uhh is this normal?

[signal wind]

yo my obs kinda cooking my gpu

[drifting jackal]

https://ctt.cx/output high fps section

[signal wind]

yeah i meant like for my gpu usage to be 20%

[drifting jackal]

Yes

It uses your GPU

[signal wind]

so is it the fps or the codec?

its 60 fps btw

[latent cliff]

its your settings.

[drifting jackal]

It’s the encoder

The encoder uses your GPU encoder

[latent cliff]

hes recording in 650 Mbps

lmfao

[drifting jackal]

Your settings also suck, yeah

Not very efficient

Fix them https://ctt.cx/output

[latent cliff]

specifically this stuff

i dont know why it says p4 because id personally not really go above p3 and usually im at p1 in the first place

but if youre recording in 60 fps then p4 is probably fine

also while recording at 600 fps im only getting 3% gpu usage soooo

maybe thats just FFMPEG VAAPI stuff tho

[signal wind]

can u share ur settings if u dont mind?

i use lossless for clipping games like bo2 for editing any i just use replay buffer for val to get normal clips

[latent cliff]

my encoder changes the settings

[latent cliff]

dont use lossless

use visually lossless

[latent cliff]

this is what it looks like for em

[signal wind]

oooooh i see

[latent cliff]

just do this

also use limited colour range

not full

[signal wind]

good now?

take 10% now instead

[latent cliff]

yeah

[signal wind]

is it normal for the bitrate to be that low?

[latent cliff]

depends on your scene and what youre recording

also why is it 1584x1080

[signal wind]

val res

[latent cliff]

interesting

if you want true visually lossless, cq 16 will look slightly better

[signal wind]

scene as what? i use game capture btw

[latent cliff]

what youre recording

more action = more bitrate

[signal wind]

i just use lossless and record bo2 using dolly cam

[latent cliff]

what is dolly cam

also wdym u use lossless

heres a comparison of actual screenshot from game, cq 16, cq 18

actual game has better colour accuracy but theres is basically no difference between cq 16 and 18

recording in 16 isnt a bad idea tho since you have more headroom when reencoding stuff

CQ 18 is 6,319,845 bytes
CQ 16 is 6,332,704 bytes

the main reason for that difference is probably just because the scene isnt IDENTICAL

[signal wind]

@latent cliff 1 sec pookie in a game sorry

[signal wind]

thats what i use for bo2 and editing clips

also u use p1? i found out p4 makes gpu usage higher

[signal wind]

a mod editors use basically free cam

but with postition keyframes ( like in blender )

[latent cliff]

i use p1 equivalent because otherwise i wouldn't be able to record in 600 fps otherwise

for 60 fps p4 is fine

also you really dont need to be this focused on a 2% gpu usage difference

p4 is fine for your use case since you get fine performance with substantially smaller files

[latent cliff]

dont det rate control to lossless cuh

set it to constant qp

and set ir cqp 18 or 16

[signal wind]

its 10%

i get 0.2% on p1

nvm will just stick to p4

with cq 16

[plush ledge]

how can u record at 600, i have an i5 14th gen and a 5060 and i can barely record at 360

[latent cliff]

because i have a better gpu?

the 5060 is a midranger

600 fps at 1080p isnt even anything crazy tbh

[latent cliff]

you could probably do better

maybe like 480 fps