#ue4-general
1 messages Β· Page 172 of 1
Your reference image has lighting in the scene, try adding some to yours and then add a post-process volume and play around with the bloom settings. It should look similar
just adding bloom and upping the intensity has gotten me this so i think im on the right track, thank you https://i.gyazo.com/4c9f6e4e46646828fb54a7fb2a887a2e.png
Sure, np. The emissive parameter may be a bit intense for that level of bloom, try lowering it to find a balance in the glow. Good luck!
Will do! I set it really high to test if it would help, thanks again!
might wanna check out the spiral blur material node to add a bit of warmth to the bloom. It looks more like banding without it at high levels
@gleaming shore
So far messing with post processing ive acheived this: https://i.gyazo.com/0a4ab3919a2c68370bd61bbc01a2e4b7.png
Using the Timeline node is there a way to start it for exmaple on 1 second (like in level sequence)
The New Time input @gleaming shore
ANYway
Do you know if the Devs from Wildcard hang out here? I am having some Ark Dev Kit issues that I think are related to the dev kit itself.
Eh they might but the better place to ask about that stuff is through their devkit help support
Im building a modified version of 4.17 right now for the first time.
Up until now I've only used pre-built engine versions and created new projects via the launcher, so I prefer working in editor/BP rather than VS.
Anyone know if it's possible to create a BP project that uses a modified engine? I couldn't find much of any info about this.
sure
just start the Editor of ur Modified Engine and u can create projects like with a normal engine
@trim terrace Sweet. Thanks a lot for trying to help . Sorry, but do you know of a tutorial that walks one through the process? Like, Im not exactly sure where I would "start the Editor". Only way I know how to start the editor is by launcing a project from the launcher.
if u have compiled your source of the Engine, there is a Bin folder with 32/64 bit binarys. there u can start the Editor itself π
Awesome thanks!
I've been working with UE4 for 2.5 years. Never knew that was in there... Guess that makes me an old newb. Thanks a lot.
π No problem ^^
if u are using the Source Version from Epic, there should also be a file to register the sorce engine to ur system, so u can select the engine on ur Project (Switch unreal engine version) or something like that. then u can run the Project directly and it will launch the Editor of ur selectet engine
to do so, make a right click on ur Projects .uproject file
This is how it looks if u run the UE4Editor.exe
there u should know how to go π
@trim terrace Thank you! I'm going to copy what you said here and add it to me UE4 notes memo. Thanks for the screenshot. (I've seen it a thousand times now, so that one's burned in my memory)... Really appreciate the help!
the file you want is UnrealVersionSelector.exe located in Engine\Binaries\Win64
π
yeah. thats what run if u select "Switch UnrealEngine Version" of ur uproject file π
@spark garden Got it! Thanks a lot guys, this'll definitely come in handy.
A small Australian Company has found a way to give computer graphics unlimited power - no one believed them, but they are now using this technology to make r...
So if Euclidean isn't bullshitting, what is the tech they're using to render so much?
Voxels with some kind of realtime dynamic LOD system?
Hasn't anyone, like a journalist, been to their hologram thing and like, put their head to the ground or in a leaf to see if they were bullshitting?
Wouldn't that reveal if they were scammin or not
of course not, they don't actually let people use their stuff
they just keep releasing dumb videos and marketing materials to sucker in investors
it's getting more and more sad, because they're still comparing their graphics with games from 2007
and their 'floating point atoms' (aka voxels) are in mainstream use in a lot of game engines for more practical purposes
It just seems unbelievable with all the minds in computer science
that they are the only ones to have found this "Technique"
It sounds like voxels to me, but even then they don't really have a way to prove it in realtime over a youtube video since it's so easy to fabricate
it's a sparse voxel octree algorithm and something like marching cubes, I think
so it does render in real time
but it's very limited
My understanding was animation especially was super limited
they've never shown anything like animated characters, as an example
and would essentially be required for alembic purposes right?
I thought it was raytracing?
They showed some animation in that demo
Elephants running at the folks in the VR room thingy
But again, it looks like potato graphics
aha, skip to 10 minutes, looks like they're using VR for their 'holodeck'
and yeah, it's shite
it's a single object, without materials etc xD
Read the related dev-blog entry: http://www.atomontage.com/?id=dev_blog&pid=mar17_2015 This work-in-progress example shows high-resolution voxel geometry bei...
Speaking of voxels
Wow are you euclidean
ah yeah, they do have animation now (11 minutes into the video)
only took them a decade to get there
They probably use model swapping
there's no way to efficienctly move that many points of data in realtime
there just isn't
so you need to either unload and reload on the dime
or it's fake (probably this one)
not sure what game they're using to compare their animation, but it's some MMO or RPG that's at least 15 years old, if not 20
gotcha
If it was made of "Atoms" why didn't they pause the anim, zoom in, and prove it?
Atoms? Psh! With raytracing and analytical shapes we can have INFINITE PRECISION. (technique circa 1979)
Right up until you try to do a mandelbrot zoom and run out of bits.
Indeed π https://youtu.be/q6flyIrvKCA
Coming soon to Playstation 4. Claybook is a unique physics based puzzle game in a world made entirely of clay. Shape your character and the world around you ...
Atomontage has been sad for a long time
I think I've first saw and laughed at one of their trailers, maybe 6-7 years ago
they have some reasonable tech, especially when compared with 'Infinite Detail'
but it's still a long way from useable in any practical game development sense
Unless your ambition is to remake From Dust.
6 hours later... still going π
@bitter iris jesus how big is that terrain ?
@cloud cobalt Atomontage is not suppose to be practical, the dev is exploring limits of such approach and what overall can be done in rendering and authoring of assets. There are plenty of good things came out of it, like the way how he creates landscape. Which have actual different materials and objects inside of it.
@tame bluff if you follow trace of founders of euclidean, you will find couple of papers on fast algorithm for rendering point cloud data
@bitter iris what's that?
it's not extremely innovative or a game changer. Have no idea if its even practical for point cloud visualization when compared to other solutions. So they are trying to find investors somewhere else
Ah cool stuff
we all know that PHYSX doesn't support quaternion sweeps, what would you guys propose for yawing while standing in place and not going through objects when moving a custom pawn that wouldn't fit on a capsule? I have my own work arounds, but going the "real way" through binary search wouldn't suit my necessities and would be expensive as hell.
@vale osprey - it's just snake oil. They change the pitch every few years and try to extract some more money from someone else
I have no idea how they've managed to keep going so long, tbh
@cinder iron are you on something like a rocking boat in the sea?
they've barely progressed in over a decade
just a long hover bike, doing experiments currently
@tame bluff its a 64km^2 map
@cinder iron - what are the actual requirements, because chances are you can fudge what you need?
@cinder iron you could approximate with linear sweeps, or do an overlap test on a separate custom mesh
everything is solved aside the yawing in place, overlapping doesnt sweep when it comes from a full rotation movement, no problem when there is forward or sideways movement.
What I'm doing right now is not allowing the player to yaw while idling, but, let's say I don't want that
or if you are fine with a bit of low level code, check GetDepenetration() function in PhysX - its basically an overlap with collision data
thing is that hit doesnΒ‘t get registered without any z, x and y movement, not including rotations
yes
will GetDepenetration() even register a thing?
it doesn't "register", you give it two objects, it tells you direction and amount of force to apply to depenetrate them if they overlap
if they overlap
but they aren't in main terms, at least if I don't include the binary search for sweeps which is (at this point) kinda meh for me to do
if they don't they don't, you can use it incrementally to find range of yaw, or just use overlap test
I see, well I don't know how would you do it without binary search
make you bike a physics body?
line tracing is still will be binary search to approximate angular movement
you could just "ragdoll" it when bike is not moving
yeah but even though it is supposed to detect when player can yaw or not if bike is close to an object
mh, ragdoll seems interesting, but currently because of design choices the bike does not update any phys when its not moving
performance decissions
and network.
maybe sim it with physx intermediate - aka put bike into intermediate scene, with some objects that are around and apply torque in one direction and then another to test for collision
it all sounds convoluted π
hahaha, no joke I though about the torque solution to force a positive situation xD
It all came to my mind because I saw hovered vehicles in the new battlefront not having these problems, and that lead me to wonder if they are going full low-level or, maybe they don't use PHYSX at all which could be an option
I don't think DICE ever used physics middleware for their vehicles, only as query/collision system
but that doesn't mean much, like with tanks, I don't use PhysX for anything but raytraces /collision with rigid body simulation
yeah no I do the same, I don't actually handle hovering with forces and physics, I just fake it
thing is that I do use physx for collisions
for queries I suppose, as if you don't simulate
I don't know why there is no angular sweep, but it's like that in PhysX itself, figure π
yes, actual collision queries, I don't simulate
could just google what people who use PhysX do in this case
I did before coming here but didn't find quiet reliable answers π€
basically only "make things fit on a capsule"
yeah I don't see much either
PHYSX supports speculative angular CCD but apparently is not exposed to the engine on a friendly way, I might investigate a bit more, if not I'll go ahead with my silly trace solution hehehe
CCD works only for simulating objects as far as I know
maybe CCDSweep will act correctly enabling it, if i don't test I can't know
well thanks for your input though! nice talk
Hello
@cinder iron CCD Sweep? I think that is something low level that PhysX does, not sure if that is even an actual function
I phrased myself a bit akward there, It's not a function no @vale osprey
There, I edited it
ahhh, I don't have high hopes for it working as it's based on velocity. Good for projectiles in theory, catching something that collides as result of angular movement - don't know. But that should be solved automagically if you simulate. Regardless of ccd
ππΌ
Isn't that an incredibly old piece of software that still uses .dds textures?
@keen birch unreal engine uses dds textures
dds textures is a texture format meant for direct gpu usage
Well yeah except iirc this only uses dds
to "load" a dds, you literally memcopy it to GPU
pngs need to be uncompressed, then compressed, and then uploaded to gpu (or upload uncompressed)
@fierce tulip Thanks for the link, looks like a useful tool.
Is it better to start with blueprints? Or just jump right into coding?
blueprints
Why's that?
Because it's basically programming without syntax
There's two parts to learning programming
(Imo)
Syntax and logic
Blueprint lets you do logic without needing to worry about syntax
Which can be really great
I started with C#, 90% of my initial errors were "Whoops, forgot a semicolon somewhere"
Except I didn't know how to read errors properly so I never knew what was going wrong
Still, I'd say the answer is "it depends" ;3
hmmm guess it makes sense. In my case I'm only just starting to relearn C++ (started it like 4 years ago then stopped cuz..life) so i figure i might as well do blueprints while i also learn C++ that way i at least understadn how to use UE by the time i got it down
wanted to make sure blueprints werent a waste of time π
LOL the answer is always "it depends"
They're not
they are not a waste of time
but their sane usage is limited imo
however you need to know them to know where to use what
E.g. what I'm doing now (procedural foliage spawning) isn't a very good idea to do in BP
But still
If you fight the engine a bit it's do-able
i assume actual coding would be "better" tho
turns out blueprint functions in unreal map directly to C++ functions
Yeah, they do
wich means that blueprint and C++ is the same thing, but on a different syntax
thats why you allways start with blueprints
Well, with exceptions
its easier to learn and prototype with them
later down the lane you try to add your own functions coded in C++
Basically, Blueprint is C#
BP is way harder to maintain and fix tho
i can tell you first hand, blueprints are a pain to mantain
In what sense?
C++ can be copypasted around, and you get the IDE, all that stuff
You can copypaste BP
you cant copypaste a function
so i SHOULD go through any and all blueprint related tutorials from epic?
you cant copypaste a function from one class to other
Well, you can create a new function and copy-paste the contents ;3
you need to remake the function, paste the contents, and rewire all variables (becouse they are missing)
I've personally never really had issues
Ah right
also, blueprint can fail silently
Yeah, true enough
in C++ if it compiles its fine
if something changes in the engine it will fail to compile
(when you upgrade versions)
blueprint can unplug things and the like
wich is a huuuuuuuuuge pain
a pure C++ project is super easy to upgrade to newer version
@noble solstice I'd say no, I'm guessing you know how variables and stuff work, so look at one or two things to see how it works and then just google if you don't get stuff
just fix some compiling errors and done
but blueprint can break
and break silently
@keen birch yeah variables, loops, genereal logic stuff i pretty much have down. Mostly need a refresher on how C++ specifically handles that. I also need to learn the UI tho and how blue pritns operate so i think ill at least go through a couple of the tutorials. even if its just the third person one and the loopty loop car one
go with blueprints
to learn ue4 itself
even serious C++ programmers shouldnt jump into UE4 C++ directly
^ It's pretty different from the actual thing
becouse now you need to learn the way UE4 architects stuff + the API + the C++ weird shit it adds
blueprints is just"the way ue4 architects stuff + API", and lots more docs about it
you deal with weird shit in C++ later down the line
I second that
also a typical experienced c++ programmer mistake is not to check BP at all, which is a bit stupid imo
I come from a C/C++ background, and at the very least, I find prototyping features in blueprint much quicker than in C++. After things start to come together, I'll refactor to C++ class and expose variables / events / methods.
This general workflow feels like the 'right' way to do it in UE4.
(At least for me :p)
for me it depends on the stuff I want to prototype
if it requires e.g. a state machine
then I won't try to do that in BP like ever again, tried it a few times, that never worked out
what I like is having dynamic universal stuff in C++ and build on that in BP - all i do in bp is basically to setup different default values and call some functions
Overall, my biggest problem with BP is that they're not version friendly. I want an engine-independent text format that I can diff and merge with my regular tools and check years later with a log history.
IMHO it's the strongest argument against Blueprints
Obviously, there is a lot of stuff that works well with Blueprints too
Blueprint should be able to be output into a text editable format for diffing
That's not nearly enough imho
What I'm considering for my next game is a data asset with an automatic synchronization to text, for example for quests
I use something like that for AI and dialogues π
I am not sure how it works yet tbh, system is ready but wasn't used much (yet)
im going to spam JSON
for everything
blueprints will hold the "master" copy of a class
and json edits the stats and stuff
makes sense
i really dont want what i have in DWVR
I'm definitely going to do something like JSON sources for quests + reimport tool
DWVR has a "Characters" Folder
with 10 blueprints or so
Base Drone, Fast Drone, Rocket Drone, Heavy Drone, Weak Drone, Sword Skeleton, Heavy Sword skeleton, Pistol Skeleton, Rifle Cyborg weak Rifle cyborg strong...
etc
actually more than 10
on my next project i would have a Drone blueprint, wich holds everything
and just json to change the stats for the fast, rocket, heavy, weak, etc ones
We went with data assets for that (Blueprint just do the 3D stuff)
We should definitely have used a json source + importer for these
data assets are still binary and a fucking pain
isn't it a pain to setup stuffs like meshes?
you have to reference them via the path, right?
those are setup in the blueprint
aren't the different drones have different meshes?
is there anyway of accessing unpacked files? I wanted to see how another game made bones for their character bodies.
even if they did, then i would have a few meshes to set visible/invisible, or just to swap
setup in the blueprint
enum
morning all
I don't trust in blueprint enums and I don't want to recompile, I think I would go with FName π
on the other hand I would have to type that, hmm
enums are a pain but trustworthy
life sucks
strings are far more untrustworthy (typos)
exactly
enums are a dropdown box
but it would get caught in my tester
but you can catch wrong ones for enum
sure but you type it once.
so yep I see your point, it's a good one
i never like to use FName for such things
the marging of error of typos is way too big
you should never replicate strings/names as well
and my game would have coop
im going to try to script NPC AI for the player
it could be interesting to have AI teammates that can be played by a real human in some cases
I say never like you can always get around it. Strings are fine for like playername but don't use fname for like bonename replication
ah yeah thats pretty cool
reminds me of csgo where you can control the bots
i can then do stuff like allow the player to select wich NPCs go with him and recruit them
like in shadowrun
of course gotta test how hard would that be
at least it "unifies" coop vs solo missions
now all of them are coop
depends on the game I guess but I think it is a rather hard task in general
to make a not so annoying coop ai mate
not the first time i do player squad bots
of course ill experiment
but they would mostly just kind of follow the player and attack everything on sight
i dont think you will be able to give them orders
its a vr game its not like i have a fuckton of buttons or weird commands
what i really hate tho is that I can't use a single anim bp for more skeletons
best I could do is to create a template one
and I have to duplicate it for each skeleton
annoying as hell
just use one?
because of course I can do a skeleton independent anim BP, i just store the anims in variables and get those from owner
I can't use the same skeleton
I have stupid monsters with very different skeleton setups
and the anim BP has to have its own skeleton
actually I just assumed that I can't play anim(skeleton B) in an anim bp (skeleton A), but i'll check it quickly π€
no, I can't
this is a nightmare
maybe the best way would be to automate the anim BP retargeting and duplication or idk
you can do anim tree in C++ for the "general" logic
and subclass it for different skeletons
but I don't want to subclass it
just you will have to remake the anim tree part...
that's the whole point, I managed to do a general logic, which works with the majority of my little monsters
but for some strange reason anim BP is skeleton bound
so if I change my general logic I have to "duplicate and retarget" my template anim bp to each monster
I don't have that many atm but it is already annoying
I guess if I could move up the general logic to C++ I would not have to do that, hmm
if I modify it, that is
yo im rendering a cinematic in Unreal and only like the first few frames actually show the scene
rest of it is just black
Ive done it before and now for some reason this is happening
How do you guys share 20+ gigs worth of data with fellow team members if you are not physically sharing the space?
perforce
I cant seem to figure out what their plans are. Just descriptions, TRY FOR FREE and other marketing stuff.
Ikr?
Apparently Microsoft offers this thing with unlimited data through GIt
Haven't looked at it too much yet, but it's interesting at least
Well yeah
Github does
Git is just the implementation
The limits depends on whoever's hosting it
And in this case, microsoft is apparently hosting it with some interesting stuff
Do you have a link?
I'll see in a sec
thanks
@weak thunder
Currently working on trying to set it up
Thanks. I will forward it to my people and see what they think.
Free for 5 or less peeps
I wonder if people could account share π
I working with a bunch of hobbyists like myself so people come and go.
guys
anyoine know a way to turn multiple actors at once to "use complex collision as simple"
cus there so many of them
what's it called when you map an existing folder location to another physical drive on windows? can't remember how i did or what it's called!
That'd be a junction @keen parcel
(There's more names for it though)
(I know it as a junction)
yes!!! symbolic link
cheers bud, so many similar things come up when i searched
just discovered the DDC has 36gb of my system drive
You know what's awesome?
YOU?!?!
Mapping to a NAS or something like that
Huh
Not what I was going for
But I'll take it
:3
building a new pc atm with m.2 ssd
havent checked but i'm guessing that a NAT would be a very slow way of managing compared to a local m.2
i got 1tb m.2 i think it was a waste
You could dump a bunch of m.2's in a NAS
insanely fast but doesnt help UE much
Well no, but you could've seen that coming
another SSD would be good enough
Typical data transfer speeds range from 3MB/s to 20MB/s for consumer level NAS devices
@half mango rly?
Why'd UE4 benefit that much from an m.2?
i think so.. but idk
but what about loading large projects? marketplace assets etc?
its almost too fast
everything is instant
there's another PC here with an old SSD
seems barely quicker
i bought too much RAM too l0l
64gb 3500mhz
definitely didnt need the speed
of the ram
again doesn't help UE
even in production bakes that take 40+ hours
the ram only uses 30% max
but threadripper.. always 100%
You didn't order yet? @keen parcel
i wish i got 2x threadripper somehow
not yet.. but spec'd it up with a 1080
gcard price has gone up cos of crypto mining
Oof, if only I had 1k lying around for hardware...
Β£1,800 predicted spend + Β£1k on new super-ultrawide 49" samsung
trust me though.. that's not extravegant when you spend 18+ hours a day glued to a screen
why the monitor?
those m.2 would be much better spent using for intel optane...
so i'd prefer bigger screen without the bezzles but curved
hey guys, when i write NativeParentClass:"InventorySystem.Item" in the content browser, it includes InventorySystem.ItemActor too
any way to make it only find the InventorySystem.Item ?
much rather have a solid cpu cache then a ssd that's speed is negligible
you see that announcment about samsung's 150" tv?
@honest rune good points btw
see that being like 10 grand
The CHG90 is available to preorder for $1,499 on Amazon, with a slated release on June 30.
too long to wait
and i think 49" looks about right.. any bigger and you are constantly turning your head
i dunno.. you can't really tell until you are working at it
That's the 49"
null
i got it in my cart for $1299.99
thought you were linking to a newer model
49
SUGGESTED RETAIL:$1,499.99
YOUR PRICE:$1,299.99
YOU SAVE: $200
hehe
it's almost 100% processing power
@half mango ^^ really?
i can see that having more cores would be very useful
the building yeah
well i bought high end everything
and it uses 100% cores
and still takes 40 hours build
with a 1950 threadripper
lighting?
π
i want 80+ cores
u doing game or archviz?
So after looking at some socket locations of ue4 weapons, i was wondering how the hell they line that up with animation its so sort of, random?
ah man, I made 2 seriously awesome and helpful APIs over the weekend.
A nice File System wrapper and a really sweet SQLite API π
did you make me a sandwich?
Should be able to wrap both we a easy to use blueprint api but I don't really do any more than necessary in BP so not sure if I'm gonna
If the sandwich is made of 1's and 0's then yes, I did
101 - sandwich (sideways)
Seems like it dunnit
Thanks for the link!
?
Seems a bit too awesome to be the case
Probably bandwidth limitations
But really, it is the impression I'm getting
well, they do own their own gigantic cloud so, I would assume their servers should be fine
TS is cloud is MS-hosted yeah?
if*
Yep
Well yeah
I just don't really expect them to give away their gigantic cloud for free
Pushing your source? @silver crown
thatd be such a dick move though lol let someone have 50gb storage but give them like 5gb bandwidth lol
xD
That's what github does
even with paid?
damn
The worse part is that once you pushed files on a repo with lfs
You can't remove them
I guess its the initial push that kills eh
Seriously?
Yeah
Contact support there?
The issue is that the solution is to remove the repo
and they rub it in iwth a "Purchase More" button right there too
Hey, is there a point where I can hook into the application start and end events (or the equivelants) without touching the message pump?
only thing I can think of without using the message pump is relying on a starter map to fire off some stuff at start before loading other maps. Still doesnt help for application exit though
Here goes
My only gripe with using VS integrated stuff for this so far is that I don't really get the amount of information I'd want
command line for life π
Because I'm a newbie with SC
The worse in VS is when it's stuck
VS's handholding is nice for now
Its lack of information because of it is annoying
I'd like to see an upload speed and that sort of thing
You could just git push in commandline for that
Progress in terms of size instead of only amount of files
See I wouldn't know the difference. SC is scary.
Lol
Actually that's kinda stupid
?
50GB is going to take some time to upload
Half an hour I think
π
gigabit connection?
Yeah
nice.
1Gb/s / 200Mb/s
200 up? wow. I've got 1gb down but only 30 mb up
woops
Cmon guys get your capitalization right
lol
1MB = 8 Mb
Yeah
The caps lacking makes it really weird
He really have 3Mb π
ye.. i do π¦
"slow WiFi" is now a synonym for low bandwidth ? π
well spotted
it's hard to get educated when there is no internet
#firstWorldProblems
On Windows, Git does not handle files in the working tree larger than 4 gigabytes.
ffs
Duplicate a 1GB file 50 times?
Can't say I particularly mind that as a cap
That's quite a bit above your average person's upload
...
git lfs
its still garbage
Ah?
hi, what should I do after NPC dies? For some reason, he gets convulsions.
any idea what sould I do?
@frank escarp Yeah right
also, git lfs scales HORRIBLY with number of files
if you take a small sized ue4 project
and try to LFS it all
LFS itself will choke to death
it would kill any git program you try to use
not really, only the content and source and config folders
but yes
you ned to do LFS add btw
andd it will choke to death
It's not an UE project
I'm not even commiting or pushing here
that code files or binary @keen birch ?
code works fine
binary LFS chokes to death
nvm I'm disk bound
No, I mean my own UE4 project
including Content folder?
is this the first time you push it, or you been working with it for a while?
First time
check your .git folder
after a week
and the responsiveness of the repo
my DWVR repo was able to choke literally every git client
It keeps everything in there? :p
Can't you just clear the git folder?
Like "I'm confident everything works in this version" -> Delete all old stuff
i tried to use LFS for the content folder btw
And uh
@keen birch I don't think
That would remove your history
@keen birch non lfs means every time someone clones your project, it clones every version of every file ever
pretty well
WITHOUT COMPRESSION
10 gb will go to shit in a couple weeks
even on a 2.5 gb project
how?
btw VSTS is pretty awesome
my git LFS (properly setup i think, on Content folder, Source and Config were in text normal git)
to start with, i was unable to push my whole content folder at once
LFS, around 6 gb
lots and lots and lots of assets (marketplace)
so i had to split it per folder, in around 6 separate commits
team size: 4
3 of us actually uses the repo, there is a code only person and 2 of us modifies levels and stuff
we managed for more than a year - no marketplace assets here tho, not even starter content
self hosted on a LAN server
2.65 gb
content folder is just 2,49
Idk how is that even possible tho
actually I know
we switched repo a couple of times π
when we were in bitbucket
there was a 2gb limit
we created a new repo when we reached the limit π
it's like the 6th repo
history isn't that long
but we are on gitlab for at least 5 month now, idk why is it so small π€
Few changes in the big assets?
maybe, the only big assets which are changed regulary are the levels
and they are not that big
yeah, biggest is 10 MB, that explains it π
anyway if we reach 10 gb we can just create a new repo for free 
@surreal viper You should try VSTS
Seems really awesome
Unlimited storage
Amazing bandwidth
@frank escarp You're right, lfs is trash
will take a look, thx for recommendation
told you
π
But it's expensive?
free for teams up to 5 epople
Nice
btw, i think "git lfs prune" command cleans up the .git a bit
but you now need to do it after every single commit
and, btw
are there other limitations for perforce? or just the 5 person?
you cant prune the server
no
just the 5 people thing
@silver crown means your server is now 25 gb
Yeah prune removes 5 GB on local
I know that π
Have to pay 5$ per month github for lfs because I tried it once on the wrong repo
repo that I don't want to delete
man for not much more than that you can get a server
and put perforce on it
with 40 gb storage
or more
git clones are by default a total copypaste of the server
thats very cool
Well yes and no
No because I can't delete it
Else I loose all the stars
I'm not using the repo anymore
what's the min and max z-order?
@silver crown just looked at your MP video of the voxel thing
man thats amazing
it makes me think of a PvP capture the flag style game where both teams try to wall off their flag
Thanks π
This one? https://www.youtube.com/watch?v=vF0IGAe6LJw&t=4s&list=PLjYKpKPK8E-_lxP5EuRxizhNuNBDYnT15&index=26
If you want to try it: https://drive.google.com/open?id=1uZ5ud83fjUl1jApYjY-O9nr2g5Of4Dy- Server: need to click Play, then enter his local ip (often 192.168....
π
Also have some nice features
Like custom graph editor
and my own C++ nativization
any UI/widgets experts here? I'd have a few questions π
already asked but nobody cares π€·
maybe output the zdepth of the widgets to the log and see what is going where and when
Wait perforce has free hosting for <5 peeps?
They really need to fix their damn website
Because I've looked at it and tried to set up a local server
Because I didn't know it was free for small teams π
I don't even know how to get the z-order of widget and the internet doesn't help
Oh
Yeah their website really is set up by a retard imo
@latent terrace if there is a valid way to set the z order it would be on the widget assuming your parent lets you set it
so the question would be how is your UMG widget set up for the hierarchy
correct as its parent is a canvas panel which respects z order
when I added the widget somewhere and I can set the z-order there, I hope it's setting it for the blur too
what is your root, the PAUSE ?
yup
no, what is it
a widget
it's another widget you created?
I've created a widget named pause and sent you the tree
gotcha so its going to be a default root which has no z order options
any way to make it have z-order options? xD
well it's the root so wherever you place this would control it's Z order
are you adding this to the viewport and its not going on top like you want it?
it's going nuts
i'ts randomly showing on top of shit that's supposed to have like 10k z-index
it's so weird
unless I specify the z-index in addToViewport?
if your adding a widget, its at 0. The items inside of it have their own z order based on that 0
so how should I set up my z-orders?
inside the separate widgets or on the main widget that's the whole UI?
yeah
basically I'm trying to create a pause effect that'd blur everything
and it's missing some of my widgets no matter how high my z-order is
are you just adding in widgets separately at different points to the viewport?
Saving to a file from unreal - what is the error log - result was zero, mean?
@latent terrace basically if you have a main UI, lets call it the UI, and it's got different parts to is there any reason the pause menu is not part of it that you toggle on or off? Alternately if you have a main UI that is lets say just a canvas panel that you add all your other widgets to at design/run time you can then be sure that your pause menu is higher than the other children of that main UI due to all parts of the UI being children of it.
but in general if you use the add to viewport and set the z order it "should" respect that if you put the pause widget at a higher number than the rest
well at some point you use the add to viewport to add your UMG?
the UI you mean?
I guess, I don't really know how your setup is setup. You have some UMG widgets you add to your viewport, say your stats and such. You want the pause to go over that which is a separate UMG Widget so your using Add to Viewport to add that one when you pause?
this sounds like how your setup is setup?
which is another UMG widget?
what is the parent of your PAUSE widget then inside your main UI?
and the Pause_UI widget has a higher Zorder set there than any other widget?
and all other items are children of that Canvas Panel?
it definitely looks right =/
when you have it unhidden in the hierarchy there does it blur like it's supposed to in the preview?
no
weird because it should. All of those items are children of the parent canvas panel so they should all respect eachother π¦
all I could think of is to start debugging each widget that is showing up above it in that preview window and see what might be causing it
well one of the items that doesn't work is being added by AddToViewport
it gets blurred when I set the z-order to negative or 0, but even 1 makes it stay visible
and the pause has z-index 99999
this widget has it own z order when you add it to the viewport. Any settings inside of it for z order are for inside that widget only
adding 2 widgets, one with 1 and one with 0 will put every part of the widget with 1 above the other.
when you use add to viewport
so any way to blur a widget that is being added by addtoViewport?
wherever you have this "blur" at would have to be higher when added to the viewport than any other widget you want blurred basically
this is one of the reasons having a main widget you just add sub widgets to at design or runtime can be helpful as you keep it all together
any reason you can't add the main UI itself at like 100 and let all the others you add to viewport be at 0?
Everyone I have an official announcement.
Mac is for pretentious assholes who think they are superior when really they are paying for shitty overpriced 1.5 - 2 grand hardware that equals to a 500 dollar Windows PC.
Apple products are locked down like alcatraz and you can't do shit with them, no one develops for them and that is the only reason they rarely get viruses.
Windows master race.
I develop for mac on my windows machine. Never had a problem on our deploy iMac tho.
You can emulate a mac on a windows but you cannot emulate windows on a mac, is all I need to say basically.
I can emulate windows no problem on a mac
parallels and bootcamp both work fine
I can even get bored and use Wine
But why?
Why would you ever want to do that, why do you have an apple product in the first place.
sometimes I want to run Everquest when on the mac
so viewport has its own separate z-index?
like if I use addToViewport, I manually specify a z-index for the stuff?
if I change the z-order on the addToViewport with UI to 100
how will my z-indexes change?
it will put that widget at 100 which should be above everything else
yeah but how will the children behave?
think of the add to viewport as adding a piece of paper to the desk. the paper has your writing on it which is your ui
changing the index on your paper does not change what is on it
will they all get +100 to their z-indexes?
they will not have any of their indexes changed
the renderer draws the widget added to the viewport with the lowest order first
it goes thru that widget and draws everything in it like it asks to be done
then it does the same for the rest in index order
just because a widget has children that have a billion indexes does not change the order the actual widget is drawn to the screen compared to any other widgets you add to the viewport
basically the z order/index INSIDE a widget is for that widget only to determine how that widget is put together. Once it is all done it is basically flat and drawn in the order you tell it to add to the viewport.
so if I don't expand that hidden part of the AddToViewport, the z-index is 0?
thanks, it made stuff a little bit more clear, I'll try a different approach π
yep default is 0
I guess a question would be why are you adding something else to the viewport other than your main ui?
maybe it should be a part of the UI somehow
yeah I'll try to put it in the main UI
as an excuse I'll say that it's not my project : P
heh
WOOO!
I think we are going to see some really cool stuff at GDC this year
Just saying
cloud gaming is impossible to do
physics laws are against it
(for full gaming, AI/physics sims are ok)
@frank escarp That is what Cloudgine is AI/Physics, etc.
I am curious to see what it offers with the networking side of things
are they going to turn into UnrealCloudgine? xD
honestly, i would be interested on a cloud hosting service
ive seen GameLift, but a cheaper alternative would be neat
hey guys quick question, not sure which channel to ask in. How do I make the struct UI be side by side like this? https://imgur.com/a/RAPXv
Just a casual word of warning
Your blueprint structs will inevitably break
And when they do there's very little you can do about it besides manually rewiring everything that uses them
Eh, not exactly true, but they're kind of fragile and it takes really getting them to know how to a) not break it and b) unfuck them.
That caveat there is enough to hold up the previous statement IMO.
Though I don't really know how to unfuck them, other than not saving anything else and rolling back. So if there's some secret to it, that'd be cool knowledge to literally never use.
Following my usual policies, I've never had reason to ever, ever create a UDS, so... yeah.
uds?
Er, sorry, that's the native class name for those. User Defined Struct.
I don't know what else to call them. "BP Structs"?
They're not BPs. They don't really share anything in common with BPs. They're both editor designable data structures though.
Ah. Yeah always hear them called "BP Structs". Reason being that its closer to that context (BP), I think.
And C++ projects will just define their structs normally.
Yeah. It makes sense on the surface that they're called that way casually. I never use them, rarely ever see them, and know them more from the engine side than from the editor side, so it don't jive with my perspective as well, but I get it.
They've gotten better over time, for sure. but I can't see myself using them.
Just too uh. "fragile"
@keen birch it's a coincidence you say that, because i am recreating my master struct since it broke due to an engine bug
and broke most of the project with it
master struct
So you know how doing that came back to bite ya?
Why don't you do the exact same thing again
Yeah I've sort of gotten the hang of un-fucking my structs at this point
You break them by changing anything about them while they're in use in Blueprints
So that's not really something you can avoid
Oh migration also breaks them half the time
Hey Sion. Do UDS which contain other UDS which then go onto have yet another UDS give you nightmares?
Hey that's what I do
Hey that's what you shouldn't do lmao
mine broke because i saved new default values while they were in use
Like I said, breaking them's kind of unavoidable
But I've largely learned how to un-fuck them
Save, restart, move stuff around in them so they recompile, recompile blueprints using them
what if I told you I know a way to never break a BP struct
even if I actually do?
nah, nah. you can use your flimsy little toy
how do you get around using the bp structs if theyre so fragile and seemingly necessary? (im not familiar with cpp)
but I know how to not break them
You use Cpp structs
Always create a new struct and move every blueprint over to using that one any time you change it. π
Well since you seem keen on abandoning them... go for it.
Afaik that breaks foreachloops @floral heart - Foreachloops lock to the struct you initially connected, so if you change the struct type in the blueprint it'll say it's a different thing and break
that's not what I was gonna say.
@worn granite Tell me all about your way, master β€
'tis a simple two step process.
1. Plan out your game thoroughly. You want as much detail as possible.
2. Implement everything in a data driven fashion. Do not nest structs, rather give them an FName with which you can lookup DT rows. Use them to init props.```
ayy now it can't break cause you won't be changing the actual structs, and the use is minimized.
But uh.
Does using C++ structs work ?
Afaik it does
wait so i should stay away from nesting structs?
because my S_ItemInfo contained S_WeaponInfo which may have borked it
Think of a block-built tower
You've built it too high so it's unsteady
So instead, you just place another tower on top of it
And maybe another one
With each tower you're exponentially increasing the odds of it completely falling apart and dying on you
π
Think that's a pretty good analogy on why not to nest BP structs
if you have smaller structs which you can slap into a datatable
you can use FNames instead of nesting them
Imma have to lean towards @livid haven's word here.
And I already explained why lol
lmao no
Sorry, was out of the loop for a minute.
In that context, User refered to editor-user.
Admittedly, sometimes user means "the engineer using this API".
I'd say that's just as arbitrary as BP struct
Except people will understand what you're saying when you say BP struct
It really doesn't make sense to call C++ structs user defined in this context.