player info

MinG3012

Hong Kong

Global Ranking

#45,746

Country Ranking

#603

Medals

213

6,198

Total Play Time

42d 4h 15m

384

210

345

Ranked Score: 14,497,117,626

Hit Accuracy: 91.87%

Play Count: 77,410

Total Score: 129,871,707,477

Total Hits: 16,537,748

Maximum Combo: 6,070

Replays Watched by Others: 58,872

101

Joined

Last seen 25 days ago

Plays with Keyboard, Tablet

Contributed 41 forum posts

Posted 14 comments

s͂ẻοụleͫgeͫn͈̙͔̹̦͠d

FayeurS ♥

Overranked Player

@ming3012

icryeverytime.com/

me!Recent Ranks Medals Beatmaps Historical Kudosu!

me!

semi semi afk
nah cant fix fingerlock + doubletap issue

Tablet: CTL-672
Pen Modifier: Tennis grip + 2 AAA battery + Kapton tape
Keyboard: Ducky Zero 2108S Cherry Red

🔥🔥🔥🔥🔥TOO HOT🔥🔥🔥🔥🔥

~~Target acc: 69.69%~~

Box of Replays

Google Drive Backup
For replay please find google drive

some G FUEL Rating (Extremely out of date / Archival)

(Depends on Personal Taste) Your mileage may vary Incomplete List
For better serving experience: JUSTaddICE

Don't expect watermelon, cherry, grape flavor in this list

Tier Overview
S ≧ A >>> B > B- >> C ≧ D (might swap around)
S / A = good stuff
B = less good stuff
B- = something wrong about it
C = no
D = big no
also drinking the same flavor everyday sucks
high humidity will surely make tub taste bad overtime
---Tier S

Blue Ice blue rasberry
Nemesis Tea Flavor better than peach iced tea imo
Shiny Splash Blue Ice alternative
Sour Blue Chug Rug different blue ice with sour they are not the same
Spicy Demon'ade not that spicy, powder is good for CC IRL
Strawberry Shortcake only superb with milk
HYPE SAUCE ~~not hyped~~
PewDiePie atleast on par with Chug Rug the sour really makes it great
Pink Lemonade Lemonade but better

---Tier A

Strawberry Banana + Milk compared to shortcake this lack the smell portion of it
Battle Juice less sweetness pineapple
Black On Blackberry focus on the berry taste
Blood Orange orange
Bobby Boysenberry that's what a grape flavor suppose to be
FaZeberry sweet berry
Green Apple sour apple
Kiwi Strawberry powder is strawberry but drink mostly kiwi fruit
Lemonade Lemonade
Peach Mango sweet = sour
Peach Iced Tea great flavor but only when its iced
Pink Grapefruit grapefruit juice 🍹 feeling
Tropical Rain Fruity sweet / Smells like ~~washing powder~~ idk
ŦYⱣE O Berry but much lighter sweet also less "dense" compared to other flavor
Snow Cone JUSTaddICE and now perfect for summer, blue ice but not blue ice
Star Fruit refreshing sweet starfruit
Twisted Kandy unique candy unlike cotton candy, this one with water is great
T-Tetrimino Berry Punch glad its not fruit punch

---Tier B+

Wumpa Fruit the smell isnt great (medicine), otherwise A

---Tier B

Bahama Mama mainly orange little bit other stuff.
FaZe X drinkable
Golden Apple Pear Decent Pear, but no apple and not refreshing
Ragin' Gummy Fish there is gummy taste but i prefer chewing stuff

---Tier B-

Cotton Candy cotton candy in a water form isnt the most brilliant idea at least they kinda fix their mistake from shortcake
French Vanilla Coffee coffee time but without sugar. plastic feel
MOON PIE with 1.5x powder still too watery
Peach Cobbler smells weird, taste weird, I rate this above Strawberry Shortcake
Strawberry Banana atleast not worst than shortcake
Strawberry Shortcake (15 Cal) atleast it is drinkable now

---Tier C

Cherry Limeade half medicine half lime
Classified Personally I hate watermelon but imagine watermelon candy it is ok
Strawberry Shortcake (25 Cal) disappointment with water
Fruit Punch some may find it is great but for me it is another type of medicine
J-Tetrimino Raspberry medicine

---Tier D

Grape compared to Rainbow Sherbet, this is sweeter
Rainbow Shebet Medicine
Sour Cherry Medicine + Sour (obito said its taste good???

not useful HxD hex code

C0 20 00 00 01 BE 07 DC 0F

-------------random stuff that may work----------------

osu! stable 720p "Veryslow" +ssim params (x264)

-------------
Better down scaling method (modify fps if needed):
such as fmtconv

#resize.vpy
import vapoursynth as vs
core = vs.core
import math
import sys
core.max_cache_size = 10000
clip = core.lsmas.LWLibavSource(source=target)
clip = core.fmtc.bitdepth (clip=clip, bits=16)
clip = core.resize.Spline36(clip, 1280, 720, dither_type="error_diffusion")
clip = core.fmtc.resample (clip=clip, css="420")
clip = core.fmtc.bitdepth (clip=clip, bits=10)
clip.set_output()

vspipe -r 1 -p --arg "target=in1080p.mkv" resize.vpy - | ffmpeg -f rawvideo -pix_fmt yuv420p10le -r 24000/1001 -video_size 1280x720 -i - -pix_fmt yuv420p -c:v libx264 -crf 0 in.mkv
ffmpeg -vsync passthrough -err_detect explode -i in.mkv -pix_fmt yuv420p -profile:v high -level 4.0 -c:v libx264 -b:v 1200k -preset veryslow -tune ssim -x264-params ref=9:bframes=16:deblock=-1,-1:fast_pskip=0:keyint=480:aq-mode=3:threads=8 -pass 1 -an -y -f mp4 NUL
ffmpeg -vsync passthrough -err_detect explode -i in.mkv -pix_fmt yuv420p -profile:v high -level 4.0 -c:v libx264 -b:v 1200k -preset veryslow -tune ssim -x264-params ref=9:bframes=16:deblock=-1,-1:fast_pskip=0:keyint=480:aq-mode=3:threads=8 -pass 2 -an -f mp4 video.mp4

-------------
Less ideal scaling:

ffmpeg -vsync passthrough -err_detect explode -i in.mkv -vf scale=-1:720:flags=lanczos -pix_fmt yuv420p -profile:v high -level 4.0 -c:v libx264 -b:v 1200k -preset veryslow -tune ssim -x264-params ref=9:bframes=16:deblock=-1,-1:fast_pskip=0:keyint=480:aq-mode=3:threads=8 -pass 1 -an -y -f mp4 NUL
ffmpeg -vsync passthrough -err_detect explode -i in.mkv -vf scale=-1:720:flags=lanczos -pix_fmt yuv420p -profile:v high -level 4.0 -c:v libx264 -b:v 1200k -preset veryslow -tune ssim -x264-params ref=9:bframes=16:deblock=-1,-1:fast_pskip=0:keyint=480:aq-mode=3:threads=8 -pass 2 -an -f mp4 video.mp4

-------------
tune=ssim usually gains tiny vmaf on low bitrate, but blur darkest area, while banding looks better (unless the grain is "part of the feature" then don't use tune=ssim), personally prefer this for general anime since <1500kbps is not enough bitrate to maintain dark detail anyways, especially in osu! where the video usually in opening/ending.
aq-mode=0 if dark scene is not a concern, this usually provide small vmaf gain but the result is even blocky in dark area especially on low bitrate

-------------
Although no-asm can produce different bitstream, it does not improve compression by any meaningful way other than waste of time.

no-asm   43,263,984 byte  vmaf 82.50683  psnr 33.45213  ssim 0.9801727  msssim 0.9828552
with asm 43,265,224 byte  vmaf 82.50312  psnr 33.45284  ssim 0.9801658  msssim 0.9828504

-------------
use threads=1 because why not extra 0.xxx% gain

Plex Web sucks:

Plex Web Client Direct Play 1080p "Veryslow" (x264)

if subtitle is .ass with Styles requires setting "Plex Web — Player — Burn Subtitles" to "Only image formats", otherwise Plex will transcode higher res video and burn subtitle / covert to .srt and only srt exist for converted resolution, this also solves mobile client not showing .ass subtitle

Edit the 720p code with the following
and change the responding section of code from 720 to 1080

-c:v libx264 -preset veryslow -x264-params ref=4:bframes=8:fast_pskip=0:aq-mode=3:keyint=480:threads=8

Plex Web Client Direct Play 720p "Veryslow" (space saving) (x264)

crf 27 + qaac tvbr V82 can get like 70% filesize of crf 23 + aac 128k
and get much better audio 18khz ish curve
instead of 16.5khz lowpass

tune ssim is not really worth the ~10% bitrate saving

if subtitle is .ass with Styles requires setting "Plex Web — Player — Burn Subtitles" to "Only image formats", otherwise Plex will transcode higher res video and burn subtitle / covert to .srt and only srt exist for converted resolution, this also solves mobile client not showing .ass subtitle

with qaac tvbr + vapoursynth fmtconv scaling

#resize.vpy
import vapoursynth as vs
core = vs.core
import math
import sys
core.max_cache_size = 10000
clip = core.lsmas.LWLibavSource(source=target)
clip = core.fmtc.bitdepth (clip=clip, bits=16)
clip = core.resize.Spline36(clip, 1280, 720, dither_type="error_diffusion")
clip = core.fmtc.resample (clip=clip, css="420")
clip = core.fmtc.bitdepth (clip=clip, bits=10)
clip = core.std.AssumeFPS(clip, fpsnum=24000, fpsden=1001)
clip.set_output()

ffmpeg -i 01.mkv -map 0:a:0 -dither_method triangular_hp -sample_fmt s16 -ac 2 -ar 48000 -f wav - | qaac - --raw-rate 48000 --raw-channels 2 -q 2 -V 82 -o 01a.mp4
vspipe -r 1 -p --arg "target=01.mkv" resize.vpy - | ffmpeg -f rawvideo -pix_fmt yuv420p10le -r 24000/1001 -video_size 1280x720 -i - -i 01a.mp4 -map 0:v -map 1:a -c:a copy -pix_fmt yuv420p -profile:v main -level 4.0 -c:v libx264 -crf 23 -preset veryslow -x264-params ref=9:bframes=16:fast_pskip=0:aq-mode=3:keyint=480:threads=8:rc_lookahead=250 -movflags faststart -f mp4 "01 - 720p.mp4"

How to waste electricity and ram (x265)

Approx 2GB RAM usage per instance of 1080p encode, have to pin individual core, ~~with enough instance you will have good argument for needing >32GB RAM~~

Avoid pinning 2 instance of pools="1" x265 into same core, might get slow enough to move into page file, CPU L2 L3 Cache already suffer enough.

Audio use flac 1.4.0+ then -c:a copy then mkvmerge
-r 8

flac.exe -8p -r 8 -A "subdivide_tukey(10)" _01.wav

ffmpeg -vsync passthrough -err_detect explode -i 01.mkv -map 0:v -pix_fmt yuv420p10le -profile:v main10 -level 4.0 -c:v libx265 -crf 17 -preset slower -x265-params level-idc=4:ref=5:bframes=8:aq-mode=3:min-keyint=1:keyint=360:rc_lookahead=80:no-open-gop=1:no-sao=1:ctu=64:min-cu-size=8:rdoq-level=2:subme=5:limit-refs=0:limit-modes=1:max-merge=5:high-tier=1:frame-threads=1:pools="1":analyze-src-pics=1:rd-refine=1:b-intra=1:no-wpp=1:pmode=1:no-strong-intra-smoothing=1:no-amp=1 _01.mkv

mkvmerge -o 01.mkv _01v.mkv _01.flac

aomenc (AV1) tbd

useful links:
https://www.reddit.com/r/AV1
https://www.reddit.com/r/AV1/comments/lfheh9
https://www.reddit.com/r/AV1/comments/l4ge0w

osu!lazer (with ffmpeg 4.3 dlls) cant realtime decode 10bit 720p without fg
guess x265 placebo go brrr...

Av1an is great if you able to make it work...
fill in number:
--cpu-used= (0 slowest, 6 faster, 3 or 4 is reasonable)
--kf-max-dist= (10x the fps of video, 20x for better compression for bad seek time)

pick one:
--target-bitrate= (in kbps)
--end-usage=q --cq-level= (it depends, the lower the better but higher bitrate, 20 is great until 50)

copy paste into the following:

ffmpeg.exe -i "in.mkv" -map 0:v -pix_fmt yuv420p -f yuv4mpegpipe - | aomenc.exe - --enable-tpl-model=1 --lag-in-frames=48 --enable-cdef=0 --auto-alt-ref=1 --tile-columns=1 --tile-rows=0 --bit-depth=10 --enable-fwd-kf=1 --frame-parallel=0 --threads=4 --enable-qm=1 --enable-chroma-deltaq=1 --quant-b-adapt=1 --passes=2 --pass=1 --fpf=firstpass.log -o firstpass.log
ffmpeg.exe -i "in.mkv" -map 0:v -pix_fmt yuv420p -f yuv4mpegpipe - | aomenc.exe - --enable-tpl-model=1 --lag-in-frames=48 --enable-cdef=0 --auto-alt-ref=1 --tile-columns=1 --tile-rows=0 --bit-depth=10 --enable-fwd-kf=1 --frame-parallel=0 --threads=4 --enable-qm=1 --enable-chroma-deltaq=1 --quant-b-adapt=1 --passes=2 --pass=2 --fpf=firstpass.log --webm -o out.webm
ffmpeg -i out.webm -c copy out.mkv