SH3/CV1000 emulation in MAME: CPU and blitter settings for emulating slowdown

Started by el_rika, March 28, 2020, 05:03:05 PM

Previous topic - Next topic

EOJ

More testing:

Ibara Kuro: I see Plasmo said (on a shmups forum thread that is on the same topic as this one) he set the CPU to 50% and the blitter off. Unfortunately, doing so produces slowdown that is far from very accurate.
Mushi: my ROM has sprites/bullets randomly disappearing, but I used Pearl's settings (37.6% CPU, blitter at 56%) and many parts ran too slow and/or felt too janky. I'll have to find a clean ROM somewhere and test again.
Pink Sweets: Again, I used Plasmo & Pearl's settings (37.2% CPU, blitter OFF). I played HARDER mode. Pretty good, but much slowdown is missing and some slowdown is too slow.

I tried various settings with Muchi Muchi Pork but so far I can't find one that plays accurately. Most emulate the boss slowdown well, but then there are spots in the stages that are too slow or too fast.

BTW the thread title on shmups forum, "Slowdown accuracy for cv1k games on MAME is almost perfect", made me chuckle after testing out these settings myself. If you change the "almost perfect" to "better than before", "pretty good", or "better than most people thought possible", it would be a more accurate assessment of the current situation.

Here is a good post about this topic from 3 years ago, with lots of settings listed per game:

https://shmups.system11.org/viewtopic.php?p=1276872#p1276872

Unfortunately, he doesn't seem to have a good understanding of the slowdown on the PCBs, so these numbers aren't of great help. Setting the blitter to 63% in DFK BL, for example, produces less accurate slowdown than the port (by quite a significant degree): I'd put it at about 60% accuracy.

My score archive
twitter: @cavexstg
youtube: @cave-stg
Xbox gamertag: eojx9999

EOJ

Topic split, so we can continue this discussion here.

When examining the slowdown accuracy in an emulator, the following factors need to be considered:

1) location: where the slowdown occurs in the PCB. Some slowdown is purely locational.
2) causation: what triggers the slowdown to occur? A shot type, a rank level, a bullet threshold, etc. Some slowdown is non-locational and thus is purely causational. Some slowdown is both locational and causational.
3) transitional-in slowdown: speed of the slowdown as it begins at a location or causation and moves into the durational phase.
4) durational slowdown: speed and duration of slowdown after the transition in.
5) transitional-out slowdown: speed of the slowdown as it moves out from the durational phase back to normal game speed.

I don't think anyone else has laid these out together like this before, but anyone who wants to do this seriously has to fully consider each factor as they tweak the CPU and blitter settings. So far everything I've tested has been inaccurate in all five factors at least some of the time and at least one factor almost every time.
My score archive
twitter: @cavexstg
youtube: @cave-stg
Xbox gamertag: eojx9999

Pearl

Quote from: EOJ on April 03, 2020, 07:08:24 PM
Mushi: my ROM has sprites/bullets randomly disappearing, but I used Pearl's settings (37.6% CPU, blitter at 56%) and many parts ran too slow and/or felt too janky. I'll have to find a clean ROM somewhere and test again.
Pink Sweets: Again, I used Plasmo & Pearl's settings (37.2% CPU, blitter OFF). I played HARDER mode. Pretty good, but much slowdown is missing and some slowdown is too slow.

Mushi is the single cv1k game that doesnt play nice when blitter is turned on during game boot. I have to disable blitter, boot the game, savestate at the copyright screen, close the game, boot the game again, turn on blitter and THEN load the save state.

what did you play for Mushi? ive only been able to notice maybe 2 or 3 sections with incorrect slowdown on Maniac with W-Power. I'm sure you know but mushi slowdown is really weird, a lot of it will feel slow but it has matched what I look at on pcb footage.

as for Pink Sweets, that's in it's early phases. I plan to play a lot of the game after calice cup and when that happens I'll be tweaking that game a lot more.

EOJ

Thanks for the Mushi workaround. I will try that.

I played Maniac with W-power. I have scored over 300 mil on the PCB (no rapid fire), so I know the game well. I have also owned the PCB many times (even within the past year) and have spent countless hours playing it. I know the slowdown in it like the back of my hand. Watching videos isn't the same as the tangible feel and reaction of the game in your hands. Believe me, I have tried to rely on PCB footage before in measuring slowdown accuracy and it leads to some false conclusions (usually, an overestimation of slowdown accuracy in a port or emulated version).
My score archive
twitter: @cavexstg
youtube: @cave-stg
Xbox gamertag: eojx9999

Pearl

Then I hope you're able to help us find an even closer setting for the game!

EOJ

My score archive
twitter: @cavexstg
youtube: @cave-stg
Xbox gamertag: eojx9999

EOJ

I used Pearl's Mushi workaround and got it to run properly with the CPU and Blitter settings he specified. Unfortunately in my estimation the slowdown accuracy overall is no better than 60%, compared to the PCB.

The main problems I encountered were:

-Slowdown missing in ST1 and the ST1 boss.
-Slowdown missing in ST2 (especially on the plants in the second half!) and the ST2 boss. The midboss is good, though.
-Slowdown missing in the first part of ST3. However, the slowdown after that is surprisingly accurate throughout the stage! (This is the real highlight of the entire MAME experience, it is better than the port in this part.)
-Stage 4 is a mess. Slowdown is missing in many spots in the first half, but the midboss has slowdown in its last phase that is too slow, making this tricky part a lot easier. Worst of all is the last part of the stage (everything after the midboss), which has WAY too much slowdown that lasts way too long. It is horrific.
-Stage 5 slowdown is stuttery, either too fast or too slow, and just feels rather janky overall. On the PCB it is smooth as butter. On the plus side, the midboss has the proper slowdown.

In addition, all of the transition-in and transition-out speeds are off (usually too fast in emulation).

I will try tweaking things up and down in the Slider menu, but it is possible what we have is already "as good as it gets" in MAME. I will be pleasantly surprised if I can get it significantly better, but like I said, I'll try.
My score archive
twitter: @cavexstg
youtube: @cave-stg
Xbox gamertag: eojx9999

Pearl

Quote from: EOJ on April 03, 2020, 10:05:39 PM
I used Pearl's Mushi workaround and got it to run properly with the CPU and Blitter settings he specified. Unfortunately in my estimation the slowdown accuracy overall is no better than 60%, compared to the PCB.

The main problems I encountered were:

-Slowdown missing in ST1 and the ST1 boss.
-Slowdown missing in ST2 (especially on the plants in the second half!) and the ST2 boss. The midboss is good, though.
-Slowdown missing in the first part of ST3. However, the slowdown after that is surprisingly accurate throughout the stage! (This is the real highlight of the entire MAME experience, it is better than the port in this part.)
-Stage 4 is a mess. Slowdown is missing in many spots in the first half, but the midboss has slowdown in its last phase that is too slow, making this tricky part a lot easier. Worst of all is the last part of the stage (everything after the midboss), which has WAY too much slowdown that lasts way too long. It is horrific.
-Stage 5 slowdown is stuttery, either too fast or too slow, and just feels rather janky overall. On the PCB it is smooth as butter. On the plus side, the midboss has the proper slowdown.

In addition, all of the transition-in and transition-out speeds are off (usually too fast in emulation).

I will try tweaking things up and down in the Slider menu, but it is possible what we have is already "as good as it gets" in MAME. I will be pleasantly surprised if I can get it significantly better, but like I said, I'll try.

Ultra or Maniac?

also, strange. Where is it supposed to slowdown on stage 1? and nothing ive checked has slowdown on those plants either..
i know its dumb to ask but are you sure you enabled blitter in machine configuration? i had to help someone with that earlier.

also, the stage 4 midboss does not have too much slowdown to my knowledge.

EOJ

Quote from: Pearl on April 04, 2020, 12:21:19 AM
Ultra or Maniac?

also, strange. Where is it supposed to slowdown on stage 1? and nothing ive checked has slowdown on those plants either..
i know its dumb to ask but are you sure you enabled blitter in machine configuration? i had to help someone with that earlier.

also, the stage 4 midboss does not have too much slowdown to my knowledge.

Maniac.

(All on the PCB:) ST1 has some slowdown on the large beetle thing that comes out on the left side of the screen after the midboss, as well as other bits here and there (none I'd classify as "important"). Yes there is slowdown on the plants if you score on them properly (tap, tap, tapping away). MAME emulation is also missing slowdown on the sand beetle things in the second half. Yes I enabled blitter at the rate you specified.

RE: ST4 midboss. Last phase has slowdown on the PCB/port and MAME. It's very important because if you're serious about scoring you need to jack up your counter to at least 100k in the previous phases and if you die in the last phase your whole score is killed for the stage. Port is a bit too fast here, MAME is a bit too slow, (so far) PCB alone has it "just right".

For some perspective, the differences in slowdown in ST1 and ST2 are relatively minor. None of that missing slowdown strikes me as very important. But the differences in ST4 and ST5 are significant.
My score archive
twitter: @cavexstg
youtube: @cave-stg
Xbox gamertag: eojx9999

Plasmo

Great we got some discussion going here! :righton:

Quote from: EOJ on April 04, 2020, 02:17:23 AMBut the differences in ST4 and ST5 are significant.

Could you maybe link to the replay you've used to determine this, ideally with timestamps? I think transparency is the most important thing here. It would be even better if people who have the PCB could record some footage to put it directly side by side to Mame.

It's great that you know the game that well EOJ, but please let us noobs also understand what exactly you are referring to. Otherwise we will not be able to tweak it ourselves.

el_rika

Quote from: EOJ on April 03, 2020, 06:14:38 PM

Groovymame with lag reduction ON still feels like it has about 4-5 frames of lag on my PC. This is quite noticeable when I play DFK BL, for example. It feels about the same as the lag in the SDOJ port. This is a big improvement over how MAME used to feel -- I mean, it felt like 6-8 frames before for me -- but it doesn't come close to the PCBs or the good X360 ports.

I have 2 frames of lag with Retroarch and mame core in android (though an extra ~ 5 from the touch screen latency), so i'd be surprised if Groovy alone had double that  :-\  Be sure to enable nobuffer patch in mame settings/cfg, it should cut an extra frame of lag.

xygax

Quote from: EOJ on April 03, 2020, 06:14:38 PM
Groovymame with lag reduction ON still feels like it has about 4-5 frames of lag on my PC. This is quite noticeable when I play DFK BL, for example. It feels about the same as the lag in the SDOJ port. This is a big improvement over how MAME used to feel -- I mean, it felt like 6-8 frames before for me -- but it doesn't come close to the PCBs or the good X360 ports.

4-5 frames, getting that with a 2 frames source like cv1k sounds typical of broken settings, meaning the lag reduction isn't working. Maybe 9 in 10 new users experience that as I've witnessed (typically they changed settings in the mame.ini they shouldn't have, or introduced another parameter that's breaking everything, like misconfigured multi-monitors, or they downloaded the wrong build, etc)
There is a quick configuration guide which has crucial info but the site is currently down so check later.
http://geedorah.com/eiusdemmodi/forum/viewtopic.php?id=433

When you use lag reduction (frame delay slider from the UI menu) you're supposed to see tearing appear, that means the feature is active, and as you increase the frame delay level you'll get where it affects performance negatively (frame/sound drops)
[often this is where users go and mess with sync settings and make a fatal mistake]
Set at a stable level before it, like if 7 is too much set 6 or even 5.
On my PC with cv1k games I can set 5 for global cv1k stability. 7~8 is better of course but it's really heavy on the CPU and GPU at times. and yes with Groovy the GPU matters a bit too, in particular with high-res displays like full-hd, wqhd, 4k, so I wouldn't trust much things like Intel iGPUs)
tip: keep an eye on F11 when you set frame delay, make sure the meter stays at 100%.

Then create a specific .ini file for a game, for instance pinkswts.ini
in it write:
vsync_offset              250
and drop it into the 'ini' folder.

That 250 value is just for illustration, you may need a much lower or higher one, play the game to see if it gets rid of the tearing.
typically this takes a few tries to get right.
Note higher values can mean it's too tight for your hardware and you should maybe drop frame delay 1 level then try a different offset value.
tip: the higher resolution your display, the greater values you'll need, think about your monitor's vertical resolution, you're moving the tearing line up and down it.

Alternatively you can use a single ini named cv1k.ini, and drop it into a 'source' subfolder in the 'ini' one.
but using a single offset value for all games don't always work perfect...

Anyway if you did that right you now have your vsynced and lag-reduced video working the way it should, tear-free butter-smooth scrolling and low lag close to that of the pcb (2~3 frames)
Levels of frame delay kinda represent setting chances of occurences of having no unwanted lag at all, it's not really a percentage but you could picture it like with 7 you have 70 chances of an input being lagless within the targeted frame time.
It's more subtle than that but Calamity would explain better.

For the CPU slider, remember there's ten steps in between each % , hence the slowness, and that's not visible from the UI (go to each game's individual .cfg file, open and witness your CPU setting on a 1000 scale to be absolutely sure of the slider position, eventually edit manually there if that's your thing)
As I've said before, even 0.1% can matter in places, some in-game behaviours will trigger or won't by a hair.
If nobody watches this, then 10 people who've set for instance 50% could have 10 different set values without knowing it.

I don't know why you're experiencing a game missing the CPU slider, I'd have to know everything about your setup, build, settings, roms, etc but I won't really go and ask that then provide the inevitable hours of support like I did tens of times on the farm and other places, paraphrasing the official guide(s) with additional explanations, because it's led to people most times failing, essentially because they were very uncooperative and a few other reasons, and I've worked my ass for nothing. I would maybe for you EOJ but there are people here I really don't like and I don't want to help them along.
You can ask anything in private though.


PS: about Calamity, few people who would really have benefited from his help actually went and asked at BYOAC (el_rika gave the link), and those who did often wouldn't bother producing a log file like Calamity demands so he can provide efficient help.
For anyone who's having trouble with Groovy that's the place to go and where to receive support, in fact this is where all the ppl serious about this current topic should go now, start a thread similar to this one and get aid from the most skilled people there along Calamity's.
That would be the smarted and most valuable move, much, much better than going for the dead-end made with old builds and retroarch trash.

However you know the situation right now, and he's in Spain (ouch), so it's possible there won't a be a response before long (plus he was already often absent from the forum even before that)

pps: oh and if you need fresh roms go to r.e.t.r.o.r.o.m.s , besides the 'dome' it is the only place carrying up-to-date rom sets, which is absolutely crucial these days the way modern MAME works, and therefore for Groovy too (wrong/old roms = some things might not work properly if at all)

el_rika

Quick [crappy] run of Mushihime-sama ultra up to level 4 mid boss.

https://youtu.be/5BWclKbSbKk

(the small annoying stutters are from youtube rendering, not the video itself)

Though maybe a tad slower in some moments (ex: the level 4 mid boss seems to start the last pattern a bit slower than pcb..or not?), it is noticeably closer to pcb than the ps2 port.
  As a small observation, the ps2 port also lacks the on-the-fly blitter management that exists in the pcb hardware, that affects a few spots in the game (very few luckily). Just like mame, it uses a pre-set CPU and Blitter for the entire game. {probably the same case with Ibara}

el_rika

Quote from: xygax on April 04, 2020, 04:36:23 PM


However you know the situation right now, and he's in Spain (ouch), so it's possible there won't a be a response before long (plus he was already often absent from the forum even before that)


Let's hope he and everyone will be ok in these pretty dark times. Stay inside as much as possible guys, play some Caves, and we and ours will be safe  :righton:

EOJ

Quote from: Plasmo on April 04, 2020, 03:56:38 AM

Quote from: EOJ on April 04, 2020, 02:17:23 AMBut the differences in ST4 and ST5 are significant.

Could you maybe link to the replay you've used to determine this, ideally with timestamps? I think transparency is the most important thing here. It would be even better if people who have the PCB could record some footage to put it directly side by side to Mame.

It's great that you know the game that well EOJ, but please let us noobs also understand what exactly you are referring to. Otherwise we will not be able to tweak it ourselves.

I didn't use any replay.  However, the official superplay DVD should be sufficient to show the differences I'm talking about. I listed the parts of the stages that were particularly bad in a previous post. As you must know yourself, if you play an arcade game enough (recently) and you have a set route you know right away where it feels different in an emulation or port. And people who are intimately familiar with the way the PCBs play should be the ones judging each new setting proposed (ideally, two or more people per game). I am doing my part here, and I hope other PCB players will join in. In the meantime I hope people will keep posting cpu/blitter settings that they think are good, and I will keep testing them.

@xygax: thank you very much for your detailed reply. It is very helpful. I didn't know about the lag reduction stuff you wrote, but I did know how to properly tune the CPU settings in each game's .cfg file. I play on a Dell Inspirion 5475 All-in-one with the latest build of groovymame (groovymame64_0219.017p_win-7-8-10). It has a 3.1GHz AMD Ryzen 7 CPU and a 4GB Radeon RX560 GPU. I suspect the screen itself introduces at least a few frames of lag.
My score archive
twitter: @cavexstg
youtube: @cave-stg
Xbox gamertag: eojx9999

el_rika

Later edit: much better numbers that result in way better accuracy are on the next page.

Futari 1.5 maniac Palm Abnormal quick test (CPU 37.9 Mhz, Blitter 54%) Reco needs a lower CPU speed for the smooth slowdown.

https://youtu.be/h-5qT1hOWRQ  (sorry for the potato quality, be sure to select 320p)

Youtube seems to be having an issue, here it's another upload that works better:

https://vimeo.com/406256643

Observations:
Just a bit overall slower (ex. Boss 1 lava speed is just like in BL, Boss 4 third pattern slows down ~ 1s earlier, though feels less jumpy than pcb, some stage bits of slowdown last 2s longer).
Some 2 - 3 Mhz more should get it as close as it can be, though a bit of stutter occurs at times (seen in pcb/360 recordings as well).

el_rika

Deathsmiles quick test (cpu 52.5 Mhz, blitter 59%) Lvl 3 forest & cemetery:

https://vimeo.com/407969615

Observations: very good overall, not easy to tweak, as even on pcb, many slow parts don't always trigger the same way, due to very very small yet drastically affecting factors (player position, type of shot, quantity of moving objects).

[Only] Rosa needs a lower blitter (57%) and a higher CPU clock (+ 1.5 Mhz) due to the more complex shot/laser graphics that produces more slowdown. (Pcb is inconsistent in some instances with Rosa as well, sometimes it slows down drastically, sometimes not as much).

EOJ

I will try the Deathsmiles settings this weekend, but the video does not look very close to the PCB (several sections have added slowdown, slowdown transitions in- and out- are too fast, etc. The usual problems!). It's misleading to claim "even on pcb, many slow parts don't always trigger the same way, due to very very small yet drastically affecting factors". In fact, the slowdown on the PCB always triggers the same way. That is to say, if you play the same pattern with the same character, the slowdown will be exactly the same every time. As on every SH3 PCB, slowdown is causational, locational, or a mix of the two.
My score archive
twitter: @cavexstg
youtube: @cave-stg
Xbox gamertag: eojx9999

el_rika

My wording was a bit wrong maybe  :bigsmile: This game's timings seems more affected by the smallest of factors, almost impossible to observe, compared to other games.(ex. on pcb, Rosa and the Dragon boss, first pattern can be slightly slow, or very slow without obvious differences in gameplay [the amount of falling rocks you randomly destroy might be an important factor]. Another ex. forest Boss big apples pattern slows down earlier/more sudden, if both apples appear ~ the same time, or ~ 2 sec later if one appears first).

I've seen this spasmic slowdowns ins and outs on many pcb Deathsmiles gameplays (all sh3 games have them). I think it's overall a bit slower too, and also much closer than the port (my God that's bad).

EOJ

Quote from: el_rika on April 16, 2020, 01:11:31 AM
also much closer than the port (my God that's bad).

The X360 port is one of the more accurate CAVE ports (certainly more accurate than any MAME emulation I've seen thus far). Perhaps you are thinking of the Steam port, which is horrible.

My score archive
twitter: @cavexstg
youtube: @cave-stg
Xbox gamertag: eojx9999

el_rika

Was it maybe patched at some point? Because i have it on 360 and it's very off in many instances (forest boss slowdown is off, graveyard as well, level sections are either too fast or too slow).

https://youtu.be/Or2uHxS0o2s  (forest and graveyard bosses are very eloquent).

EOJ

JP version is excellent (no need for a patch). US version was messed up on release but received a patch later on that fixed the slowdown. You should download the patch.

Also, the video you linked to is 360 mode. Arcade mode has more accurate slowdown.
My score archive
twitter: @cavexstg
youtube: @cave-stg
Xbox gamertag: eojx9999

el_rika

I assumed as much, as it was really messy. Will do that sometime in the future when i get to my xbox. Thanks for the info  :righton:

el_rika

Quick question: does Deathsmiles MBL slowdown more than vanilla (i'm talking similar conditions, say both level3)? Black Labels have a tendency to do that.

Could anyone link me a pcb (preferrably Windia) MBL playthrough (not lvl999)?

EOJ

There are a lot of videos on Nico douga:

https://www.nicovideo.jp/search/%E3%83%87%E3%82%B9%E3%82%B9%E3%83%9E%E3%82%A4%E3%83%AB%E3%82%BAMBL?sort=h&order=d

I don't recall there being more slowdown in LV3 in MBL compared to vanilla. The shots/lasers are basically the same in each game. Bullet patterns are very similar too.

However, there are parts with LESS slowdown in MBL compared to the vanilla. Such as the Jitterbug boss fight. This is apparently due to the higher spec of the CV1000D ("SH3-b") PCB. DSMBL is the only BL to be on different hardware in comparison to the vanilla version.
My score archive
twitter: @cavexstg
youtube: @cave-stg
Xbox gamertag: eojx9999

el_rika

Oh, thanks so much for answering. I didn't know that MBL used slightly different specs; this might explain the small speed differences.
I will check the link you provided for some down to earth pcb Windia vids!

I'be been testing for a couple of hours and it seems the overall slowdown is indeed similar, with some exceptions. Most importantly, Windia's full power laser (BIG option circle shots) definitelly slows down the game more than in vanilla (which has a smaller laser @full). Also some bosses have slowdown eliminated or modified, see forest Tree Boss, where big apples pattern slows down much less and only if you keep shooting the apples. Also last pattern doesn't have the vanilla slowdown anymore (nor the safespot :-[)
Lava dragon also has most of the 1st pattern slowdown removed.

I'm only talking level3 here, as 999 i way out of my league atm, but it's getting clearer ~ 2 Mhz higher and ~ 2% faster blitter is starting to look right for MBL compared to vanilla.

Thanks again

Will return with more findings  :righton:

el_rika

So, Deathsmiles MBL is indeed a bit different in terms of slowdown especially when you activate 1000 Power mode. Windia's full laser slows down the game much more agressively than in vanilla. Capser's shot as well.

1.5 mhz more and 2% less blitter usually does the job. For Windia though, 54% blitter seems to be the sweet spot. Taking a little break from it for now.


Meantime here's an Espgaluda 2 quick test with Cpu at 49.76 mhz and blitter delay 63%. There are spots when the game slows down, depending on how you play and what shot you use, especially in later levels, but most of the slowdown occurs in awakening and ultra-awakening, mid bosses and bosses.

https://streamable.com/mhvaqf

(Any audio delay or slight stutters are due to streaming conversion)

el_rika

Daifukkatsu 1.5, what a gorgeous game   ^-^

Different ships/modes produce more or less agressive slowdown, with B/C Strong ships' shot producing the most. Unlike Futari though, it seems on pcb Cave didn't use different CPU/Blitter values, to have similar timings with all ships/modes.

On pcb, mostly with green and blue, the game has certain parts where it exhibits that CPU dependant spasmic slowdown seen in Futari (and most others).

Too high CPU clock value, eliminates some of the subtle slowdown found on pcb, while too low produces too much in-out slowdowns (spastic). There are some spots affected exclusively by blitter (ex. 2nd Boss penultimate attack), so it's a matter of fine-tuning it.

Quick test with B Power:
https://streamable.com/nppkmt

Quick test with B Strong (a bit longer):
https://vimeo.com/422407474

Cpu 43.62 Mhz 
Blitter 59%

As always, the upload conversion produces a little stutter in a couple instances, so ignore it.

el_rika

Could anyone (EOJ?) link me some pcb Akai Katana gameplays? Prefferably not superplays. This is another game where C ship slows down the game much more than A & B ships, so careful comparison is needed.
Youtube doesn't like this game so much, there aren't many pcb videos of it.

Thanks very much!

EOJ

I don't notice any extra slowdown with Type C in Akai katana compared to the others. The slowdown in the game is based mainly on how much gold you have rotating around as well as how many enemies and bullets are on the screen. For PCB videos there is youtube and nicodouga and that's about it AFAIK. Not a whole lot out there.

Your DFK settings produce slowdown that is not very accurate to the PCB (too much slowdown, slowdown is too slow, etc. The usual problems).
My score archive
twitter: @cavexstg
youtube: @cave-stg
Xbox gamertag: eojx9999