#Team Fortress 2 Megathread - winter is the most chill time of year
1 messages · Page 59 of 1
*i guess this is what
they meant by it being a
"consistent cone now"*
its a consistent cone. Just as long as you dont look in any diagonal direction
btw does it come from the eyes now instead of the pyro's centre
it's using the eyeposition vector, so yeah probably
someone should make a debug overlay for this
it's the same except for more broken decomp
show anyway
it might be something really stupid
different compiler, different codegen
im wondering if it's compiler shit that causes the different behaviour though
well that's probably most likely the cause
i think it's a bug on the vphysics side
were there any physics changes
they hated jesus because he spoke the truth
different compiler settings(!)
cone op
cone strong enough to cause a ping noise
how did it reach when you crouched pointing down
no idea
im assuming the last one was directly upward
looking straight up and straight down
srraifhr up makes sense but straight down is mystical
it seems fine to me, it looks more sane in firstperson
again, someone should make a debug overlay for this
ngl i only now realized jumping means you extend the airblast range if aiming downards 
did you actually suggest a cvar that detects cheaters?
politely request that cheaters set cl_is_cheating 1
yes because thats essentially what every single client side suggestion is
any in-game solution
the only real solutions are steam trust or improved client side vac
but client side vac isnt gonna get much better
so steam trust is the way
actually how long do you think it takes for someone to log into steam, download tf2 and run it?
because if all these bots are using the same hardware with different accounts just prevent joining multiplayer games of tf2 on accounts using the same pc.
how do you detect something as "the same pc"
thanks redditor
what’s a virtual machine
prevent hackers by making tf2 cost $$$ to play
kernel level vac would fix this
no one is using a vm to bot.
and phone number
ow2 player
if they're connecting from the same ip, same mac address and most hardware specs are the same. what if it used those to determine if it's the same pc?
all of that can be spoofed
how can an ip be spoofed?
vpn
because there's no need to, because detection doesn't care about being on the same PC. But if detection did start doing that, there would be a reason to virtualize/spoof/vpn/etc.
that's not spoofing, that's basically a redirect
whats your point
and even then how many vpns do you think 20 different accounts can connect to before their isp thinks they're getting DDosed?
more than 20 vpns
that it's not easy to connect to multiple vpns at once and sounds expensive.
they dont need to be paid services. You could just ask your buddies to act as proxies for you and that would work in this case
wouldn't that just give data on whom are bot hosters?
im not following
and even then how many of them do you think have friends at all?
@languid haven has leveled up! (28 ➜ 29)
if your security only works if people work alone, its not very good security
so what? you're saying some organized group of people are bot hosting to kill a game that means nothing to them?
that is essentially the status quo, yeah
ok, tell me when you find sewer squad
would it be impossible to bring watchdog over to tf2? it was working pretty fine in CSGO i think they ditched it though
and judging by the booming cs2 problem they could really use watchdong rn
lol i mean watchdog
or was it called overwatch
they called it overwatch and tbh I think they ditched it because it was just getting diluted by people giving bad faith demo reviews
its overwatch
for example there was someone near the end of overwatch's life cycle who setup a neural network that just watched overwatch cases... which defeats the point
not to mention the fact that because the majority of the playerbase didn't use overwatch it probably started to get more bots than real people on it
thats my guess tho
ogresnotch
its actually a fantastic system for a game like dota
*its actually
a fantastic system for
a game like dota*
yet again other valve games fail so that dota can thrive
valve should launch account trust factor and then base the weight of an overwatch report on that
planned
planned for 2095
there are whole neighborhoods using 1 single IPv4
this is common practice by ISPs on many parts of the world
even some powerusers find it desirable because when an ISP does that it can plasably deny companies seeking to send end users nasty letters they can point out they don't know which of the users did that because the company wouldn't specific a port number
does steam support ipv6?
probably, its 2024 ffs
I do wonder if tf2 somehow supprts it
I know that Minecraft went a long time without supporting it
to its credit, I don't think the idea was that IP addresses would be the only thing to identify a specific machine. It'd have to match IP and mac address and hardware and etc.
So the issue is "all of those can be changed or faked fairly easily for a bot network operator, which defeats the entire idea"
just solve the problem, it's that easy!
I see
Some people REALLY have HUDs that have the auto aim hud from HL2 (CHUDAutoAim) as an element which stopped working on the 64-bit port of TF because it relied on getting lucky treating the C_TFPlayer as a C_BaseHLPlayer to early out and do nothing x_x
Joys of any 64-bit port! :D
wtf they fixed the dxlevel crash
autoaim hud?
you specifically requested the opposite..
serious question, why are bounding boxes so prevalent? seems extremely weird to use a hitbox that changes based on the direction you face for so many things (melee, projectiles, airblast, etc) and I'm sure there's a reason but I can't figure it out
chud auto aim
its cheaper and works well
did you mean doesnt change based on the direction?
ye
but yea its just easier to calculate or smth
easier to calculate and to players its more consistent
imagine if you missed a rocket because someone's arm was in a different spot
saving on calculation I get but like, can't really understand how it would be considered more consistent. I can see the argument for explosives but definitely not something like the huntsman for example
me when he calls for medic so my flare misses 
putting aside the bugs with how it calculates headshots, I just mean the standard hit detection
hunstman is like the only exception
that I can't answer
hunstman is its own enigma lowkey..
I mean. any projectile weapon runs into this imo. guillotine and sandman can both feel pretty random at times
I hate the cone I hate the cone I hate the cone I hate the cone I hate the cone I hate the cone I hate the cone I hate the cone I hate the cone
I want my funny box back
cone better
i mean they fire at a delay by intention so they always feel kinda odd
no way did they really forget to put in references to the localization strings for other speaker settings
?
its not that bad bro
THE CONE IS EVIL
is there something im missing..
i am a certified cone hater™️
actually wait follow up question. if the argument is that it's meant to save resources, doesn't that benefit not apply with airblast? since it's doing two checks now if I understand correctly
is it a gameplay issue or is it a whats going with it issue
the cone check only runs if the box check passes
oh I see
it always was doing 2 checks but it was fucked up
Airblast was sad ugly cone -> became box that allowed silly by airblasting behind -> back to cone now
It still allows silly airblast, you just need to look up or down
The way they did the cone check is pretty expensive though iirc
The old cone was just a dot product check
also why does every shot with the dragons fury chug my frames for a moment
Was it?
DF is silly
Pretty sure, read the leaked code
and does this happen when im fighting one too?
unless im missing something
Spawning dynamic lights in general causes silly in the game
*Spawning dynamic
lights in general causes
silly in the game*
At least in 64bit
nah it was always likej that
dx8 and everything df always causes frame drop
doesnt that do nothing anyway
I get a stutter when I use funny flashlight command on servers that enable it
they should bring back the coneless airblast so i can pop people into the air with less thinking 🛐
no way did they actually add dxlevel to linux
there are mods to fix this
vanilla spawns 50 billion lights in front of you when you fire
it's stupid
Awesome!
fuck off github
what the hell lol
microsoft penny pinching
can someone on linux test dxlevel
i can real quick, what should i do and what should i look for
I refuse to believe this isn't just like removing things from OS specific compiler flags and making them generic
does it look like dx8
does mat_dxlevel report dx8
etc.
don't remember what dxlevels look like which, never tweaked them
it'll look bad thats how you'll know
build a teleporter and screenshot it or smth
the casual medal will be broken looking
fireballs have dynamic lighting
mat_dxlevel is 92
as expected
it does nothing
@balmy root i'd post your message here so more people see it
ah whatever i'll screenshot it from that help thread
intersting change
I wonder what effects it has
even tho it says "def. 100"?
yes, it gets reset to 92 it seems
huh
it's because before this update the cvar had a min. value of 90 and max. value of 92
maybe I was right when I thought it was a weird half-baked thing to remove OS specific behavior
I talked to a person on that thread and it's probably related to dead code from ToGL, which forced dxlevel 92
I think it's something related to compatibility with OpenGL
death to dxvk, togl is the best 🫡
we're gonna start seeing fixes for bugs decade olds in tf2 (because maybe now is already a part of tf2) or is just gonna be like fixing regressions problems with porting to 64bit?
i doubt theyll change nything too major in gp like that
unless it really is just negative and unintentional
cs2 native vulkan 
I think seeing bug fixes from decade-old bugs is a side effect of porting TF2 to x64 and fixing the x64 bugs
valve stonks
IIRC it is
but performance on Linux compared to Windows is bad
and I mean it
that was the joke
also from this thread
now it reports m_nDXSupportLevel: 100 from mat_info
before it only reported 92 on Linux
it is a step forward for my issue which requests 95 to be the default
dx7
yes dx8 should be dead by now
d8vk in tf2 when
IIRC I saw a youtube video many years ago about a person forcing dx7 on tf2
it was horrendous
lemme see if I can find it
comp players wet dream (it gives you -10 fps but your game looks even shittier)
modern-looking games made to resemble pieces of shit from 2002 is my new aesthetic
_
song is some vaporwave
edit: song is VGM #4 sound shower by Vektroid
they probably just removed this ifdef
this
and set max to 100
@craggy ridge huge
oh hello gorgeous
wait this update means no more directX 69 in the settings :(
need that rn
on this video it's possible to notice the lack of support for transparency
which makes me think it's DX7 or lower
it seems they still require it to be 92
even after removing the code
I mean we don't know exactly what they did
rip directx 69
it probably is that the ifdef was removed and replaced with a general check
hmm
what I noticed
is that mat_dxlevel previously on my machine was like "92.0000000"
so the cvar is the same defintion on both OSes (note the fact that it now has its description from windows)
after this update it is only "92", and also the description was updated to mention competitive
before the update, it only was "Current DirectX level"
now, it is the same desc. from Windows
yes
exactly
I heard there was a way to use DX10 but on the steam website it said that was for xbox or something.
Ah
on TF2 XBOX afaik the DX level used is 98
ye
how fucking dense is this guy holy shit mods can you please send a ballistic missile to their house https://github.com/ValveSoftware/Source-1-Games/issues/5868#issuecomment-2068378795
dxlevel 100 seems to be a workaround for removing device limits
Explain
ok so
looking at the dxlevel code
essentially the various dxlevel values set specific variables, usually disabling things
except 100
100 just ends without setting anything
That is....kinda funny. So wouldn't that cause a crash or would it default to something else?
Or does it just use whatever was previously set?
so, they call the variables "caps", so I assume that it is essentially just lowering maximums and disabling things, and the default is "always on", or the highest possible value
judging from dxsupport.cfg it's theoretically possible to use dxlevel 50 on tf2, but that would require various hacks I assume
for example, all of the dxlevel code (except 100) tells the game that the GPU does not support DX10
theoretically, but the code i'm looking at only goes down to 60, so anything else would get clamped I think
i thjnj the game would crash due to all of the shaders not being possible on directx 5
I heard dx9 and dx10 were similar but i'm not so sure.
also a possibility
are there any visual differences specifically on dxlevel 100
Or performance differences? Or any at all?
no way this the slender im thinking of 
also a big ass particle that gets stretched over a space greater than ur whole screen
IT SO MUCH BETTER NOW!!! It doesn't flash bang my screen everytime i fire and reminds me of rockets from quake when you fire them.
supposedly dxlevel 100 uses "dx10 blending", idk what exactly that is but I assume its just a faster version of dx9 code idk
tbh the only game I know that uses DX10 is GTA 5 (but the default there is DX11)
it seems many games jumped from DX9 to 11
10 is quite "unpopular"
11 is much more laggier though
it appears to do something regarding converting to SRGB color space?
And 12 ain't much bettee
maybe on dx10 supported cards its not necessary so its faster
from a technical perspective 12 should be better
iirc 12 is better but harder to setup well due to being more low-level
its basically a completely different beast
ye
from Wikipedia:
Lead developer Max McMullen stated that the main goal of Direct3D 12 is to achieve "console-level efficiency on phone, tablet and PC".
I usually had it set on windows when i used to use it and i have pretty similar performance to the 32but. ToGL with it except a lot more stutters.
do you have an intel gpu?
achieve "console-level efficiency on phone, tablet and PC".
keeps code proprietary and closed-source
or did at the time?
It's not, believe it or not. You get better graphics emulating the same game on PS2 with Vulkan or OpenGL.
whatd u do
i mean
consoles are proprietary and closed-source
because it seems like on intel cards they force disable it, but it probably doesn't take into account newer intel GPUs
intel fards
Nothing, it's just better now and idk who changed it but i love it even with the no lag dynamic lights.
Bro
💀
this is also just one use case. vk and dx12 is better comparision, but ogl 
tf2 dx12 update tomorrow
i was sitting in spawn firing that thing and still digging trenches in my frames
unless that was thursday
yeah but you get more control if you can manipulate the entire code and not just use dev blogs/guides
microsoft restricts the devs choices to what they want to do, whereas in vulkan you can do anything
Tf2 now
do you think devs are using vulkan forks for their games?
no they aren't
is the df pause your game by left clicking thing within the scope of what joshie would fix or nah
vulkan programming is quite different from dx
🎉
and it's just simpler to program in dx and then slap dxvk on top of it
(which is partially why valve is using dxvk instead of native)
finally
comparing old versions of dx with vulkan is kinda silly
we can be gordon froman in fort team double
I mean they don't even fix simpler things
u a time traveler or something
imagine they wanting to create a native vulkan render for a Source 1 game
not happening
:(
its the 23rd in some parts of the world
different timezone
impossible. nothing exists east of me
OpenGL has less bloat. What causes most lag is post processing and not understanding what's causing lag.
what
linux mindset 
Arch has less bloat. What causes most lag is post processing and not understanding what's causing lag.
It's why opengl es has better performance then opengl on Similar versions. It's not better and supports less features but it makes it faster.
makes you think
how does opengl have less bloat
it's open
Compared to DX12?
so all the bloat walks out
its not opensource
LOL
windows 98 has less bloat than windows 10 because its older and doesn't support modern hardware
*windows 98
has less bloat than windows 10
because its older*
that has nothing to do with "bloat"
actually called it
Actually idfk why they call it openGL.
Open Graphics Library
opengl probably bloated with that ancient 90s codebase
But it ain't open, maybe it is compared to direct3D
tf2 is bloated because it has code from quake

voodoo 2 graphics cards from 1998 have less bloat than ti1080s
TRUE
tf2 is bloated and need some tums
this is so real they should remove the original
how is opengl not open?
closedgl
closedgl
invitegl
FUCK
HAHAHHAHA
How is it open?
gdi
Idk why a sega genesis emulator ran fastest with that.
So vulkan but worse.
Didn't the company that made opengl make vulkan?
it’s foss, why would you do that
it’s immediately better
i still gotta try out mesh shaders at some point, hardly even touched compute shaders. Basically only worked in vertex and fragment shaders
this is why I uninstalled google maps and use osmand now
osm stans are next level
i like foss but god damn i want to get to work at a reasonable time
or be able to find places
On windows, at least on nvidia, alt tab should be just about instant in every configuration
If it's not instant, you're doing something wrong
like leaving dxlevel in launch options or something weird like that
i think joshie actually blamed nvidia drivers not supporting something for slower alt-tabs on vulkan
it is im very impressed
and same
mine flickers black for a split second before tabbing in/out
what gpu?
GTX 1070
Yeah I have the black flicker too using dxvk
I still considered that instant, it's fast enough
But no black flicker without dxvk
For me it flicks black for maybe 0.1 seconds using dxvk
to be completely honest I do kinda disagree with joshie here because i haven't had this issue in any other vulkan game so it kinda... shouldn't be up to GPU drivers idk
like yes the driver should support this but also you could fix this i think
FWIW toggling the nvidia setting for vulkan present method between "prefer native" and "prefer layered on DXGI swapchain" doens't have any effect on alt tabs for me
Both are very quick but both have a short black flicker
yeah I was just about to check that since I googled it
fascist
maybe we wait and hope that they add some specific driver settings to the new tf2 exes
I just used a profile inspector and added the new EXE names to the "Team Fortress 2" profile
what driver are we talking about?
nvidia windows drivers
idk they have all sorts of wacky behind the scenes settings
source has its own antialiasing compatability setting!
it’s the linux in him
lmao maybe
this regularly happens there
Yeah it's kind of crazy how much application specific shit there is in the windows gpu drivers
No wonder they're like 1gb lol
I think it's funny how the "Antialiasing Fix" in Nvidia's drivers is ONLY used on TF2 and all it does is make some AA color sampling less aggressive on certain generations of cards and nothing else
every driver does this
https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/src/util/00-mesa-defaults.conf?ref_type=heads
guess TF2's art style was too vibrant or something
linux drivers also arent much smaller. nvidia’s for example is pretty comparable to nvcleaninstall
and mesa is in the kernel anyways
well interestingly, Pc gaming wiki actually has a tutorial for setting up this specific thing, but it doesn't work
so i'm gonna go out on a limb and say that its not necessarily nvidia's fault
especially considering it says this
I mean the alt tabbing is fast for me
I wonder why it would be slow for some people
Maybe differences in nvidia generation but I kinda doubt it's that
alt tab is a little slow for me but before during 64bit beta it caused my pc to lag, stutter and freeze. way better now.
It was instant before and instant now
Did you have -dxlevel in launch options or anything else like that
or change antialiasing mid-game
or anything causing a materialsystem reload
now it just takes 3-6 seconds to switch onto and off of the fullscreen tf.
nope
What could the difference be?
native ogl would fix this
nah
have y'all confirmed whether or not Windows is applying it's "Fullscreen Optimizations"
?
native glide support would fix this
it's more of a non-issue right now for alt tabbing. there's bigger problems really.
I have this off, if it matters
so normal exclusive fullscreen
but when I use the fullscreen optimizations, still instant alt tabs
*but when I use the
fullscreen optimizations,
still instant alt tabs*
oh also comfig is not using the precomfig so i'm stuck having to use autoexec on linux instead because tf2 doesn't even detect precomfig either.
what
if your keyboard has volume controls it's easy to test by seeing if the Windows volume overlay shows up or not since sometimes fullscreen optimizations don't get turned off even if you tell it to be off
you mean pre_comfig.cfg?
tf/cfg/overrides/precomfig.cfg
i did not name it
i think your problem is that precomfig is the wrong name
dafuq
Learn how to do advanced customization and scripting with mastercomfig.
thanks
Turns out this isn't true
Black flicker is a bit longer on "prefer native"
it's barely noticeable but definitely just a bit longer
sounds like a shit post like when sniper eating his rifle caused a crash
*sounds like a shit post
like when sniper eating his
rifle caused a crash*
I mean it makes sense
the tf2 hud uses some of the hl2 hud as a base
and I think some of the tf2 player code uses hl2 player code as a base
Some people REALLY have HUDs that have the auto aim hud from HL2 (CHUDAutoAim) as an element which stopped working on the 64-bit port of TF because it relied on getting lucky treating the C_TFPlayer as a C_BaseHLPlayer to early out and do nothing x_x
Joys of any 64-bit port! :D
what the heck does the autoaim hud even do?
I barely played HL2, and I only recall autoaim being present only on easy difficulty and on the game's console ports?
wait i thought tf2 didn't have autoaim
sentries and wrangler doesn't count
it doesn't
i don't think the autoaim hud actually worked
so why fix it?
did you read the patch note?
if you want to get rid of hud things that never worked and do literally nothing, you'll have to pry them from my cold dead hands
i'd prefer if fov setting worked. it's forced on 90
it works fine for me
it just changes model fov now for me
Normally I just have to restart the game when that happens
fov_desired
they seem to be the exact same.
or at least they are now
hey are wait events still causing issues in 64bit?
they already save that info over multiple files
Damn this new update is cool
I'm cranking up more settings on my steam deck config and I'm still consistently reaching 60fps now
remember that I would usually only hope to reach a peak of around 40~ fps before this update
maybe don't use max settings?
in tf2?
how are you getting that low of fps in tf2 on a steam deck
idk that's just what I've known since getting my steam deck in 2022
My steam deck config before this was mainly centered around the comfig low preset with every graphics setting that mainly taxes the GPU set to max
Because the only issue with the steam deck is its relatively weak single core performance
what have i done
how annoying is it to get working
your answer decides if i will work on it or not
It's manual math I did based on what I was told lol
oh i mean vscript in a listen server
Rn you just execute the script then call SetupWeapon()
when i did the airblast thing years ago source didnt have a way to show cones i had to do a bunch of math it was annoying
vscript is arguably easiest in a listen server because you can just call script_execute scriptfilename from the console
or do oneliners with just script
I'm getting food rn but it takes in a cone angle and max distance and color iirc it's been a minute since I wrote it
Doesn't work if you want spaces cause source command parsing says fuck you
*Doesn't work if you
want spaces cause source command
parsing says fuck you*
iirc the way i did it was using big triangles for the transparent part and then drawing the little horizontal lines across it
seemed like the simplest way to show a cone in debugoverlay
i don't tend to do oneliners cause what I do is more complicated than is possible with that
you can't do triangles i don't think
you'd have to fake it
wait wdym you cant do triangles
vscript says no triangles for you
???
that's cringe
My box is done with lines instead of box because I am a rebel
so you'd have to jerryrig it with math
*My box is done with
lines instead of box because
I am a rebel*
oh wait i didnt even do the lines https://m.youtube.com/watch?v=sZ0axNXvNik
Airblast detection uses a 256x256x256 cube, not a cone
Projectiles and players touching the cube will be airblasted. The cube does not rotate with the pyro, so projectiles and players can be reflected from behind at certain angles.
Before Jungle Inferno, a 256x256x128 box was used, which is half as tall. In addition, players had to be within a...
it's just triangles here
i guess the transparent part isnt really necessary
(I am too used to doing everything from scratch in vscript and forgot there was a box)
yeah lol so much stuff you gotta do yourself
can you put a tranparent colour through the box
yeah
and it'll do something like in the thumbnail here?
specifically this section
https://developer.valvesoftware.com/wiki/Team_Fortress_2/Scripting/Script_Functions#Printing_and_Drawing
idk why for only lines they have a function with a color input as vector and one for separate RGB
I wish there was a draw tri
Time to make gothub issues ig
that would be a bit too convenient
hmmm
you could always make an entity thats a plane and assign a texture to it thats a triangle 
i hate this steamhapp so much

i'm going to inject a check into my discord that looks for the hash of this steamhapp and removes it
and then i will add support for an AI to check for similar images
and it will brick my discord
Maybe a draw poly?
UB what angle did you use in your script?
if you can only have one then imo its better to draw arbitrary polygons with triangles than draw triangles with arbitrary polygons
but no reason you cant have both
🤔
36.87
Also nvm draw cone wasn't customizable
I had the numbers hardcoded lol
yeah ive noticed this a lot
on a 45 degree ramp your crosshair is off
idk if this is related to your crosshair being off on moving platforms
I forget where I pulled this out of
But yeah that's what I used
I also have 0 clue where it comes out of the body lol
@livid oracle
isn't it 37.5
Nyooom
so our hypothesis was right
thats for the old cone iirc
huh so it was decreased by a tiny little bit?
i assume because it will feel a lot bigger anyway
I guess they silently tightened it in JI and never documented it because it didn't actually properly math the check before?
Maybe the vphysics function they use after JI always returned true till 64bit?
acos(0.8)
something like that
I actually checked this yesterday, and in the old builds it returned really weird values for players outside the cone
according to replay at least 
idk if it's used anywhere else, doesnt seem like it would be
*idk if it's
used anywhere else, doesnt
seem like it would be*
So bugfix then
Guess I'll report to github issue
yeah but afaik joshie claims they never touched it
Joshie never touched the flamethrower code, might have been used in another spot?
whose pronouns are it/its
Or some overflow fuckery caused it (super super super unlikely)
flamethrower
i am flamethrower
at this point im convinced it's just compiler bullshit
compiler optiyon
what are the values
7
it would return 0x41 iirc
i dont think it's 7
gothhub
lemme see if i can get the script working again
0x41 is a weird return
especially for a bool
Mmmhm
let's just check the source code of BoxIntersectsCone 
Throwing exception and getting interpreted wrong? Potentailly?
(Literally pulling any answer out of my ass)
How do you enable this DXVK overlay?
env var DXVK_HUD
lmfao what
so you can't airblast when the server is using jolt?
wtf
probably
yeah?
Why do you have a deific bootlicker on your friends
older version?
yeah older version
reminder that @teal sand is transphobic (hates transmission)
Oh yeah, I thought about trying to implement something like Minecraft's brigadier as a chat command parser in vscript
Its probably trivial but at the same time it seems cursed to do idk
computer
Huh, dying changes your bbox position? I didn't know that
I forgot to turn on my silly name, there we go
you can move it around too
it follows you in free roam
Damn
yes
should be fine to do anything with it as long as you run it with -insecure
Wack
Oh that sc didn't capture what I saw lol
Its bugging out my monitor for some reason
The lines sometimes render over my viewmodel
wtf
your gpu will burst into flames in 10 seconds
Nesting haters rise up (nasty code* still though)
- More than two levels of nesting (war crime)
btw it's 35 degrees for sure right? i assume that's the degrees you used
Ignore the weapon.GetCenter() I forgot to remove that 
so i guess here's the JI cone changes we know of so far:
- cone degrees decreased from 37.5 to 35
- checks intersection with the player's entire bbox instead of just their centre point
- comes out of the eyes now instead of the pyro's centre point
Yep
can you check if like
if this would count
Let me enable box draw
basically checking if the new cone has infinite range
or at least just enough range to reach the box's corner
btw imo it's easier to test this with a pyro bot instead being the pyro for this
cone length is logged in the video
wait wdym kinda
256?
yeah
the box's diagonal length would be ~362
however the cone does start inside it
i forgot what the box's offset is from the pyro
i do know it's offset from the eyes, not the pyro's centre
Does the box move up/down with view angle I forget
it does
might be in this vid https://www.youtube.com/watch?v=bNnYJwW_3DI
Hi guys I was gonna make an actually edited video on this topic but a lot of things happened back to back in a row, including but not limited to discovering airblast pogoing, joining a local band and getting drunk too much, The International 8, Monster Hunter: World coming to PC, and my second monitor dying.
In the one where I airblast the dud...
nvm he doesnt mention the offset
@misty plover oh it's literally just 128 units
Mmm then no the cone isn't infinite length
Oh interesting vscript draws stay in your demos?
Anyways I'm gonna nap
I'm still running a fever and all that
so does this not reach?
this is a 256-length cone
that being blanket

are you guys edating?
@potent wolf request: can you test this 

tomorrow, am going to bed
Why would this reach.
the user known as UB claims it did
Doesn't
I thought at first you needed to be intersecting with the airblast box over a certain amount but you don't
ok so the cone is not long enough to reach the corner of the box
yeah you just need to touch it
Its hard to visualize the cone properly cause I'm not able to freecam lol
like bros name is literally ub. wait. like bros name is literally wget
you two probably stink
"robert"
robert the cat
this means this is the actual player airblast shape at 45 degrees
@next nymph how does this make you feel
Idk why but I thought you needed to be past the center of the airblastee's BB hence "kinda"
I love this actually
diamond
trippy]
Cone that morphs into a cut diamond when rotated 👍
Wait hold on another thing I want to test is projectiles let me boot up walkway
I mean if you think about it I think this kinda works out because of player bboxes not rotating
so you have the pointy end into the diamond
and the flat end into the flat end
now make an over complicated visual representation of it
or at least thats what get passed to isconeintersectingbox, check how that value gets used (h in truncatedcone_t)
cryptic messages
Okay good projectiles only care about the box still that's gamer
yeah it should be unaffected
Yeah but i wanted to sanity check just incase they meant to change that in JI and never did too lol
Wait how old is this bug then, five years it went like this? And we assumed it was intentional? Wack
yeah lmao
Oh I think I found why it always returned true
Maybe
Relies on some assumptions I can't prove rn
what's your guess anyway
tr.startsolid; is what IsBoxIntersectingCone returns. Looking at the call path it uses DoSweep which calls SweepSingleConvex.
This goes down a way untill
// contains the origin
if ( simplex.SolveGJKSet( vert, &m_pointClosestToIntersection ) )
{
// Removed for brevity
}
Which if passes it then solves along the sweep, and checks the ray's length.
if the ray length is set to zero the check fails, startsolid and allsolid is set to true.
if ( m_ray->m_length != 0 )
{
// Removed for brevity
}
else
{
m_trace.startsolid = true;
m_trace.allsolid = true;
m_traceLength = 0;
}
IsBoxIntersectingCone(Bool) returns tr.startsolid and coincidentally the ray it feeds into the trace has a length of 0
// offset the space of this sweep so that the surface is at the origin of the solution space
CTraceRay ray( vec3_origin, vec3_origin );
what makes the behaviour different between 32 and 64 bit
Gonna guess behavior was changed somewhere in this (since the comment seems to imply its not supposed to have a length of zero?)
Would need Replay to decomp it
bool CPhysicsTrace::IsBoxIntersectingCone( const Vector &boxAbsMins, const Vector &boxAbsMaxs, const truncatedcone_t &cone )
{
trace_t tr;
CM_ClearTrace( &tr );
bool bPoint = (boxAbsMins == boxAbsMaxs) ? true : false;
CTraceAABB box( boxAbsMins - cone.origin, boxAbsMaxs - cone.origin, bPoint );
CTraceCone traceCone( cone, -cone.origin );
// offset the space of this sweep so that the surface is at the origin of the solution space
CTraceRay ray( vec3_origin, vec3_origin );
CTraceSolver solver( &tr, &box, &ray, &traceCone, boxAbsMaxs );
solver.DoSweep();
return tr.startsolid;
}
But yeah, I think the ray they pass in has a length of zero pre-ji
And coincidentally ntohing uses IsBoxIntersectingCone in the pre-ji codebase (as far as I can tell)
i assume nothing other than the airblast uses it even in post-JI
Most likely
So why it got fixed would be even more confusing
And it also once again relies on simplex.SolveGJKSet returning true which it shouldn't in this case
The only other answer is its reading uninitialized memory
Actually no wait is it literally that
But why would it be consistently always be passing back then? There should be a nonzero chance of it failing and rejecting no?
Head hurt
Those are the two answers I can find looking at the code, it seems more than possible that the trace isn't initialized properly though so tr.startsolid is junk
No, CM_ClearTrace memsets it to zero

Which means this has to pass, which it shouldn't be able to
@fathom eagle has leveled up! (3 ➜ 4)
I eated it
anyway i confirm this
it only reaches when you touch the cone, length of 256
is the edge of the cone curved
@misty plover can you send your visualisation script 
Sure just ignore drawDebugLines its not a real function
This function does not exist
I would never write this
I would have used the actual draw box 🙏 surely 🙏 I totally didn't forget about it 🙏
Very soon into 2024, I was playing TF2 as normal and then I was suddenly thrust into the middle of a trial where both sides would have to argue and figure out whether the creamy penguin was cheating or not. After my very first kill of the match, the trial immediately began. Will the truth be revealed, and will I be Innocent or guilty in the eyes...
Why link his video?
What if I did something absolutely awful and made it so it draws the cones with circles instead 
i like it
Itd probably just lag.
And would look something out of roblox.
I'd have it draw in a fixed amount of circles like I do already for the line config
oh yeah idk why i didnt notice this first it's definitely flat at the bottom
Epico
apparently this is what a truncated cone is, it's definitely not this though
it's only 4 values
Nvm deleting those those posts I was wrong
1984
Are there any examples of CHUDAutoAim being used in a hud? I’ve seen the code for it but I’ve never seen the element or seen anyone talk about it.
chud auto aim
Yeh but what’s it look like?
Then why’d Joshie tweet about it?
chud auto aim
@livid oracle he tweeted about it because it does nothing, resulted in a crash, and yet HUD makers still use it
I’m surprised I didn’t see anything about it. Though it makes sense people trying to use that wouldn’t be the ones in the huds.tf discord.
Wtf why are you leaking the name of my cheat... I thought what we had was special...
chud auto aim
Chud auto aim
@potent wolf another favour
: can you check if CTFWeaponBase::DeflectProjectiles (or maybe CTFFlameThrower::DeflectProjectiles?) uses UTIL_EntitiesInBox or UTIL_EntitiesInSphere
also where'd you get "deflection_radius" from?
i want Confirmation if the airblast is also supposed to be using a sphere instead of a cube
explain...
I'm putting chemicals in your water supply so you forget about the airblast
You're in too deep it's for your own good
the chemicals in question:
looks deadly
anyone know why reflect jumping with cow mangler sends you notably further than regular rockets (even if its not a charged shot)
does it deal more self-damage than regular rockets?
like sentry rockets (deals up to 170 self-dmg)
yes
well then that's it
but why it do that
i didnt even realise it would affect cow mangler rockets too
*i didnt even
realise it would affect cow
mangler rockets too*
wht would affect
Pyro does not take mini-crit damage from their own reflected explosives. Players will never deal critical damage to themselves under any circumstance.
Reflected grenades, sentry rockets, and detonator flares deal 85-170 self-damage. Normally, grenades deal 37.5-75 self-damage, sentry rockets deal 50-100 self-damage, and detonator flares deal 15...
read desc
guess i'm adding cow mangler rockets to this 

i dont have like
scientific proof of this
but if you were to jump off a cow mangler and a regular rocket
youd notice the difference
is 89.63 the max self-damage a reflect rocket can do
@livid oracle can’t believe you’re reddit famous
https://rxddit.com/r/tf2/comments/1cax67m/unconfirmed_apparently_spycrabs_are_no_more/
woke reddit be like: rxddit
wake up baby, bots are back
They've been back for a bit
spycrabs still work
nu uh reddit said otherwise
Lying
Can every pyro vs demo/soldier interaction just be decided with a game of four squares so we don’t have to think about this cone business anymore 
what's four squares?
okay, I had some scheduling problems, but I'm fairly certain Im gonna start testing when I get home
so it's gonna be:
test one:
- TF2 highest through UI
- TF2 lowest through UI
- mastercomfig low
- mastercomfig low + hud playermodel
hud test:
- lighthud
- middlemann
- most of the others I tested before
hud test 2:
- same but no match hud
any other suggestions?
whats middlemann for
it's the name of a hud
the one with Hale doing the Anaconda pose
it seems to be a popular hud. But I'm not into huds so I don't know
i mean like does it usually have better or worse fps than the default hud
I heard people say its unoptimized, but I'm sure its just hate because of the afromentioned hale doing the anaconda pose
that is true
I feel like there are more complex huds like zees or weirdcore from the looks of it. They’re just complicated in the sense that they use lots of custom materials, which shouldn’t be absurdly laggy either way.
Could be interesting to test with/without friendslist enable if you do your testing online. Supposedly a lot of people used to get stutters that were fixed by disabling that element.
My bad meant to make that as a reply to @limpid panther
were going to kill and eat you
Don’t threaten me with a good time
vtable 2 lines up, leads to CTFFlameThrower::GetDeflectionRadius
hacker
UTIL_EntitiesInSphere
which leads to the box test
same as the leaked code
ficool weighed in on the airblast change
specifically this is in response to jane’s github post
yeah that makes sense
the question is where the issue is
i think i might have spotted an issue, but i need to confirm it
*i think i might have
spotted an issue, but i
need to confirm it*
oh btw i found a consistent issue but i need to find the map
it was the oot map
in the hole before the flag pole when climbing down death mountain the brush's geometry is so beyond fucked that it's not even visibly a room. looks like i played a n64 game with a bad emulator but it's in tf2
UB? rust would fix this
"no way to fix this, says users of only language where this regularly happens"
rust takes a longer to compile
this is an incredibly severe problem
this broken airblast installed malware onto my computer and used that to fuck my wife
ok how many iterations of the same code do you think it's possible to write before the compiler doesn't make a mistake in it? think of it like a translator.
what are you talking about?
UB doesn't mean the the compiler is making a mistake
it's a language ""feature""

you guys have been using the steam forums too much.
explain
actually, explain this first
i did not get an explanation :(
guess i'll just never understand
ok if it fucked up converting c++ to machine code because of some optimization or misread of data. it's like a translator that mistakes the spanish word for black as a english nword. that's as good as my bad explanation can go XD
No, optimizations like that aren't incorrect, they're just absolutely insane. The C++ specification explicitly allows them to do that.
Hydrogen peroxide is dangerous.....
wait is it possible to disable those specific optimizations in parts of your code but still have it used for most of it?
for some of them you can, but the problem is that C++ makes it incredibly easy to shoot yourself in the foot by default
it dissolves flesh.
yeah that makes sense
safe rust code just doesn't let you do stupid things that cause UB
have you made a game in rust?
no
by friendslist do you mean the steam overlay?
good cause i don't know anyone else whom has either
ok? what's your point
can someone explain why the top right works pre-ji but not the the center left
is it because the cone used to not be truncated? or something else
pre-ji i thought it had to be both to count.
ok now look at the image jim
am confuse, which one is center-left
Will be airblasted only on 32-bit post-JI and 64-bit post-JI
I'm not getting why that one wouldn't count pre-JI basically
pre-JI, the center of the bbox has to intersect the cone
ah yeah that'll do it lol ty
that sounds like a huge oversight and an unoptimized code
ah
The friendslist on the hud on the main menu screen.
@grim parcel
@grim parcel
@grim parcel
getchoo
is there somebody named KRUM in ur server
Krum
krum
yes there was
they just told me about u
I just joined
😭
why u leave
have you made a game in raw machine code? checkmate.
owned
my game would slow to like 10fps every time I paused lol yeah it sucks, dunno if that's fixed now
no it's definitely unoptimized, most of it is just slapped together haphazardly from other huds and I say that as someone who used it for a while
the fonts broke on it once and I reached out to the creator about it and like 20 minutes later he sent me a fixed version that just put multiple mentions of the fonts everywhere lmao, then he asked if it worked and it did and he shipped it
whatever happened to these cosmetics? haven't ever seen this and i started playing pre-ji https://wiki.teamfortress.com/wiki/Grand_Duchess
shit i wasn't meant to reply to that
Saw them the most back around when they came out.
there are probably thousands of cosmetics that you've never seen before
that's not exactly shocking
no what's shocking is no one can really use them because they don't drop.
people wear them when they drop and then most of them fizzle out from there, it's exceedingly rare for a cosmetic to stick around in popularity for a long time
*no what's shocking is
no one can really use them
because they don't drop.*
99% of cosmetics don't drop
they should
also voicelines shouldn't need a set. say if your equiped 2 or 1 of a set it'd count would be better
lmao
Eh. My inventory is already bloated with the thousand other things that drop. I don’t really need a bunch of hats dropping
Idling for the Halloween cosmetics just made Scream Fortress unplayable after the first few days.
The rate they drop at now is a “oh not a weapon or crate” surprise. Any more and it’d just be another sigh
i have over 50 cases, it i can convert even 1/3 of it to hats. it'd be a gain.
fuck even 1/4
I think I have 18 originals since my last crafting session. And like 6 modest pile of hats
i've had 2 pink paint cans randomly drop.