go"
Shit yeah. The core of my work is built on top of someone else's plugin.
Learning Python had a lot of those moments... Ohhh you want PySide or PyQT? Here are some crazy obscure guides that only old time developers know how to do
Well... I mean, my work adds super value, but it wouldn't function without that other work done.
oh yeah @dry fjord and others that were interested last night, here's a video of my zero-g movement feature in Iron Rain, bear with the development art and other unfinished features (really the movement scheme isn't done yet, but this is really fun to glide around in) https://www.youtube.com/watch?v=Rul5eB5yP6g
Yeah, I think we were talking last night about Battledome
yup
I heard of that in the Vive reddit, haven't tried it. fun?
And no shits given to motion sickness caused.
@winged shale thats pretty cool! Was it hard to do?
yes.
besides moving the fingers and doing the logic to figure out a grab, the movement was absurdly complicated
because you can collide with the environment
He talked to anti-damage ๐
actually this is before antidamage's fixes
Oh nice.
BP? C++?
does it use extra software?
nope
I thought you said that was jittering
it's my own
Looks ok to me?
no judder happens with controller models at 45Hz
I don't a. hit 45Hz ever anymore after fixing the performance
b. show controller models when hands in range
also the IK is super broken
it's a combo
You programmers be crazy
@winged shale Hey, that's pretty good
I've got this absurd AnimBP that goes from Ref-pose to actual
The IK stuff for sure.
i would buy it
Actually, I want the Unity IK solution. Looks bloody amazing.
You could always go unity ๐
Nah, it's a marketplace asset.
You even get the Lab forward renderer
it does IK on hands, then it sets the rotations of each finger bone based on a crazy 1000-line logic set for figuring out how grabbing works in the character
So that's why you want to do what @dry fjord taught me
where you do the hands seperately from the character ๐
the really complicated part is the movement by grab-draggin on the world
Final IK is the Unity one
I've got a lot of optimization to do on the anim BP
but how exactly you do the IK on the hands?
ive been doing it myself and having problems
also with the head
Using the Allright Rig
oh
Yeah the way @dry fjord and I came up with
was seperate hand mesh and two bone IK to the wrist to the shoulder
how are you setting the torso position?
@wicked oak I do two-bone, then I rotate the hand, and I have attached to each controller model a component that represents the offset for the hand, and I pass its transform into the animBP
oh that is so complicated
first I have to figure out which direction to place it
i got the 2 bone ik and rotation for head and hands, but thats not enough of course
because you can tip your head back further than 90 degrees
With the allright rig, it's just dragging the torso controller ๐
Did I mention it's free? ๐
so you need to dynamically interpolate between forward- and top or down facing vectors of the HMD
I'll be writing a dev blog for this week (first post!) about all of this crap
I'll link when I publish it and start the blog
But I just used the right vector on the HMD and rotated it back
Works fine for placing my feet ๐
lol I don't care about marketplace things, it's more important to me that I get my game out there, I'll just share my methods for the interested in my blog
Side question; so I've built a thing on top of someone's free plugin.
I can't release that on marketplace can I?
@clever sky when doing that, there's the problem of if the user rolls their head
@winged shale Yeah, I can imagine. I'm just relying on the user not rolling their heads.
I've built a state machine that figures out the proper orientation of the torso based on where the hands are and where the head is looking at arbitrary transforms
At least not more than 90 degrees.
there's a few problems in that video, one is that for some reason I left the elbow targets in world space, so they're always pointing at the ship, and another is I left another vector in world-space that I hadn't noticed before I made the video
btw, that VRGrip plugin that was being thrown about earlier, what happened to it?
so arms will oddly interpolate in strange ways
next video will fix all of these probably
VRGrip?
What's it do?
Yeah, there were some plugins for the unofficial VR template(before Epic released their own)
to handle all the grip stuff
Arhh, thats why I had trouble finding something about it.
I think? It definetly has a grip component in there.
Im just using BP interfaces and a function library to quickly add it to different actors
but it does have some issues, doesn't handle multi grapping very well right now
yeah how do you interp the object for 2 hands
Yeah, if you have an item grapped, and want to change hands
I'll show a video soon of my two-handed weapons and two-handed object interactions, I think it's pretty good ๐
Ah yeah. Did a bunch of work on that to make mine work properly
it attach/deattach is a bit... funky... So I wanted to see if someone has already been doing all the work for me ๐
They did the bare minimum in the template
The second you start adding extra functionality it just shits the bed ๐
C++ ftw
not gonna happen @winged shale ๐
But it works now. I've got rumble from the grabbed object.
also all these youngsters don't properly know how to architect a program these days for extensibility
And it doesn't transfer that rumble over
which was a thing that was happening to me for a while ๐
Hey! I learnt some C++!
@clever sky++
I already have my plate full with so many things... Learning C++ is way to big a moutfull right now
someone should make an upbot plugin for Discord
and yeah if you don't know it going in, it's going to be a long-ass road
Im doing VR dev demoes, I'm doing video production, I'm doing 3d productions, I'm doing internal tool development
so much stuff going on... BP's are a goddamn godsend because they are so quick
I know dat feel
I prototype nearly every new object interaction in BP, then transfer it to C++
Best part is that you're not switching back and forth between two programs and that BP 'intellisense' works perfectly.
it's really easy because BP functions mostly always correspond to a C++ function
yes lol
^^
Yeah... well their objective is to make BP and C++ interchangeable right?
well.. the biggest plus is that I dont need to know C++ to do the stuff I need done ๐
Having done some Unity dev a couple of years ago, it was nice that I could do small parts in JS(which I know from web dev), but it quickly changed that you neeeded to know C# in order to do anything really
Honestly, BP is the main reason I'm with UE4.
I dont feel the same limitation from BP's
So to your suggestion to try Unity, can't do!
I'm with UE4 because I don't have to pay monthly for it
and it's superior in many ways
Yeah, like BP ๐
and C++
hell.. BP's are the reason some of my colleagues are finally starting doing logic work them selves. Before that they wouldn't touch anything that required progamming/logic
That means I can focus on the large systems and "outsource" small projects to them
So... real question time - what's the overhead of doing everything in BP vs C++?
~10x
it depends
haha ๐
I read that you don't want to do basic math operations in BP?
nooooo
the more micro you go, the more slow
but if you have huge functions like sphere traces or whatever, it's nearly the same runtime as in C++
Jeez. Gonna have to rewrite my whole movement system. Wonder if I can just pay someone to do that.
@clever sky there is this guy called @winged shale I heard he is doing some crazy stuff
probskies, lots of devs that know BP and C++ well would appreciate some extry moolah
Probably not a bad idea really... I can learn from seeing my BP code reinterpreted into C++.
I'll give you advice if you get stuck, but I'm currently a full-time grad student, doing extra full time research, and also doing full time game dev in my spare full time, while being a full-time husband
So your days are 4x24 hours so you can do everything fulltime? ๐
So like Zpanzer you too wear the hats.
That's not a fez in his avatar. It's hats on hats.
The biggest difference is the stuff sphychiatrist is doing probably will earn him more skittles than what I'm doing ๐
yeah the asymptotic shape of hats as hats -> infinity is a fez
Phil Fish?
so much respect, Fez is beautiful
Yeah he does look a bit like that.
Well, to be fair, game dev is an industry that will drive most up the wall.
Or so it seems.
yeah
I'm only here so I can drop my VR ideas
I'm guessing its because indies are usually passionate people with little to no business/project management skills
the best moment for me in my game dev history was realizing that, while I'm passionate about game dev, passion is useless, and discipline is what makes your game when the passion follows its randomized sine wave pattern
which is why I feel comfortable going social-media and public with Iron Rain, because I don't feellike I can work on it regularly anymore, I know that I will.
It's just I procrastinate like hell when I have non-passionate work to do D:
life changing
Haha.
One I thing I have learned working on larget 3d projects, is that you so easily get bugged down by details because you're passionate about what youre doing. You want everything to be just right. You lose your perspective and sight of the goal
REALLY worth a read if you want to procrastinate about something
and you end up spending 4-5x the time really needed
Yeah... I get that. Gotta get all the details! I want to be a god and make a virtual world!
So... we need these marketing shots...
and done.
@tawdry dragon from a meta-standpoint, I realized it's super useful to do public reviews for my stuff, because even if nobody watches your shit, you are gonna watch it over and over and notice all the stuff you missed from being perfectionistic about stuff
like I just added several high-importance TODOs to my google keep list because I watched that video I showed over and over again last night
My process is to do a kanban diagram with all the tasks ahead of me, write a short description of what the task is actually supposed to do and how important it is for the project
and then I start doing it one after one, always checking if I hit the "its good enough" marker
But also, I'm going silent for a bit. gonna try out AliceVR
what I do is with Google Keep (I keep it simple because I can easily get bogged down by formatting a google doc :P) I keep a IRON RAIN //TODO, IRON RAIN //COMPLETED, IRON RAIN //IN PROGRESS
in TODO I have a header that says, "I just completed ____, now it's time to start working on ____", followed by a list of things I need to do, starting with the second blank
I know this is not for everyone, but I suggest you give https://trello.com/ a look
I used trello for months, it didn't work for me too well :/
too much reorganizing and labeling, I prefer this, because...
every night after I'm done working for the day, I update the stuff by putting TODO things in COMPLETED, update IN PROGRESS with stuff that is still being worked on, but MOST IMPORTANTLY I reorder the todos in TODO that night and that night only
I don't allow myself to reorder them during the day
that way, what smart me last night said I should do next is what I will do
it's important because I often am much more likely to schedule a big shitty task like "rewrite everything that has to do with this so your future is better" when I don't have to do it immediately
so I kind of trap myself
The hard part of project mangement is that we're all different. Its so hard to give advice because you don't know if it works for everyone else. The most important part, I think, is to try something and see what works and then adapt it to your own style
definitely agre
e
I like to put it out there because it might solve someone else's problems
I have the problem that I really go down the rabbit hole with doing really fun tasks, like "make a new weapon!" or "do the spaceship thruster balancing code"
and then the core mechanics and architecture of my code craps itself
Back. Yeah, don't buy AliceVR
lol
Refunded. I have zero tolerance for locomotion that sets the forward vector to the face.
Look people. I like to look around while walking around. Why you gotta make it so hard?
indeed
Plus it also sets movement to on touch.
As opposed to on click.
That surface is an ideal resting spot for my thumb.
like Onward
So why you making me move around when I just wanna rest my thumb?
ugh I can just feel the crappy code architecture when I play Onward
Haha... But it's getting so much praise!
Fuck... that's the shit (referring to the locomotion system) that the VR community is praising right now. Hopefully when I drop mine, minds will be blown.
but there's just so much that Dante for some reason chose not to do right
same feeling
Otherwise, I dunno man.
It's like people just want to be able to use gamepads in VR, everything else be damned! ๐
Or at least the classic control schema
I think my game is the space game that everyone has wanted since Freelancer
Wait, you're working on an open world elite/freelancer game?
I truly believe this, and I'm so excited to showcase it
yeah doo
it's a spaceship building game
Oh shibbssss
and you can salvage modules from other ships
brb relocating to my office
Well, I'm headed to bed. Cya later. But also, sounds exciting your thing
Wants to play something like that.
Just make sure the space ship design is tight.
Incidentally, my space ship and mecha designs are pretty decent if I must say so myself ๐
space ship design is modular ๐
every hull has attachment slots for other hulls
so arbitrary spaceship design is possible
also I'll open up workshop access after alpha
@winged shale how would you actually go about loading external assets? I guess that is only possible with C++
I would package an editor with the game, like Unreal Tournament does
so people can make mods
which is why I say after alpha
I want alpha to be all about making the game fun
it's because I've actually been looking into doing stuff like variables inside XML files that my coworkers can do without doing anything in unreal. It would be cool if I could make complete "uasset" outside the editor and have it load them in at runtime
that would be dope as a rope
but at that point my question is why don't you just use the editor, or teach them how
because that would mean we have to cook the project each time we do a change
ohhhh
it would be awesome if we could do it with on-drive xml/fbx/texture assets
maaaybe :p
that would certainly be interesting
maybe some kind of xml metadata about how to cook it by your game, and then make a lite material builder in your game
but that's a lot of C++ there
probably more time than it would save you
but certainly a crazy good learning experience
Yeah that would be my guess.. It also just thinking forward. Say we add new products to our VR experience. Updating computers could be done using XML files/zip files instead of whole builds
yup, I chose the solo path for a lot of reasons, including this situation
it's very hard to do syncing right
using perforce/git with people who don't know how to use perforce/git from their professional careers is gruelling
so you're left with using something that kind of drops assets in and out, like google drive or dropbox, and then everything is bad
Im using dropbox atm ^^'
but if it was a "real project" I would have to look at github
thing about git
is that it's reeeeeally good at storing different versions of text
but for blobs (actual term, binary large objects) like images, vertex lists (models) etc. it just screws up astronomically
It handles binaries excellently too
does it?
Yup
unless you use git-lfs plugin it's abysmal
How can it screw up when it just detects a change?
because git is written to detect and merge changes in carriage-return terminated text
when a binary object changes, it doesn't have lines to change or not change, so merging binary objects is not possible
I have not had that experience, every time my binary objects change, I must upload a copy of the new object
Sure
the point of git for me is that commits happen often and so do pushes
so, I don't want to wait for hours on my terrible internet to do it, but git with git-lfs works really well as it only sends diffs
So don't commit binaries
I don't!
Yay solved
and I use git for my code anyway
Hi everyone. How can I make a simple laser sight draw where my weapon is facing?
Draw where your weapon is facing. Done n
Google the components of your problem
It's very basic
anyway, when you work with git and multiple devs, you want people to have their own branches, and merge them together to kind of paste the project together in master as everyone works independently of each other to implement their things, as is the key to working productively with many people on one project
how do you reconcile this? and in what way is your use of git different from just using Dropbox if it had a changelog?
been testing oculus ASW
damn black magic
but it DOES artifact
its similar artifacts to things like smooth video project
subtle, but happens
yeah that's a case too, rotational motion causes nonlinear motion against the fulcrum and extreme ends
@winged shale the same way you reconcile it in P4, SVN, whatever. git is no different.
the problem is not version control. the problem is UE's lack of exposure of facets of blueprints to version control.
if blueprints were a text-based format with every node being a single line and every connector between nodes being a single line you'd be able to merge changes and review them
P4 solves it by locking. git solves it by allowing anyone to do anything and generating conflicts that you can resolve
literally nobody tries to "merge" binaries so it's a completely false argument against git
which is a shame because git is by far the best version control system when you get down to it
in all of these situations one person needs to go and talk to another person either before or after a change has been made
I prefer after, since you can see what both people intended to do because they went and did it and you can manually integrate changes and use that as the conflict resolution instead
but the problem will always be UE itself
LFS isn't a tool to get around any of that, it's just a handy filestore for binaries or other large files that might slow down the repo
That's interesting, I didn't realize blueprints are not generating human readable code behind the scenes...
they look like json or xml
you can copy and paste them
earlier blueprints compile to bytecode, new blueprints compile in the same way C++ does
there's no artificial code generation in between that you can tinker with
but ultimately that json/xml is stored in a binary format to prevent things from messing with it
and we all know how fucking shit it is merging changes to xml
be better to adopt a simpler non-nested format
every line contains a node ID and arguments or links to other nodes to populate those arguments and event execution links
and shit like positoning and comments
any VC can handle a single line changing easily
don't have to worry about the nesting becoming corrupt
and fuck it, make that format human readable
Agree wholeheartedly
But we are off the topic for VR. Anybody using the https://www.unrealengine.com/marketplace/gesture-tracker-vr ? I'd like to hear if it is all shipshape to use
A gesture tracking and recognition component designed to be used with VR motion controllers. Drop in your VR pawn blueprint and start casting spells with the wave of your hand!
@dry fjord the blueprints to c++ thing is still experimental and has to be enabled, its off by default
I know
nup
its some kind of massive state machine
like each case is a node and stuff
easier to develop that way I guess
it would be cool to have a tool to serialize blueprints to text
and be able to switch beetwen them
you can now
nah, im talking more about a true blueprints<->text interoperabilty. so they can be merged easily
its the main problem of blueprints, they are very hard to mantain
then you could have a visual change management utility
with a timeline to go forward and back through the changelist
there is already a perforce diff tool for blueprints in the engine
but i havent used it much
btw, how did you do that VR ik?
it's only for diffing not merging, isn't it? I haven't tried it in over a year
for fullbody ik in vr
with IK
oh
kinect
I'm working on adding IK constraints to it as I get time
since the kinect data can be shitty
I have tricks to get results that look like accurate data from just inferred positions though
very usable
@dusk vigil I haven't but I will this weekend, I've been waiting for a gesture tracker since I saw the unity one
took 30 mins to set up the first time
I'll let you know if you haven't purchased already
I can wait til next week no problem, we are just starting production in the last couple of weeks and are still casing things out. I'm basically planning to simultaneously do Perception Neuron mocap and run the gesture recognition mappings during the next month. Looks like a good fit for our game
@dry fjord I was originally talking to someone who is not a programmer about tools for sharing blobs across creative machines, and reconciling that. Thanks for the calmer answer, and I agree with you.
so Blueprints play nice with git nowadays?
how'd it go yday when you guys discussed about having a 3rd person view/camera in VR
that's fair, when you said third person view I imagined a third person shooter
and the VR camera is just taped to the character
swingin' around
Yeah that would get into trouble pretty fast
hehehe I'm doing a third person thing in Iron Rain where you switch to 'commander' view by pulling down a VR headset while in the cockpit of your ship
Nice
you put it on, and suddenly you see your ship and all ships in your radar range in a view like Homeworld
I love it when things are done properly in context like that
yeah I hope people like it as much as I think they will, I hope to have it so if you get sick of piloting the ship around you can go into this mode and just command your ship to do things, assuming you've got the relevant modules
and it should be useful for commanding other ships in your fleet
Making it an important part of the game - if it's just there for fun, there's not much point to have it
Like a submarine game where you have to use all the different areas, conn, periscope, engines etcetera, all are essential
can't be having fun in games
ever so spicy
man, whenever someone puts "VR" in the title of their game, it's got like a 90% chance of being bad
I actually haven't looked at any VR games for many months...
same
I wonder if that's a dev thing or vr game interest is fizzling out in general
or perhaps it's because most titles coming out are unpolished shite
I just don't want to see other games because I then don't see that maybe other games are better than mine ๐
so I just work on mine without caring about what others release, and then when I consider my game ready for release, then I might look at what others do...
that's a strategy, but maybe if you overcame that you'd be stronger as a dev
not that I'm one to be giving advice, what works for you works for you
well not sure if it works for me, I just hope so ๐
the thing I would worry about is missing out on great features that work well
and learning what doesn't
so I don't have to try it out
maybe, yeah
most of the stuff on steam is pretty garbage still
will be interesting to see the launch titles for the oculus touch, and a bunch of PSVR games coming out too that are properly polished
ok, I guess I should say its very nice to hear that most is garbage ๐
hah well its not great to hear from a consumer point of view