| smbhax [sys=PC; cat=Utility; reg=NA] |
| | | https://obsproject.com/
Free, open-source video recording and streaming software; I use it for recording video for YouTube and streaming to Twitch.
From my recording/streaming notes:
mic:
volume: -0
gain: 30.00 dB
noise suppression: "RNNoise (good quality)"
limiter: -1 dB, 60 ms
game:
volume: ~ 0 dB
Elgato volume: 76
limiter: -1.0 dB, 60 ms (default)
Type: Standard, NVENC H.264 (new), VBR, Preset: Max Quality, Profile: high, Psycho Visual Tuning, Max B-frames: 2
Audio: 320
1080p 16:9: 48K, 96K --> now use for all, since VBR will use less if not needed
1080p 4:3 (old): Bitrate 36000 Kbps, Max 72000
720 16:9 (old): 30K, 60K
720 4:3 (old): 22.5K, 45K
Streaming: CBR, 2200 Kbps, Preset: Quality, Profile: high, PVT, MaxB: 2 (only use 720p 16:9 :P)
Advanced: Priority High, D3D 11, RGB, 709, Partial,
(Network--check boxes to try to reduce frame drops--darn WiFi :PPP)
Enable Browser Source Hardware Acceleration, Disable hotkeys when main window is not in focus
desktop capture:
volume 0 (record at 100% desktop volume)
-1 db limiter
hotkey recording toggle ctrl+shift+Ins/ScrLk
streaming:
720p
stream CBR 2200 Kbps Quality high PST B=2
16:9 - transform -1,0
4:3 - -160,0
Framemeister on NORMAL2
Current game-specific recording notes (the top ones are being updated more or less constantly so these will be out of date by the time you read them; the lower ones are mostly very out of date because they're mostly games I haven't gone back to):
Fighting:
PS3 Virtua Fighter 5 Final Showdown -4 (1?:1? 2:2 was too loud) (720p)
PS4(X) King of Fighters '97 Global Match 0 (7:10:10-??? way too loud) (-228|1464x1080)
PC Capcom vs. SNK 2 0 ((45? 50 was too loud)) use SC
Capcom vs. SNK Pro 0 ((45? 50 was too loud)) use SC
Fatal Fury Special 0 (-15? -14 was slightly loud) (GC -240|1440x1080 4:3 59.94fps) window 1st
JoJo's Bizarre Adventure (flycast) 0 ((70)) use kof
King of Fighters 2000 (flycast) 0 ((55 OST)) (-241|1436x1008) internal 640x480
Marvel vs. Capcom 2 0 ((55)) use SC
SoulCalibur 0 (default 12:12 (50)) (GC) 1080p 59.94
Street Fighter III: 3rd Strike (flycast) 0 ((45?)) (50 was too loud) (-270|1380x1008)
Ultra Street Fighter IV 0 (1) (GC)
Adventure:
PS3 Wolfenstein 3D -9? (-7 was too loud--doors) 720p 4:3
PS4 Pinball FX 0 (?:?:?:?:? 5:7:7:7:10 too loud) (gamma 2.7) (3:4 4K)
PC Castlevania -2? (-1|1516x1080) t/b overscan 16 -- NTSC -- NIGHT LIGHT OFF! vsync off?
Occasional:
PS2 Virtua Fighter 4 0 (16:16) (-325|1276x888)
PS4 Gran Turismo 7 0 (game Menu BGM 4, Menu SFX 5? 7 kinda loud, Race SFX 4? 7 way too loud)
PC Double Dragon 0 (-6?) (1440x1078 use FFS) (MAME: Sliders - H/V Stretch 1.068, OPM .5, ADPCM 2)
Gunbird 2 -2 (1? 2 was too loud) (GC -1164|1512x2160) (vsync off)
TrackMania Nations Forever -8? (-4 was way too loud) (default (max)) (DC) vsync on
WCW/nWo Revenge -9 (-6 was too loud) (30fps GC -240|1440x1080 no AA) vsync off
Puzzle!
PC Dig Dug 0 (-13?) (-1110|1620x2160 - 3:4 4K) (DIPS - M diff)
Mr. Do -25? -22 was too loud (GC -44|1372x1080) t/b overscan 7/9
Ms. Pac-Man 0 (-13? -10 was too loud) (-1110|1620x2160 - 3:4 4K)
Tetris -2? (0 was too loud) (GC 1152x1080) NO overscan, DEFAULT ASPECT vsync off
Beat-em-ups I could play:
PS3 X-Men 0 (1080p)
PS4(X) Arcade Archives Crime Fighters -9? (4:3 -240|1440x1078)
Arcade Archives Zero Team -6? (-244|1432x1070) Big-O cumu.score: 18 x_x
PC Double Dragon Advance -6? (1620x1080) (was mGBA; Check sound in VBA-M) vsync on, 59.94 fps
Double Dragon II (SGB) -3? (GC 1200x1080 -- DEFAULT ASPECT)
Final Fight 2 -2? (GC -1|1408x964 -- NTSC)
Water Margin -3? (DC -256|1408x1002) 60 fps
Other games I might play:
NG(X) King of Fighters 2000 0 (-337|1232x896; thumb 885x498, fuzz 19)
C64 Way of the Exploding Fist +6? (PAL 50 fps!) (window; pt scale 1620 x 1080)
PS3 Daytona USA 0 (2:2) (1080p)
PS4(X) Arcade Archives Burger Time -12? (-840|1604x2160)
Arcade Archives Kangaroo (use AA Burger Time)
CA2S - Sat Nite Slam Masters 0 (5:5--make louder!) (4:3)
PC Bonk's Revenge -7? (GC 1240x1080) Border on, SGB2 A input, t/b overscan 7/9 -- DEFAULT ASPECT!
Capcom vs SNK Pro -0? (GC)
Final Fantasy -7? (-4 too loud) (GC -1|1408x1080) t/b overscan 8 -- NTSC -- NIGHT LIGHT OFF! vsync on
Guilty Gear X 0 (flycast) (BGM Medium) ((80?)<--too loud even w/o music) use kof
Jojo's Bizarre Adventure (Dreamcast) -0? (GC)
Mario Bros Classic Serie 0 (GC 1584x1031 50fps -- SET PAL; PAL overscan not working)
Pac-Man 0 (-13?) (-1110|1620x2160 - 3:4 4K)
Street Fighter II 0 (-9? -6 was too loud) (DC 4:3)
Street Fighter III 2nd Impact 0 (-10? -8 was too loud) (mame DC 59.94fps 1080p)
Zachtronics Solitaire Collection 0 (m20?) vol in \Documents\My Games\
Night light (60) OFF 'cept for Scuj
Other games I could play:
PS1
(x)Darkstalkers 3 -6 (game 2:3? 2:4?) SET NARROW SCREEN (1144 x 894)
PS2
SFAC - 3rd Strike 0? (game -9:-9) (-364|1196 x 894) (game 94% width)
(x)Taito Memories II: 1st Volume (Joukan) 0 (game 2:2?) (record 1276; 1274 x 896 max)
(x)Taito Memories II: Last Volume (Gekan) 0 (game 2:2?) (record 1276; 1274 x 896 max)
PS3
(x)Joust -12? (1080p)
(x)Street Fighter III: 3rd Strike Online Edition 0 (game 25?) (720p) (4:3 894 x 672)
(x)Virtua Fighter 5 0 (game 4:13?) (1080p*) *disable 720p
PS4
Arcade Archives Darius -6? [Arcade Archives generic is -16?!]
Capcom B'EUB 0 (5:5:5?) (-189|1540x1080) CUSTOM--NOT 4:3
Marvel vs. Capcom: Infinite 0 (game 3? all)
Ultimate Marvel vs. Capcom 3 0 (game 4? all)
Pac-Man Museum+ (Arc Arr) -6? (100:50:50?) (-1146|1548x2160)
Pac-Man Museum+ (CE) 0 (100:50:50?)
PC
Angband +11? winamp (4%?) (hengband ost) record at 30fps
Batman Returns -2? (dims as FF -- set NTSC)
Combatribes -13? (dims as FF -- SET NTSC)
Crazy Taxi -1? (2:2) (DC)
Death Brade -2? (dims as FF -- SET NTSC) B,Y,A
From Below 0 (use Tetris NES) no overscan, SET DEFAULT ASPECT
Gourmet Warriors -6? (-1|1408x969 -- set NTSC)
Grand Theft Auto V 0 (10:0) (DC) vsync on
King of Fighters 2000 0 (cmd: mame kof2000 -nofilter -bios unibios40 -volume -10)
Mario Bros 1986 0 (GC 1316x1053 -- SET NTSC)
Mushihimesama 0 (20:20?) (720,960center|1440x1920 (3:4 960p)) NIGHT LIGHT(45)=65
PAKU PAKU -1? (16:9) http://www.asahi-net.or.jp/~cs8k-cyu/browser.html
Tetris -3? (GC 1200x1080)
Border off, SGB2 A input, t/b overscan 0 -- DEFAULT ASPECT -- NIGHT LIGHT OFF?
Tetris 2K21 0 (plus4emu: Video Quality 0, Contrast 1.102; Volume 0.198?
Machine: SID & Blaster emu
shortcut w/ -prg [filename]
(WC -360|1192x750 50 fps)
WWF No Mercy (use WCW/nWo Revenge)
Games I could play but probably won't:
PS1 Castlevania Chronicles -8? (-376|1160x928--shift)
Namco Museum Vol. 3 - Ms. Pac-Man -17? (-379|1160x956)
PS3 PC Genjin 2 -11? (-264|1364x1080)
PS4
Atari 50 (Asteroids) 0 (game 10/10?)
Arcade Archives Dig Dug -16? (-767|1604x2160)
Arcade Archives Double Dragon -9? (-348|1224x908)
Arcade Archives Double Dragon II -7? (-253|1412x1045)
Arcade Archives Pac-Man -19? (-767|1604x2160)
Arcade Archives Sea Fighter Poseidon -6? (1436x1080)
Arcade Archives Shusse Ozumo -6? (-1068|1704x2160)
Arcade Archives Tecmo Knight -11? (4:3 -240|1440x1078)
Arcade Archives Vendetta -6? (4:3 aspect)
Kunio-kun: The World Classics Collection 0 (20%/S?) (-287|1392x1080) DDII
(x)Granblue Fantasy: Versus 0 (game 20%?)
Last Blade 2 ? (with interframe blending)
Mr. Driller Drill Land 0 bit loud (1:1?)
Sega Genesis Classics -10? (4:3 (1440x1080))
Street Fighter 6 Demo 0 (7)
PC
Akumajou Densetsu -2? (see Akumajou Dracula) NIGHT LIGHT OFF
Arcade Game Series Ms. Pac-Man -3? (-1032|1776x2160)[gamma 1.5]
Bagman Comes Back -9? (Game Capture -973|1892x2160)
Bagman Strikes Back -9? (Game Capture -190|1540x1080)
Dig Dug Revival -18? (GC -1056|1728x2160)
Castlevania II -2? (1316x1080) probably should've top/bottom overscanned -16
Fire Pro Wrestling World -11? (defaults) (GC)
Funaki Masakatsu Hybrid Wrestler: Tougi Denshou -8? (1408x1080 -- set NTSC)
switches to 1280x980 at wrestling! manually toggle back (too tiny)
Lady Pac -15? (GC -1080|1680 x 2160)
Marathon, Marathon 2, Marathon Infinity 0 (bit loud : P) (GC)
Megami Tensei Neuroheroine 2 -20? (-23 in 5) (defaults)
(switch to BLACK win bg color!)
(Capture Method: DXGI (fullscreen) but actually want 720p window)
Pro Wrestling -5? (Game Capture 1316x1080 -- SET NTSC)
RandomPac -15? (GC -1080|1680 x 2160)
Rushing Beat Ran -8? (use FF (1408x936) -- SET NTSC)
Street Fighter III 3rd Strike 0 (-10? -8 was too loud) (mame DC 59.94fps 4:3)
Super Game Boy (2) borders (Game Capture -1|1232x1080 -- SET DEFAULT ASPECT)
top/bottom overscan 7/9
(x)Tetris.com 0 (70/20?)
Tutankham -1? (Game Capture -973|1892x2160)
Undercover Cops 0 (-14?) (DC -240|1440x1080 - 4:3 profile, FFS Scene) (OST Copyright claims!)
worldsbiggestpacman.com 0? (Tetris Classix4 profile w/
4x window capture or something, Google Chrome) (2x prob okay...)
Wrestling Empire 0 (50:50?) (Speed=Smooth) (Nvidia: 8x AA, no VSync)
WWF Superstars -3? (use DDII -- SET DEFAULT ASPECT) TYPE A
WWF Wrestlemania Challenge 0 (Game Capture -41|1276x1044 -- SET NTSC)
PCE Double Dragon II 15? (-361|1168x952) Lev Select: title: OXOX,Run
Out of rotation:
dreamcast (expanded)
Capcom vs SNK 2 -? (1278 x 952) (ai weak, bosses ugh, movement clunky)
Capcom vs SNK Pro -14? (record 1280??; 1276 x 950)
Guilty Gear X -21 (game medium) (1278 x 952) (AI esp bosses not fun, can't beat w/o fireball)
JoJo's Bizarre Adventure -1? (-391|1156 x 894)
SoulCalibur -6? (-2 was too loud) (1?) (-343|1276x952)
Virtua Fighter 3tb - 58 (chng Guard to Punch, 2nd ESC to Kick) (ai can't block sweeps)
ps1
Darkstalkers -? (slowdown)
Marvel Super Heroes vs. Street Fighter - 54 -8 (game 3:3) highlight "OPTION", R1, Circ, L, Tri, Tri
Street Fighter Collection 2 - WW -? (game default) cont. to avoid game over red screen strobe! (puzzle-like AI)
X-Men: Children of the Atom - 63 -4 (game default) (slow? some flashy; not many chars like)
X-Men vs. Street Fighter - 54 -4 (game 4:4--default???) (wht scrn flsh w/ ev special move hit)
ps2
Capcom Fighting Jam/Evolution -15? (game 1:1) (1278 x 896)
Fu'un Super Combo - Kizuna Encounter - 54 -18
Garou Densetsu Battle Archives 1 - FF3 - 54 -18 (side-dodge spam breaks AI)
Garou Densetsu Battle Archives 2 - RBFF - 54 -4
Garou Densetsu Battle Archives 2 - RBFF2 - 54 -6
Hokuto no Ken - 63 -6 (game 24) (too easy--cLK-cHK-air-combo)
King of Fighters '94 Re-Bout - 53 -10 f (pao pao cafe intro flashy; also korea lightning, g over,etc)
king of fighters NESTS - 2001 - 50 -4 (MAX super meter red/orange rapid flash 8p)
king of fighters NESTS - 99 -? (slight flashes, boss sucks, eh)
king of fighters 2002 - 59 -8 (lightning stage; practice only--but too easy)
King of Fighters 2003 - 51 -24 (difficulty curve etc)
King of Fighters '98 UM - NEOGEO MODE - 55 (game 1:1) (not as fun as 2000, confusion)
Midway Arcade Treasures 2 -15? (res toggle at game switch zaps meister)
Neo*Geo Battle Coliseum - 53 (game 1:1) (bleh boss, balance)
Sengoku Basara X - 55 -4 (game 2 all) (not fun)
SNK vs. Capcom: SVC Chaos - 53 -10 480p X+/\ p2 - supers too flashy
Street Fighter Alpha Anthology - Upper - 63 -7? (game 7) - flashy, just a bit eh
Sunsoft Collection - Waku Waku 7 - 53 -8 (huge input delay)
Taito Memories II: Last Volume (Gekan) -15 (game 2:2) (1278 x 892) Warrior Blade, Bubble Symphony
Tekken 4 -8 (game 4:4) (1278 x 896)
Tekken 5 - 58 (game 8:8:8) - AI weak to fast attacks
Vampire Darkstalkers Collection -15 (game 8?:8?) (V1 playable but ugh; play others in Training to avoid KO flashes) (1278 x 896)
Virtua Fighter 10th Anniversary -? (game 8:4--doesn't save; health bar flash; oppressive w/ no sidestep to avoid edge)
Virtua Fighter 4: Evolution -11 (game 6:6) (1280 x 894) (VF5 kinda better)
World Heroes Gorgeous - WH - 53 -6 (approx)
ps3
Castle of Illusion Starring Mickey Mouse (Trial) 0 720p
Dark Awake - 54 -4 (game 4:4) (1080p & xrgb 1080/1440 scene) (flashy)
Darkstalkers Resurrection -15 (game 50:50) (720p) (894 x 672) (red KO flash)
Dead or Alive 5 Last Round - 63 ? (game 45:45:45) (720p) (camera hurts eyes, counters hurt wrist)
Fighting Vipers - 63 -4 (game 4:4) (1080p & xrgb 1080/1440 scene) (eh, iffy clothing strip)
Final Fight: Double Impact 0? (2?:4?) (1080p)
Giana Sisters : Twisted Dreams (DEMO) 0 (default) 720p
Golden Axe -18? (1080p)
Guilty Gear XX Accent Core Plus -? (game 20:20:20) (720p) (vs cpu but repick) (big sparks, counter hit char flashes red)
Hungry Giraffe 0? (1:2?) (1080p) some horrible purple screen flash FX
Jet Set Radio Demo -18 (default gar; doesn't save) 1080p 30fps
Jojo's Bizarre Adventure HD Ver. Trial 0 (3:3:3:3) (720p)
Marvel vs. Capcom 2 0 (game 3:3?) (1080p)
Marvel vs. Capcom Origins - MSH - 63 -4 (game 25?:25?) (720p) (894 x 672) (Power activation hits are red)
Marvel vs. Capcom Origins - MvC - 63 -18 (game 25?:25?) (720p) (894 x 672) (eh)
The Next Space -17? (1080p)
NEOGEO STATION The King of Fighters '94 0 (27%) (1472x1080)
NEOGEO STATION The King of Fighters '95 0 (27%) (1472x1080)
NEOGEO STATION The King of Fighters '96 0 music a bit loud (27%)
(1080p) KO RED FLASHES
Ozma Wars -24? (1080p)
Pac-Man CE (PS minis) -13 (too loud) (1080p)
Pac-Man Museum -10 TOO LOUD! (720p)
R4 Ridge Racer Type 4 -4 (game 6:5) (4:3 1080p)
Sonic Adventure 2: Trial -17 (kinda loud music sometimes) 720p
Tekken - 63 -8 (1080p)
Tekken 2 - 63 -8 (1080p)
Tekken 5: Dark Resurrection - 63 (game 10:10:10) (1080p) (AI spammable)
Tekken 6 - 63 (game default) (720p) (slower than dr, just not as light and fun)
Tekken Tag Tournament HD - -13 (game 7:7) (1080p) (I'm bad at it ; )
Tekken Tag Tournament 2 - 63 -2 (game 8:8) (720p) (feels generic)
Touch Down Fever -14? (1080p)
Virtua Fighter 2 - 62 -2 (game 3:3) (1080p)
(duds char u/d, rank st+GPK, xprt chara d2,u2,l2,r2)
ps4 (expanded)
ACA NEOGEO 3 Count Bout -13? (1536? x 1074)
Arcade Archives 64th Street -10
Arcade Archives Shanghai III -11 (-247|1424x1080)
BLURRY TILES=EYESTRAIN! :P
Arcade Game Series: Ms. Pac-Man -10 (-1120|1600x2160)
Capcom Fighting Collection -4 (game 80%)
Darius (Darius Cozmic Collection) 0 (45)
Dead or Alive 5 Last Round - 63 -16 (game 60) (camera hurts eyes, counters hurt wrist)
Fighting EX Layer - 62 (game 3:1:2:3)
Garou: Mark of the Wolves PS4 -? (game all max) (only like fire guy in survival; grades stop story ending)
Injustice 2 -4 (game default 100 all) (1080p) (copyright claims)
Pac-Man Museum+ (PM) -10 (100:50:50) (-1146|1548x2160)
Teenage Mutant Ninja Turtles: The Cowabunga Collection -3
Tetris Effect: Connected 0 (3) (-892,-285|1080x1440) tiny blurry blocks
Tekken 7 - 63 -2 (game -1 pip main) (feels generic)
Virtua Fighter 5 Ultimate Showdown - 60 (game default)
Worms Battlegrounds -4 (game -2/-3/-2?) (janky : P)
Worms WMD -4 (game -2/-4/-3) every kind of eyestrain
Yakuza 6 -10? FZ -4? others (1540x1080)
PC
Blue Wish Desire -16? (Shot=0,Bmb=7,Brst=5,FA=2,Paws=6; J2K: Btn2=z(T),Btn3=w,Btn4=c(T)) (720 x 960) Kinda 8o-gross bullet patterns/colors
Blue Wish Resurrection Plus -16? (960 x 960)
Death and Return of Superman -2? (Game Capture -1|1408x1066 -- SET NTSC)
Forza Horizon 4 0 (9) (DC--according to fh site) vsync on
Mahjong Solitaire Refresh -1? (3:3?) RETURNED--slightly blurry tiles
Sensible World of Soccer 96/97 2? loud-ish (4:3; 480p x 2.25 point scaling; 1440x900)
Star Trigon 0? (~45/~75?) (windowed 800 x 600) (cute but kind of aggravating)
Trackmania 0? (default (max)) (GC) makes me nauseous DO NOT PLAY
Worms Armageddon 0 (game ~70%?) TURN OFF NIGHT LIGHT |
|
|
| | | In the past I've had jerky capture after switching Scenes in OBS, so after a Scene switch I close the program and restart.
I record in .flv, which OBS then remuxes to .mp4; .flv doesn't get broken if the recording cuts off unexpectedly.
When recording from OSSC / Framemeister, for more-correct video I intialization (again mostly washed-out issues) I've found it's best to leave the recording-output OSSC off, turn everything else on and run OBS, then turn on the recording-output OSSC once the video mode I'm recording is showing on my monitor from the other OSSC.
Then I go OBS toggle the "HDMI Color Range" settingfor the Elgato HD60 (see entry 1635) back and forth between "Standard" and "Expanded" through OBS, because sometimes it initializes via OBS at a setting other than the one indicated, which would result in washed-out or overblown video. |
|
|
| | | Record Flycast (see entry 1615) emulated Dreamcast video at 59.94 fps forced by RivaTuner Statistics Server (see entry 1630):
|
|
|
| | | For anything through the Elgato Game Capture HD60, that says it uses 59.94 fps rather than 60 fps exactly: https://help.elgato.com/hc/en-us/articles/360027952432-720p-video-at-59-94-or-60-fps-with-Elgato-Game-Capture-HD
Trying to test with KOF '97 Global Match's (see entry 1531) flickering shadows, did not see a difference between recording at 60 and at 59.94:
DigitalFoundry lists their console videos, even for current gen, as recorded at 59.94. |
|
|
| | | Mesen has a setting for emulating at integer framerates
Video > General > Enable integer FPS mode (e.g. run at 60 fps instead of 60.1)
I use that setting and record at 60 fps. (Native NES and SNES framerate is said to be 60.1; GB and GBA 59.7.)
NEOGEO via MAME is ~59.19; other arcade games via MAME may vary, Ms. Pac-Man and Dig Dug are 60.6, for instance; Double Dragon is 57.4. |
|
|
| | | PS3 60 vs 59.94 comparisons:
NEOGEO STATION The King of Fighters '96
Marvel Pinball
PS4:
ACA NEOGEO Art of Fighting
Pinball FX
PS2:
King of Fighters 2003
N64 in Mupen64Plus:
WCW/nWo Revenge (~30 fps game)
I can't see any difference so I'm going to leave them at integer recording fps. Shadow flicker in MPC-HC in Windows is definitely less pronounced 60 fps videos. Mupen has frame stutter (visible in frame-by-frame) at 29.97. |
|
|
| | | | So, the main exceptions to using integer framerates, for me, are Flycast (59.94, forced with RTSS) and MAME (varies per game). |
|
|
| | | General note to self regarding gamma adjustment:
The gamma 2.2 images I've always used to calibrate my monitors for graphics work (making sure Photoshop is also set to gamma 2.2 in its options) are from
http://www.photoscientia.co.uk/Gamma.htm
specifically the pop-up window
www.photoscientia.co.uk/2point2.htm
I'm attaching those photos here as a back-up, with the added tag "ps_." Wikipedia also has a pretty good test image that I'm also uploading here, with the added tag "wp_," from
https://en.wikipedia.org/wiki/Gamma_correction#Simple_monitor_tests |
|
|
| | | Mednafen Neo Geo Pocket settings:
Profile: 1136x1080
Scene: GC -390 |
|
|
| | | | Window Capture captures ReShade's LumaSharpen, but ALSO captures the cursor, even when I have Window Capture's cursor capture option unchecked, dang. So for games in which I'm using ReShade--currently: VF5FS and Demon's Souls in RPCS3, and Fire Pro Wrestling World)--pretty much have to use Display Capture, I guess. |
|
|
| | | Added Compression and Expander filters to my mic audio filter chain to cut down on volume spikes:
(JBL Quantum 100 headset)
Gain: 28 dB
Noise Suppression: "RNNoise (good quality)"
Expander: Ratio 4:00:1
Compressor: Ratio 3.00:1, Threshold -5 dB, Output Gain 2 dB
Limiter: -1 dB, 60 ms |
|
|
| | | | Took out the Expander, felt like it caused a bit of an interruption when I would trail off at low volume and come back. |
|
|
| | | I was noticing audio dropout on both mic and game audio recently, when they would both be trying to play during Elden Ring sessions, for instance here
|
|
|
| | | It seems to have been caused by a feature OBS's log analyzer says was added to "Windows 10," called HAGS (Hardware Accelerated GPU Scheduling); OBS's analyzer says HAGS is known to cause problems and recommends turning it off; I did and that seems to have solved the problem.
It SEEMED like a feedback loop where sound from playing a recording in the MPC video player while OBS was open would route that sound into OBS's Mic channel for some reason, phantom sound would continue to register in the Mic channel even after stopping the MPC player, and then when recording, noise suppression would kick in on both Desktop (game) and Mic sound when both were going at once.
With HAGS turned off, MPC playback audio still registers in the Mic channel (I'm playing the audio back in my headset, which does not normally bleed any sound to the mic--and doesn't do so in Windows' sound test as far as I can hear), but doesn't continue to spike phantom readings in the Mic channel after playback has stopped. Hopefully that will mean I won't get mic/game audio interference/dropout/suppression in recordings now.
In the video I'll post next, I toggled noise suppression settings before recording, which seemed to stop the feedback and I think gave a clean recording. At the end, I spliced on a little bit about finding and disabling the HAGS setting, which I found out about and did after the game session. I'll time jump to that part. |
|
|
| | | | (Oh I found this out through OBS' nifty log uploader/analyzer feature; it also pointed out that I'm not running OBS in Administrator mode, which can enable it to capture some games in Game Capture mode which otherwise could not be captured that way; I don't have that problem with anything I'm playing currently though so I guess I'll leave Run as Admin unchecked to have myself having to okay that "Allow this program to make changes to your computer" or whatever window that comes up when running the program as Admin.) |
|
|
| | | | I could only find the previous two video sessions as having definitely suffered from the sound drop-outs; this roughly coincides with my doing Windows 11's big 2025 features update thingy in Windows Update, so I wonder if that may have toggled HAGS on. |
|
|
| | | OBS Studio log analyzer and about disabling HAGS to fix audio dropout in Windows 11:
|
|
|
| | | Actually I dunno if it was HAGS at all 'cause the problem showed up again. Basically I think it's due to bleed-over through the headset due to the noise suppression filters getting confused by the high amount of static noise from the high gain boost I have to use to get the JBL Quantum 100's mic volume up in Windows 11 now.
If I turn on Windows' mic default effects and the noise filter in JBL's sound app, I can turn off the OBS noise filter and the problem is kinda solved, but voice sounds pretty bad.
Compressor filter was munging the sound quality a bit and may have been contributing, I did have it adding 2 dB gain, with 2 off the main gain, from 30 to 28. Turned it off and put main gain back to 30. If I find the desktop sound feedback loop (from Elden Ring) gets into the mic again I'll drop main Gain to 28. Not going back to compressor though. ; P |
|
|
| | | ARGH I'd been audio sampling in OBS at 48000 Hz but had the mic sampling rate set at 44100 Hz in my computer's "Realtek Audio Console" and Windows' Microphone Sound Properties (which calls 48K "DVD Quality" and 44.1K "CD Quality"). And I see YouTube now recommends "48kHz" https://support.google.com/youtube/answer/6039860?hl=en sample rate uploads.
I think maybe it used to recommend 44.1? But probably not. Anyway my settings were all screwed up. ; P
Testing at 48 now, it does sound cleaner and more robust. So obviously that's what I'm switching everything to. |
|
|
| | |
0:27 - OBS Studio audio sample rate setting
1:01 - Windows microphone sample rate setting
2:22 - Realtek Audio Console microphone sample rate setting
4:14 - YouTube recommended sample rate
That YouTube "Encoding specifications for music videos" page is https://support.google.com/youtube/answer/6039860?hl=en |
|
|
| | | Noise Gate filter doesn't really solve the problem with the desktop (game) sound somehow feeding back into the mic in OBS, leading to the Noise Suppression filter filtering the vocals, causing voice drop-out. Seems to be interaction between high Gain and thus loud background noise/static, loud desktop/game sound, and Noise Suppression filter.
Noise Gate can technically stop it if you jack the Open Threshold value high enough, but it has to go quite high--minus 18 dB, at least, and that's into yellow range!--to catch all of it, and that would be cutting out lower-range speech, too. |
|
|
| | | | The problem with the desktop (game) sound feeding into the mic seems most likely to be due to the sound traveling through the headset itself, from the ear cups to the microphone; it gets slightly worse as applied gain is raised in the Gain filter, and--to a lesser extent--as sound playback volume is increased in the headset, but is unaffected by moving the mic closer to the opposite ear cup during sound playback (except maybe very very close, didn't try that), or by clicking the volume icon under Desktop Audio in OBS to cut pickup. It is also unaffected by putting Desktop Audio and mic on separate tracks in OBS. |
|
|
| | | The feedback/dropout loop with the Quantum (at 30 Gain) in Tetris:
|
|
|
| | | | Now that I have a 24-bit mic with the ATH-M50xSTS-USB, I'm switching Audio Encoder from "FFmpeg AAC" to "FFmpeg PCM (24-bit)." In quick desktop test, waving videos around, this increased mp4 video file size by about 50%. Sound is much more rich. |
|
|
| | | [Didn't save this video because I'd recorded it in AAC audio which is only 16-bit. Anyway it was just me doing my "Hello this is Elden Ring" intro w/ the JBL Quantum 100 and then the ATH-M50xSTS-USB, and they sounded remarkably similar--indistinguishable, to my ear. M50x would've sounded markedly richer if recorded at 24-bit, of course.]
Both at 48 kHz sample rate (M50 goes up to 96). Quantum only does 16-bit; M50x is set to 24-bit.
I had the mic boom arms straight out, not bent.
Quantum was at 30 dB added Gain (via filter) in OBS. M50x was at 21, that seemed about equal in volume to 30 (which is the max) on the Quantum.
Both at 100% volume in Windows Sound Settings. Windows audio enhancements Off for the Quantum mic; the ATH mic doesn't have the option.
Quantum also at max plus 20 dB in my laptop's Realtek Audio Console, the sound card controller. The Quantum is an 1/8" (3.5 mm) audio jack analog mic, while this version of the M50x is USB.
RNNoise noise suppression filter enabled in OBS, as well as a -1 dB, 60 ms Limiter.
I have the stock foam mic cover on the Quantum; no cover--it doesn't come with one--on the M50x. I ordered some "L6"-sized foam covers that fit its mic, according to some Reddit comment: https://www.amazon.com/dp/B016WOYASA ; they haven't arrived yet, but with the mic kept straight it doesn't sound like it gets any wind from my mouth, so I may end up going cover-less to maximize capture volume; I didn't catch any wind-y sound on these recordings with the mic straight, whereas it was immediately obvious--and pretty bad--when I bent the naked mic "near the corner of your mouth" per the pdf instruction book ( https://docs.audio-technica.com/all/ATH-M50xSTS-USB-UM-EN.pdf ). The Quantum never got THAT windy-sounding in my bent-mic-arm tests with it, but it also can't really bend as near the corner of the mouth and retain its shape there--plus it does have a modest foam wind cover. (The M50x is a condenser mic; the Quantum is a more typical headset mic, which I suppose means a dynamic mic.)
Although on the plus side, the M50x's arm is easier to bend and holds its shape easily, whereas the Quantum's (mind you the Quantum is a budget headset; the M50x is rather expensive) doesn't really hold anything beyond a mild bend, so it won't stay as close to my mouth.
While I'm at it, I'll mention that the M50x does clamp harder over the ears than the Quantum; not really painfully, but enough to leave a visible impression around the ear, and my ears definitely feel more relaxed wearing the Quantum, even with the surface worn off its old (I've had the Quantum for a few years or so, just got the M50x) leatherette earpads. I have the stock "mesh" pad earpads ("for increased comfort") on the M50x; it also comes with a pair of alternative "leather material" pads that "stress audio quality and sound isolation." I thought I have kind of a large head but just one notch up from the smallest headband size seems to fit best on the M50x.
Bent-arm, I was able to remove 10 dB of artificial filter Gain on the Quantum, and 15 on the M50x. This is potentially important because I'm probably having to move on from the Quantum because I've been getting some sort of sound/filter feedback effect in OBS in which game sounds through the headset somehow start spiking volume on the mic, visible in OBS' volume displays. I think this is due to some sort of interaction between the Gain filter, when set to a high amount of added dB, the noise suppression filter (still happens with the Speex "low CPU" noise suppression filter), and background noise--which high Gain boosts.
With Noise Suppression off and 21 dB Gain, the M50x's mono mic volume bar shows about -43 dB from background noise in OBS. The Quantum shows about -36 dB background noise at 21 dB Gain, and about -26 dB at 30 dB Gain, which is where I have to put it to equal 21 dB Gain on the M50x.
Incidentally, the Quantum, which I'm using plug-n-play, w/o JBL's sound app, only has "2 channel" options in Windows Sound Settings, & always shows two volume bars in OBS, even when Mono is checked in OBS' Advanced Audio Properties. The M50x has only "1 channel" options in Windows Sound Settings, and shows correctly as mono (1 bar) by default, regardless of that box.
That level of background noise seemed to be what triggered the feedback problem--which would result in game sounds somehow in the mic input, and Noise Suppression starting to suppress my voice, causing vocal dropout. My best guess so far is that the noise was from sound vibrating from the left ear speaker through the headset itself and into the mic.
Anyway, the goal is to keep Gain low, hopefully avoiding the feedback & drop-out. The M50x is off to a decent start, requiring 9 dB less Gain with the mic arms kept straight. If the feedback/dropout comes back, I'll have to try bending the mic to the corner of my mouth & using a wind cover; that'll mess with the clarity of the input with wind & muffling, & make keeping the volume input constant trickier since it's harder to judge the bent position than the straight one, but would greatly reduce the artificial Gain needed & almost certainly avoid the feedback and subsequent voice drop-out problem. Hopefully I won't have to resort to that but we'll see how the M50x holds up straight-arm in actual gameplay recordings. |
|
|
| | | | YouTube currently recommends https://support.google.com/youtube/answer/6039860?hl=en lossless 24-bit FLAC or Linear PCM; OBS only seems to offer 16-bit FLAC, so PCM (hopefully Linear? Google AI says yes 'p') it is. |
|
|
| | | | Windows file details say the data rate for 24-bit PCM vs (16-bit) AAC in two quick test recordings I did is 2257 kbps vs 270 kbps. |
|
|
| | | Possibly the sound feedback problem had got worse in the past few months because as the weather gets colder, my apartment's two wall heaters are on more, and their fans add some background noise. The dynamic (I think) mic in the JBL Quantum 100 doesn't pick up that much background noise really but maybe it was just enough additional general ambient to push it over the brink.
The condenser mic in the M50x is more sensitive so eh I guess I'm going to be passing the winter playing games with the heat off. ; D
Hm in the summer the fridge cooling pump or whatever kicks in more... ; )
Not confident I can use the Noise Gate filter for that stuff 'cause for instance the wall heater in this room is picked up louder on the M50x than my softest speaking voice. |
|
|
| | | I did used to turn off at least the nearer heater when I was using the Snowball, come to think of it. Can't remember if I turned off the heater in the other room too, but I was pretty paranoid about background noise with that thing so I may have.
Not going to unplug the fridge though. :P |
|
|
| | |
