It seems to be a problem with the d3dcompiler_47, install the native version with protontricks

It seems to be a problem with the d3dcompiler_47, install the native version with protontricks

tested and that didnt do the trick. i see kisak added the XAudio2 label so i tried installing xact via protontricks as well. but that didnt do the trick either

I know this is probably really unhelpful, especially as I'm my on a phone right now and can't give meaningful rapports, but the game seemed to run fine for about an hour at whatever proton version and settings were pre-defined by Valve. Solid 60@1440, with NPCs glaring at us the whole time.

Specs are: GTX 1080ti, 9900k@stock, 2TB SSD NVMe drives.

I promise I'll update with proper specs and related etc. when I wake up.

on launch i got a black screen with an audio pop

installing xact64 using protontricks got me into the title screen but as above the game will crash when it tries to play the titlescreen video. Starting a new game resulted in a flat green screen with audio. i didn't try going further.

5900X, 6900XT, 5.11.16-2-MANJARO, mesa-21.0.3

log

Given the issue template asks me to confirm not having found any other reports for I game, I take this to mean we should not open multiple reports for the same game. While I don't think I observe the same issue as MagZu, I'll add my report here as a comment. Please advise if I should make this its own issue instead.

Compatibility Report

• Name of the game with compatibility issues: NieR Replicant ver.1.22474487139
• Steam AppID of the game: 1113560

System Information

• GPU: AMD AMD Radeon RX 5600 XT
• Driver/LLVM version: 4.6 Mesa 20.3.5
• Kernel version: 5.11.15-200.fc33.x86_64
• Link to full system information report: https://gist.github.com/Gambloide/f24b669ef1765086610016f86178ec72
• Proton version: experimental-6.3-20210423 AND proton-6.3-2

I confirm

• [X] that I haven't found an existing compatibility report for this game.
• [X] that I have checked whether there are updates for my system available.

Symptoms

Normal Steam installation. Both with proton experimental and 6.3-2 the game tries to launch, briefly shows an empty black window and then closes again. Journalctl shows unhandled page faults on read access to to 0xffffffffffffffff in 64-bit code at the time of the crash, with other apparently non-critical errors shortly before.

• 6.3-2 logs: https://gist.github.com/Gambloide/97a9d5f37f3cfb6734e8110ca1ac852b
• Experimental logs: https://gist.github.com/Gambloide/5ff532a27acb1d5465bf270ee22f3dec

Reproduction

Try to launch the game

@Gambloide from your log https://gist.github.com/Gambloide/5ff532a27acb1d5465bf270ee22f3dec#file-gistfile1-txt-L481:
It fails to find a suitable vulkan driver for some reason. I've seen report from at least one other person that this happened to them with mesa 20.3, but was solved when upgrading to mesa 21.0. Not sure why this error only comes up with this game and mesa 20.3 though because it fails before even initializing vulkan.

Compatibility Report

• Name of the game with compatibility issues: NieR Replicant ver.1.22474487139
• Steam AppID of the game: 1113560

System Information

• GPU: RX 5700 XT
• Driver/LLVM version: Mesa 21.0.3-1
• Kernel version: 5.11.16
• Link to full system information report as Gist:
• Proton version: 6.3-2 and experimental

I confirm:

• [x] that I haven't found an existing compatibility report for this game.
• [x] that I have checked whether there are updates for my system available.

Symptoms

Installed game via steam when launching via steam play the game will load and show a black screen for about a second before immediately closing, this happens with all version of proton I tested 4+

Reproduction

Install game and launch via steam

• 6.3-2 log: Gist
• Experimental log: Gist
• 5.13-6 log: Gist

Also having this crash on an AMD 4700U laptop with Arch Linux. Game boots up fine, initially the logs indicated that it needed the native d3dcompiler_47 but kept crashing. Based on this thread, I'd guess it crashes because of the cutscene. I'm not seeing any errors about missing graphics drivers. Can't test on my main PC though.

Replicant uses a lot of RAM, moreso than other games I've played even on my main PC. The usage jumps up to almost 8GB before crashing, with Replicant using most of that! I haven't seen it mentioned, so something that other players could keep in mind.

If anyone is getting black or white screens and crashes immedietly at startup this might help

"If any of you have non-standard english characters in your username, try removing them. I had Japanese quotation marks in mine and that prevented me from being able to boot the game."

I might say, those .arc files could be just compressed assets with the movies in them. I wouldn't be surprised if this turns out to be yet another Media Foundation error.

My issue does involve spoilers for the game. I didn't see a way to do hidden text through markdown on github, so symptom is attached as a gist.

Compatibility Report

• Name of the game with compatibility issues: NieR Replicant ver.1.22474487139
• Steam AppID of the game: 1113560

System Information

• GPU: GTX 980Ti
• Driver/LLVM version: Nvidia 460.67
• Kernel version: 5.11.0-7614-generic
• Link to full system information report as Gist:
• Proton version: 6.3-2

I confirm:

• [ x] that I haven't found an existing compatibility report for this game.
• [ x] that I have checked whether there are updates for my system available.

Symptoms

Game crash to desktop after selecting story option and load screen.
More detail, contains spoilers, gist

Reproduction

Choose option, game crashes after load screen.

• 6.3-2 Log steam-1113560.zip

I'm baffled by what I encounter with the game: when the intro starts the fps go down to 28, when in control it stays at around 80 when turning the camera or when certain animations play it goes back down to 28 and stays there or sometimes at specific angles it shoots up again. I've tried the fps limiting, monitor refresh rate to 60, disabling G-Sync but the low fps problem persists.
This is with Proton 6.3 and Experimental, Ryzen 9 5900X, Nvidia 980Ti (465.24.02)

Interestingly with Glorious Eggroll's Proton 6.5-GE-2 the fps go down to like 17.

System Info Gist

My issue does involve spoilers for the game. I didn't see a way to do hidden text through markdown on github, so symptom is attached as a gist.

Exactly what I encountered too, @etw10

Choose option, game crashes after load screen.

If this is about halfway into the game after a bossfight, I am having the exact same issue. I'm testing to see if I also get crashes at the title screen if I let the game run too long as I suspect its the same issue.

EDIT: Yep, getting the exact same issue when I let a cutscene play on the title screen. Going to launch the game with Steam opened from commandline to pull logs.

EDIT 2:

Unhandled exception: page fault on read access to 0x00000000 in 64-bit code (0x00000001407ae12e).
Register dump:
 rip:00000001407ae12e rsp:000000002a9aeee0 rbp:000000002a9aefe0 eflags:00010206 (  R- --  I   - -P- )
 rax:0000000080004001 rbx:000000000dfe8210 rcx:0000000000000000 rdx:000000002a9aef20
 rsi:000000000dfe8218 rdi:000000002a9af7e8  r8:0000000000000000  r9:0000000000000080 r10:0000000000000000
 r11:000000002a9af7e8 r12:00000000004b2a98 r13:0000000000000001 r14:0000000000000001 r15:0000000140abff00
Stack dump:
0x000000002a9aeee0:  000000002d4795f0 0000000359224d39
0x000000002a9aeef0:  000000002aace450 00000000004e2050
0x000000002a9aef00:  000000002e3adb10 000000002a9aef90
0x000000002a9aef10:  0000000000000000 0000000000000000
0x000000002a9aef20:  0000000000000000 0000000000000000
0x000000002a9aef30:  fffffffffffffffe 000000000dfe8210
0x000000002a9aef40:  000000000dfe8218 000000002a9af0b0
0x000000002a9aef50:  00000000004b2a98 000000002d4795f0
0x000000002a9aef60:  00000000004e2050 000000035908e218
0x000000002a9aef70:  000000002d479620 000000002a9aefb8
0x000000002a9aef80:  000000003c88df37 0000000000000000
0x000000002a9aef90:  000000002e3adb10 00000000004e2050
Backtrace:
=>0 0x00000001407ae12e EntryPoint+0xe013110e() in nier replicant ver.1.2247448713 (0x000000002a9aefe0)
  1 0x00000001407aa91c EntryPoint+0xe012d8fb() in nier replicant ver.1.2247448713 (0x00000001426e3e98)
  2 0x00000001403d3e51 EntryPoint+0xdfd56e30() in nier replicant ver.1.2247448713 (0x000000002a9afaa0)
  3 0x00000001404ef1df EntryPoint+0xdfe721be() in nier replicant ver.1.2247448713 (0x000000002a9afaa0)
  4 0x000000014008322e EntryPoint+0xdfa0620d() in nier replicant ver.1.2247448713 (0x000000002a9afaa0)
  5 0x0000000140086e28 EntryPoint+0xdfa09e07() in nier replicant ver.1.2247448713 (0x0000000000000000)
  6 0x000000014003c6c9 EntryPoint+0xdf9bf6a8() in nier replicant ver.1.2247448713 (0x0000000140ac2488)
  7 0x000000014003c941 EntryPoint+0xdf9bf920() in nier replicant ver.1.2247448713 (0x0000000140ac2488)
  8 0x00000001400693e6 EntryPoint+0xdf9ec3c5() in nier replicant ver.1.2247448713 (0x00000001415add60)
  9 0x0000000140067f7a EntryPoint+0xdf9eaf59() in nier replicant ver.1.2247448713 (0x0000000000000000)
  10 0x00000001407b0619 EntryPoint+0xe01335f8() in nier replicant ver.1.2247448713 (0x0000000000000000)
  11 0x00000001407bc66c EntryPoint+0xe013f64b() in nier replicant ver.1.2247448713 (0x0000000000000000)
  12 0x00000001407b0525 EntryPoint+0xe0133504() in nier replicant ver.1.2247448713 (0x0000000000000000)
  13 0x00000001409598d0 EntryPoint+0xe02dc8af() in nier replicant ver.1.2247448713 (0x0000000000000000)
  14 0x000000007b62c939 EntryPoint+0xffffffffffffffff() in kernel32 (0x0000000000000000)
0x00000001407ae12e EntryPoint+0xe013110e in nier replicant ver.1.2247448713: movq       (%rcx),%rax
Modules:
Module  Address                                 Debug info      Name (69 modules)
PE              3b400000-        3b441000       Deferred        steam_api64
PE              7a850000-        7a854000       Deferred        opengl32
PE              7b000000-        7b0d5000       Deferred        kernelbase
PE              7b600000-        7b812000       Export          kernel32
PE              7bb10000-        7d288000       Deferred        dxgi
PE              7bb10000-        7d288000       Deferred        ntdll
PE             138000000-       13948b000       Deferred        steamclient64
PE             140000000-       16073a000       Export          nier replicant ver.1.22474487139
PE             180000000-       1801ea000       Deferred        gameoverlayrenderer64
PE             1c1ef0000-       1c1f1c000       Deferred        atl100
PE             1c8890000-       1c88f9000       Deferred        d3dcompiler_47
PE             1c8b40000-       1c8b60000       Deferred        msacm32
PE             1c8db0000-       1c8e3c000       Deferred        msvcrt
PE             1d97a0000-       1d97a7000       Deferred        api-ms-win-core-fibers-l1-1-1
PE             1f51e0000-       1f51ef000       Deferred        hid
PE             205950000-       205958000       Deferred        api-ms-win-appmodel-runtime-l1-1-2
PE             21a380000-       21a390000       Deferred        xinput9_1_0
PE             21a7e0000-       21a854000       Deferred        setupapi
PE             228450000-       22847a000       Deferred        propsys
PE             231ae0000-       231b62000       Deferred        rpcrt4
PE             236df0000-       236e49000       Deferred        dsound
PE             23d820000-       23da47000       Deferred        user32
PE             2419a0000-       2419df000       Deferred        wbemprox
PE             24f470000-       24f48f000       Deferred        mpr
PE             26b4c0000-       26b642000       Deferred        gdi32
PE             2739c0000-       273af6000       Deferred        oleaut32
PE             28ebe0000-       28ebe8000       Deferred        api-ms-win-core-sysinfo-l1-2-1
PE             29e400000-       29e435000       Deferred        msctf
PE             2bb750000-       2bb894000       Deferred        comctl32
PE             2bc640000-       2bc650000       Deferred        imagehlp
PE             2c83b0000-       2c8403000       Deferred        winegstreamer
PE             2e3540000-       2e3591000       Deferred        shlwapi
PE             2e8f10000-       2e9028000       Deferred        ole32
PE             2f1fa0000-       2f1fad000       Deferred        version
PE             2f7230000-       2f725d000       Deferred        uxtheme
PE             30c980000-       30c988000       Deferred        api-ms-win-core-synch-l1-2-0
PE             3126f0000-       312709000       Deferred        shcore
PE             317d60000-       317e88000       Deferred        dxvk_config
PE             319900000-       31992f000       Deferred        mfreadwrite
PE             327020000-       327072000       Deferred        combase
PE             32a700000-       32a729000       Deferred        sechost
PE             330260000-       33029f000       Deferred        advapi32
PE             335260000-       335272000       Deferred        rtworkq
PE             336850000-       33687a000       Deferred        mmdevapi
PE             341d30000-       341dfd000       Deferred        quartz
PE             34abc0000-       34abe1000       Deferred        msdmo
PE             351680000-       351688000       Deferred        api-ms-win-core-localization-l1-2-1
PE             359050000-       359303000       Deferred        d3d11
PE             3785e0000-       3785ec000       Deferred        vulkan-1
PE             39a620000-       39a644000       Deferred        msvfw32
PE             3a0440000-       3a04c0000       Deferred        wininet
PE             3a6d40000-       3a6d9c000       Deferred        mfplat
PE             3ad720000-       3ad74f000       Deferred        devenum
PE             3af670000-       3af728000       Deferred        ucrtbase
PE             3afd00000-       3afd1a000       Deferred        imm32
PE             3b8f00000-       3b8fb7000       Deferred        winmm
PE             3be590000-       3be5f0000       Deferred        dbghelp
PE          7ff588620000-    7ff58862b000       Deferred        winspool
PE          7ff590320000-    7ff590324000       Deferred        iphlpapi
PE          7ff590410000-    7ff590414000       Deferred        xaudio2_9
PE          7ff590440000-    7ff590443000       Deferred        winepulse
PE          7ff5b8010000-    7ff5b8013000       Deferred        avicap32
PE          7ff61b3d0000-    7ff61b3d3000       Deferred        lsteamclient
PE          7ff61b8c0000-    7ff61b8c4000       Deferred        winex11
PE          7ff61bcc0000-    7ff61bcc4000       Deferred        ws2_32
PE          7ff61bd20000-    7ff61c5f7000       Deferred        shell32
PE          7ff61e070000-    7ff61e0ba000       Deferred        dinput8
PE          7ff61e110000-    7ff61e114000       Deferred        wined3d
PE          7ff61e510000-    7ff61e514000       Deferred        winevulkan
Threads:
process  tid      prio (all id:s are in hex)
00000020 steam.exe
        00000024    0
        000000bc    0
00000030 services.exe
        00000034    0
        00000038    0
        00000044    0
        00000060    0
        00000078    0
        00000084    0
        000000a8    0
        000000e8    0
0000003c winedevice.exe
        00000040    0
        0000004c    0
        00000050    0
        00000054    0
00000058 plugplay.exe
        0000005c    0
        00000064    0
        00000068    0
        0000006c    0
        0000008c    0
        00000224    0
        00000238    0
00000070 winedevice.exe
        00000074    0
        0000007c    0
        00000080    0
        00000088    0
        00000094    0
        00000098    0
        0000009c    0
        00000220    0
        00000228    0
000000a0 svchost.exe
        000000a4    0
        000000ac    0
        000000b0    0
000000b4 conhost.exe
        000000b8    0
000000c0 explorer.exe
        000000c4    0
        000000cc    0
        000000d8    0
000000e0 rpcss.exe
        000000e4    0
        000000f0    0
        000000f4    0
        000000f8    0
        000000fc    0
        00000100    0
00000110 (D) Z:\home\semperverus\.local\share\Steam\steamapps\common\NieR Replicant ver.1.22474487139\NieR Replicant ver.1.22474487139.exe
        00000114    0
        00000118    0
        0000011c    0
        00000120    0
        00000124    0
        00000128    0
        0000012c    0
        00000130   -2
        00000134   -2
        00000138   -2
        0000013c   -2
        00000140   -2
        00000144   -2
        00000148   -2
        0000014c   -2
        00000150   -2
        00000154   -2
        00000158   -2
        0000015c   -2
        00000160   -2
        00000164   -2
        00000168   -2
        0000016c   -2
        00000170    0
        00000174    0
        00000178    0
        0000017c    0
        00000180    0
        00000184    0
        00000188    0
        0000018c    0
        00000190    0
        00000194   -1
        0000019c   -1
        000001a0   -1
        000001a4    0
        000001a8    1
        000001ac    0
        000001b0    1
        000001b4    1
        000001b8    0
        000001bc    0
        000001d0    0 <==
        000001d4    0
        000001d8    0
        000001dc    0
        000001e0    0
        000001e4    0
        000001e8    1
        000001ec    1
        000001f0    1
        000001f4    1
        000001f8    2
        000001fc    0
        00000200    0
        00000204    0
        00000208    0
        0000020c   -1
        00000214    2
        00000230    0
        00000234    0
        0000023c    0
System information:
    Wine build: wine-6.3
    Platform: x86_64
    Version: Windows 10
    Host system: Linux
    Host version: 5.11.16-zen1-1-zen

I might say, those .arc files could be just compressed assets with the movies in them. I wouldn't be surprised if this turns out to be yet another Media Foundation error.

Look like those video is encoded in wmv2 format.

hexdump of MOVIE_ATRACT_GESTALT.arc include this

|.........m.a.j.o|
|.r._.b.r.a.n.d..|
|.....m.p.4.2....|
|.m.i.n.o.r._.v.e|
|.r.s.i.o.n......|
|.0...$.c.o.m.p.a|
|.t.i.b.l.e._.b.r|
|.a.n.d.s.......m|
|.p.4.2.m.p.4.1..|
|.(.W.M./.E.n.c.o|
|.d.i.n.g.S.e.t.t|
|.i.n.g.s.......L|
|.a.v.f.5.8...2.0|
|...1.0.0........|
|........ Se.....|
|.......M[......_|
|\D+.W. U[......_|
|\D+........7....|
|.............8..|
|..,.,.......8...|
|....WMV2..^.....|

Just got bit by this crash to desktop bug after getting to the half-way point in the game. It ran silky smooth up until then. :\

After enabling extra tracelogging, by adding +mfplat to usersettings.py, the log contains the following prior to the crash:
14854.850:0114:01cc:trace:mfplat:MFCreateDXGIDeviceManager 0000000020AEEF10, 0000000020AEEF20. 14854.850:0114:01cc:trace:seh:dispatch_exception code=c0000005 flags=0 addr=00000001407AE12E ip=00000001407AE12E tid=01cc 14854.850:0114:01cc:trace:seh:dispatch_exception info[0]=0000000000000000 14854.850:0114:01cc:trace:seh:dispatch_exception info[1]=0000000000000000 14854.850:0114:01cc:warn:seh:dispatch_exception EXCEPTION_ACCESS_VIOLATION exception (code=c0000005) raised
I can see that the function MFCreateDXGIDeviceManager hasn't been implemented, which explains the crash

Hit the halfway spoiler bug. Ouch, that's a long way from a savepoint.

I wonder if <Workaround removed by moderator> for the media foundation stuff will work with some combination of proton versions and it. I tried 6.3-2 and 4.11-13. With 6.3-2 it gives an identical log to running in that version without the script. In 4.11-13 the game wouldn't launch at all. Maybe an older version of Proton GE when some of the media foundation workarounds were baked in? Next time I get a chance I will try some more.

I recall Borderlands having a similar issue where a video being played on a monitor in-game gave people problems. Looking through protondb for that game, someone mentioned that Glorious Eggroll version 6.1-2 made him able to continue the game.
For the heck of it, I tried the same build here, and although the game still crashes when it tries playing the intro video, the produced trace contains significantly more information.

I do realize that Glorious Eggroll is a fork of Proton, so this information probably should go elsewhere, but I think the trace may still provide some value for whomever are tracking this issue, and for anyone currently working on extending the Media Foundation implementation shipped with Wine.

It looks like it instantiates the necessary data structure, then sends chunks of data from the video into a buffer which the device might have to use to play the video, but I suppose the implementation back in Glorious Eggroll 6.1-2 just discarded that data.
Regardless, this seems to run without a problem, and after calling various release and free functions to free up allocated memory, the game then crashes when it calls mfplat::MFllMulDiv

Trace attached from the moment it attempts to play the video
Proton-6.1-GE-2.log

i just want to add. i tested the title video on windows with DXVK and it fails to load the cutscene there aswell.
Game doesnt close it just seem to freeze at a black screen. which confirms the dxgi problem with MFCreateDXGIDeviceManager not being implemented.

Should we perhaps create an issue at the dxvk project? and add a trace log etc.

@MagZu
If the game crashes on WIndows when using DXVK, and works when you aren't using DXVK, then it might be worthwhile opening an issue.
If by WIndows, it's a WIndows 7 install, then it'll probably crash even without DXVK since the version of Media Foundation shipped with WIndows 7 doesn't have the necessary functionality either.
Minimum supported version is Windows 8, as stated by the API documentation

I'd like to point out that MFCreateDXGIDeviceManager is implemented in Wine's mfplat.dll, has been since https://github.com/wine-mirror/wine/commit/645cf84677428fd4541387eed79ae7dbdedb5a4d (2019, made it into a release in Wine 4.15) and the implementation seems fairly complete as well. The issue must lie between these lines because the 2nd TRACE is never printed.

@JacobSvenningsen in my test i did it with Windows 10, but thats good information regarding win7 and 8 in regards to Media Foundation

You're very right.
Proton 6.3 is built against a version of WINE where the function returns E_NOTIMPLEMENTED
Why this patch got merged, I cannot say. Must have had a reason

Interesting. I suppose it'd be worthwhile testing the game with that patch reverted, although I'm almost certain it won't solve the issue, as transplanting MF libraries from a working Windows install doesn't do it either; the game still crashes somewhere along that code path.

After reverting this patch, the game no longer seems to crash when playing videos, but it gets stuck on a black screen.

winegstreamer: error: decodebin0: Your GStreamer installation is missing a plug-in. 12642.632:010c:0210:err:gstreamer:decodebin_parser_init_gst Failed to play stream. winegstreamer: error: decodebin0: ../src-gst_base/gst/playback/gstdecodebin2.c(4719): gst_decode_bin_expose (): /GstBin:bin0/GstDecodeBin:decodebin0: no suitable plugins found: Couldn't set protonvideoconverter0 to READY

steam-1113560.log

@jpgay

Replying to https://github.com/ValveSoftware/Proton/issues/4777#issuecomment-829025920

It seems that the game makes use of the 32bit Media Foundation implementation, so it might be possible that gstreamer fails here because you lack the ugly and bad 32bit plugin packages for gstreamer. This is assuming that Proton makes use of the host machine's gstreamer and not the bundled one.
Based on @Billli11's post, the video is of the WMV2 format, which should require the bad or ugly gstreamer plugins.

Replying to https://github.com/ValveSoftware/Proton/issues/4777#issuecomment-829242199

Proton uses it's own gstreamer, and only comes with the good plugins. I think the avdec_wmv2 plugin is a part of gst_libav, which would also require ffmpeg.

After building GE 6.7 (for bad and ugly gstreamer plugins, I'm almost certain this would also work with upstream Proton) with the patch reverted, I ran into the issue described by @JacobSvenningsen in https://github.com/ValveSoftware/Proton/issues/4777#issuecomment-828713669 where the game tries to call MFllMulDiv and fails. As the function is more or less trivial to implement, I added

// dlls/mfplat/main.c
LONGLONG WINAPI MFllMulDiv(LONGLONG a, LONGLONG b, LONGLONG c, LONGLONG d) {
    return (a * b + d) / c;
}

and replaced

// dlls/mfplat/mfplat.spec
@ stub MFllMulDiv
// with...
@ stdcall MFllMulDiv(int64 int64 int64 int64)

yielding a different error: now the audio plays, but DXVK fails to render the video with

err:   D3D11: Cannot create texture:
err:     Format:  103
err:     Extent:  1920x1080x1
err:     Samples: 1
err:     Layers:  1
err:     Levels:  1
err:     Usage:   28
err:     Flags:   0

Plugging the error into DXVK issue search got me this issue https://github.com/doitsujin/dxvk/issues/1676. I'd dare bet that this is the same issue causing the crash on Windows when using DXVK and maybe even the reason for the original patch stubbing out MFCreateDXGIDeviceManager. I believe this is pretty much the brick wall when it comes to the cutscene issue, and that it should probably be reported to the DXVK developers once it is reproduced by someone other than me.

@aurieh

As the function is more or less trivial to implement, I added

I've wondered about this too. There's so much functionality already implemented for MF, so I really wondered why that function hasn't been. The API documentation even states what it needs to do.
Although in case of overflow it should return LLONG_MAX, while returning LLONG_MIN in case of underflow, but even then it's still pretty trivial.

I believe this is pretty much the brick wall when it comes to the cutscene issue, and that it should probably be reported to the DXVK developers once it is reproduced by someone other than me

At this point, I don't think so. Both here and on Steam, it's been reported that it crashes when using DXVK on Windows, and now it crashes for you too when using Proton after patching it a little yourself. I think it's fair to say it's a DXVK issue. Although from the linked issue, working around it may require a hack.
Opening an issue there may still be worth it

Although in case of overflow it should return LLONG_MAX, while returning LLONG_MIN in case of underflow, but even then it's still pretty trivial.

Indeed; I didn't bother implementing the checks as I was just testing whether the game would run, and there's only one call to this function in the entire media pipeline. With these additional guards added, the patch should be submitted to wine-devel.

Opening an issue there may still be worth it

For sure.

Building proton with all the gstreamer plugins from proton-ge (and ffmpeg), and adding aurieh's MFllMulDiv, results in different errors than with a proton-ge build. Plays audio, displays subtitles, and returns to title screen after the attract movie finishes, but still no video.

8640.549:0110:01b0:fixme:mfplat:source_reader_setup_sample_allocator DXGI device manager is not supported.
8640.549:0110:01b0:fixme:mfplat:audio_mediatype_GetAudioFormat 00000000543680A0.
8640.552:0110:01b0:fixme:mfplat:audio_mediatype_GetAudioFormat 0000000054463E20.
8640.589:0110:01b0:fixme:mfplat:audio_mediatype_GetAudioFormat 0000000054414540.
warn:  D3D11DXGIDevice::QueryInterface: Unknown interface query
warn:  10ec4d5b-975a-4689-b9e4-d0aac30fe333
8640.610:0110:01b0:trace:seh:sigsys_handler SIGSYS, rax 0xf08a, rip 0x14744794b.

steam-1113560.log

So no crashing?
Does that mean you can proceed into the second half of the game after you've made a certain story-related choice?

I should be able too, but I'll have play up to that spot again to test.

I've got a save right at the event. Could test it for you if you have a means of sharing the build.
I've had problems building the project myself, and due to that I tried substituting compiled dlls with the same changes people have done here, just to see if that would work. Those dlls all failed to load, causing instant crashes.
Been quite annoying not being able to help out on that front :(

I've uploaded my build here (mega)
Excuse the build's name, I never grew up.

Confirmed. I checked the video on the main menu to see if it would crash here or not, which it didn't.
After making the story choice, the same thing happened. The video starts playing (although only the sounds are actually played). This allows you to skip the video, save the game and proceed to stage two of the game.

Of course it would be best if the video frames could be rendered too, but this works as a workaround I guess. At least people can now play the game, and probably finish it too

I expect at least the endings of the game will have a similar problem, though. There are a bunch of other ARC files in the movies directory.
(Also, the game just got a 1GB update. Seems to be DLC files only, though.)

If it helps any, the .arc files in the data/movie folder seem to just be plain ASF files after the first 24 bytes. They don't contain any audio though, only video.

If it helps any, the .arc files in the data/movie folder seem to just be plain ASF files after the first 24 bytes. They don't contain any audio though, only video.

would it be possible to perhaps reencode those files with a different codec?

Had a friend try out the MFllMulDiv function on his Windows 10 install, and despite the specification stating that it returns LLONG_MAX if an overflow happens, it only does so if the final result cannot fit in a 64bit integer. It seems like the intermediate result might be stored in a 128bit integer. I assume it works the same way for underflow.

Basically it calculates each step and stores that in 128bit integer, then checks if it can cast that value to a 64bit integer without causing overflow/underflow, then returns the result if it can successfully. Otherwise LLONG_MAX if overflow or LL_MIN if underflow
Anyone planning on making a patch? It seems like quite a chore, especially for first-time contributors

I tried using @jpgay 's build and instead of the game crashing when when trying to load the title screen movie I just get a black screen with no audio and no way to skip whatsoever. The same thing happens for the cutscene after the specific choice you make at the end of Act 1 (i.e. black screen, no sound and no way to skip or remove anything).

I tried using @jpgay 's build and instead of the game crashing when when trying to load the title screen movie I just get a black screen with no audio and no way to skip whatsoever. The same thing happen for the cutscene after the specific choice you make at the end of Act 1 (i.e. black screen, no sound and no way to skip or remove anything).
That's different from what others are experiencing.

Try renaming the user_settings.sample.py file and add +mfplat to the WINEDEBUG line. Be sure to uncomment the line as wel. Then post the log
Maybe something else is necessary on your system

Thanks @JacobSvenningsen, here's the log

steam-1113560.log

Sadly I don't see much from the log. This may be a bug with Navi2 chips, though I cannot say.
Others here that are reporting success, are running at a 1080p resolution. This may be far-fetched, but try with the intro video like that just once.
If that doesn't help, I guess there's a bug somewhere in RadV for Navi2 chips

Yeah switching to 1080p didn't fix the issue either. Probably a Navi2 issue. Hoping that somebody finds a fix then :(
steam-1113560.log

Sadly I don't see much from the log. This may be a bug with Navi2 chips, though I cannot say.
Others here that are reporting success, are running at a 1080p resolution. This may be far-fetched, but try with the intro video like that just once.
If that doesn't help, I guess there's a bug somewhere in RadV for Navi2 chips

Just want to add that i run on 4k resolution and with jpgay's proton build it just skips the cutscene but no frozen black screen.
Using RX 5700 XT

I also tried @jpgay 's build and the game won't start any more. It crashes instantly after clicking the 'Play' button :(

Here's my system specs:

• GPU: NVIDIA GeForce RTX 2060 Rev. A
• Driver: 465.27
• Kernel: 5.11.16-arch1-1

BTW, where can I find the proton log file? Would be helpful if I can provide my log file for you.

in your home directory
~/steam-{AppID}.log
So the log for this game will be located at ~/steam-1113560.log

in your home directory
~/steam-{AppID}.log
So the log for this game will be located at ~/steam-1113560.log

@JacobSvenningsen it seems that there's no log in my home directory. I also checked ~/.steam/logs and there's no file named steam-{APP_ID}.log. All files in ~/.steam/logs are:

bootstrap_log.txt  configstore_log.txt  connection_log.txt  content_log.txt  shader_log.txt  stderr.txt

pull-request up on WineHQ for a proper MFllMulDiv implementation https://source.winehq.org/patches/data/205057
The MFCreateDXGIDeviceManager function isn't stubbed out in mainline Wine either.
So soon enough, the game probably won't crash anymore with mainline Wine+DXVK.

Just built that patch against a community flavour of Proton (Proton-tkg) and the game no longer crashes when playing the cutscene after idling at the "Press Any Button" screen. It doesn't play the video or audio, I can't skip the cutscene, and it doesn't look like it ever "finishes playing" in that it would return to the previous screen. Could be to do with the custom build of Proton though.

At the very least, I can confirm this patch seems to fix the hard crash when playing cutscenes.

Replying to https://github.com/ValveSoftware/Proton/issues/4777#issuecomment-826027599

Have you solved it? at first i could play normally but now i happen to have this green screen. @maverck

Replying to [#4777 (comment)](https://github.com/ValveSoftware/Proton/issues/4777#issuecomment-826027599)

Have you solved it? at first i could play normally but now i happen to have this green screen. @maverck

i was on manjaro and i installed fedora and the game runs fine aside from the cutscene crash.
my assumption was something getting messed up from messing around with mesa-git. maybe a bad lib still linked or something i don't know.

pull-request up on WineHQ for a proper MFllMulDiv implementation https://source.winehq.org/patches/data/205057
The MFCreateDXGIDeviceManager function isn't stubbed out in mainline Wine either.
So soon enough, the game probably won't crash anymore with mainline Wine+DXVK.

Looks like the patches was added in experimental-6.3-20210507 but I still seem to get similar issues with crashing on the title screen. In the patchset from May 4th the the MFCreateDXGIDeviceManager was reverted and readded for some reason, so I guess there are some troubles with removing the stub?
So there seems to still be some issues here, attaching my PROTON_LOG with experimental-6.3-20210507 steam-1113560.log

@axhav
And even if that stub is removed, then upstream proton will probably run into the issue described by @jpgay earlier about missing gstreamer plugins.

@kisak-valve
Do you happen to know why Proton doesn't ship with the bad and ugly gstreamer plugins? License/Patent issues that makes You unable to? Or just want to play it safe?
If so, is it possible to statically link the plugins from the Host if they're present?

It seems like this patch from GloriousEggroll's Proton-6.8-GE-1 fixes the issue for many. I seem to personally not be able to launch the game under anything except proton-experimental, but some have suggested this is a denuvo issue.

If the game launches for you, videos will work for you under that version. Do note that this requires backing up your games wine prefix folder, deleting the original, and then copying your saves back in (or download from steam cloud) as the current prefixes break due to the wine 6.8 update, as well as controller hotplugging being broken due to that same update

@kisak-valve is this something that can be incorporated or are there legal/logistical issues that would prevent this?

That seems like such a trivial patch, I hope it could be incorporated into Valve's Proton. Would be nice to see this game actually join Automata as an officially-supported SteamPlay title.
Thank you for the head's up, @Semperverus. I don't use Proton-GE but as NieR is one of my favourite games of all time, I'm about to install it and give it a go.

@adolson absolutely, happy to help. I hope it gets incorporated too. (Let me know if it boots for you with GE proton)

Latest Proton does help here. Do read the release notes though.
I had to delete the proton prefix, otherwise it would just crash when videos started playing.
DON'T JUST SHIFT-DEL it, since that'll delete your save files too.
Save file is stored within /1113560/pfx/drive_c/users/steamuser/My Documents/My Games/NieR Replicant ver.1.22474487139/Steam/
Keep that folder but delete everything else

@adolson absolutely, happy to help. I hope it gets incorporated too. (Let me know if it boots for you with GE proton)

Yeah, it worked. I re-did the mid-way battle and progressed as expected. :)

Latest Proton does help here. Do read the release notes though.

Latest official Proton works, now?
My save file was sync'd to the cloud, I guess. (I moved the entire old prefix, rather than deleting, just in case I needed to restore it.)

For those that don't check the DXVK issue, DXVK can now read the video in it original format (on Window at least)

mfplat still needs some fixes for this to work; with current upstream wine + the DXVK branch above it renders video, but it's all green because wine seems to forget to upload the chroma plane for NV12 textures.

Just got the game and wanted to start playing, tried what was suggested, like proton-ge in the latest version, which got me after a few tries into the main menu and even into the game where I was met with the green screen

Tested with Proton-6.8-GE-1 and Proton-6.9-GE-2 and Proton Experimental got the same issues on all of them

Specs:
Linux tyrus 5.12.9-arch1-1 #1 SMP PREEMPT Thu, 03 Jun 2021 11:36:13 +0000 x86_64 GNU/Linux
AMD Ryzen 5 2600 Six-Core Processor
AMD Radeon RX 590 ( Driver Version: 4.6 (Compatibility Profile) Mesa 21.1.1)

Replying to https://github.com/ValveSoftware/Proton/issues/4777#issuecomment-826027599

like here, my screenshot:

my game log: https://gist.github.com/ottersarecool/adfc332dd739724db7a78979f8f001e9

sadly can't test further for the next 24 hours since Denuvo already locked me out, god I love these Austrians (NOT)

Replying to https://github.com/ValveSoftware/Proton/issues/4777#issuecomment-854466008

I was able to resolve the greenscreen issue by removing the package amdvlk

The game has become hardly playable for me as it fluctuates around 30fps and runs very choppy. Used to run just fine with stable 60fps tho, but haven't payed it in a while since then.

It ran fine again once I switched from Proton Experimental to Proton GE 6.10, but that only lasted until I restarted the game. I also fixed it once by simply reinstalling it, but that also only worked once. Switching around between different Proton version didn't help either, removing the prefix and shaderache (in the Steam directory) also didn't do anything.

So, no idea what causes it or what exactly fixes it :(

I'm on Arch Linux (Kernel 5.12.14) and Radeon VII with Mesa 21.1.3.

EDIT:

I've noticed desktop animations being choppy as well while the game is running. I've disabled vsync in Kwin via environment variables (since AFAIK with Kwin X11 there's no other way anymore to disable it) :

KWIN_X11_REFRESH_RATE=144000
KWIN_X11_NO_SYNC_TO_VBLANK=1
KWIN_X11_FORCE_SOFTWARE_VSYNC=1

For now, this seems to have fixed the issue, even after restarting the game.

@alex9k1 Kwin's desktop effects should be disabled while playing, otherwise you'll get stutter and increased latency pretty much everywhere.

That said, the game is known not to play nice with any sort of GPU power management because it synchronizes with the GPU multiple times per frame (which is something no game should ever do, but oh well), so having too much GPU power on something like an RX 6000 GPU will cause issues because it doesn't clock up properly. On my system I need to do echo high > /sys/class/drm/card0/device/power_dpm_force_performance_level in order to be able to play the game at 60 FPS.

@doitsujin Yeah, I usually just disable compositing when I run into issues, which also used to disable vsync as well a few versions back. Although I have to admit, that it hasn't caused any issues for me in quite a while.

I've set set the powerlevel to high, but it didn't help. After disabling vsync with the environment variables however, powerlevel auto is just working fine.

Still wondering why it was running fine 2 times before (one time after changing the Proton version and one time after reinstalling it), but was running bad again after I restarted the game 🤔

EDIT: I've also tried it on Wayland with Kwin - also choppy :/

The issue is back :( Even tried to run it with vblank_mode=0 and Proton GE 6.12-1 but nothing helped.

One thing I've noticed tho is, that this time the game was ~45fps initially. I deleted the prefix and started the game again and now it's at ~35fps

When I change the refresh rate of my display from 144Hz to 60Hz it's back to ~45fps. Noticed, when doing this, the FPS can actually go to stable 60 if I just look at the ground. It kinda doubt that my system (Ryzen 2700X, Radeon VII) isn't fast enough (running the game on medium - high settings), since it ran fine before. GPU load also seems rather relaxed.

Updated to Proton GE 6.13-1, had stable 60fps again even after restarting the game. A day later it was ~45fps again :/

• GPU: nvidia 1070
• Driver/LLVM version: 495.29.05
• Kernel version: 5.14.12

This game appears broken on all recent proton releases. The last working release appears to 5.0-10 (Broken releases as of my testing: 5.13-6, 6.3-7, experimental). Performance on 5.0-10 seems fine though.

On broken releases the game crashes before a log can be written.

Can't confirm, game runs fine here on Experimental.

Denuvo was recently removed, right ?
I can also confirm that the game still works... and I still have the issue with it running in slow motion and capped at 20-30 fps on my 980Ti with low gpu usage :/

It seems I was affected by #5118. After testing older proton versions my prefix was rendered unusable for 5.13+. Deleting the compatdata (warning: loses your save data) fixed the issue if the game not starting before logs could be written.

--- edit
Performance is also fine on 6.3-7 after rebuilding the compatdata directories (At least in facade and seafront where I had saves)

Hello, I am affected by this issue on Proton 6.3.8. I only rushed through the comments on this issue, so I'm not sure if this was fixed already and is waiting to be merged
https://github.com/ValveSoftware/Proton/issues/4777#issuecomment-829025920

Hi,

I'm also affected by the problem with cut scenes.
I've tried both proton 6.3.8 and proton-experimental, cleaning the pfx each times (also tried various Glorious Eggroll version without success).

What I experienced with the video at Start Screen (when you wait 30 s without touching anything) :

• If I enable shader cache, the video begins to play with sound for like 3s then return to Start Screen
• Else, the game just freeze

As for the cut scene at the end of Chapter 1, the game freezes all the time and I just can't pass it.

I remarked an interesting log when trying proton-experimental :

winegstreamer error: decodebin0: Your GStreamer installation is missing a plug-in.
winegstreamer error: decodebin0: ../src-gst_base/gst/playback/gstdecodebin2.c(4719): gst_decode_bin_expose (): /GstBin:bin0/GstDecodeBin:decodebin0:
no suitable plugins found:

When trying proton-GE variant (for all plugins) :

GStreamer-WARNING **: 12:24:22.252: Failed to load plugin '${HOME}/.local/share/Steam/compatibilitytools.d/Proton-6.8-GE-1/files/lib64/gstreamer-1.0/libgstdash.so': ${HOME}/.local/share/Steam/compatibilitytools.d/Proton-6.8-GE-1/files/lib64/gstreamer-1.0/libgstdash.so: bad ELF class : ELFCLASS64

So from what I understand, there is a problem with Gstreamer,
From what I read, the videos are in a wmv2 format, so what plugin does it need ? Shall I install it on my computer or it should be included in proton ?

Thank you

Specs :

Graphic Card : AMD Radeon RX 580
O.S. : Solus OS
Kernel : 5.14.16
Mesa lib : 21.2.5

@Keyhaku Try using a newer release of Proton-GE to start with, the logs you posted indicate you're using 6.8-GE, the most recent version at the moment is 6.21-GE-2.

As for plugins, try installing the gst-plugins-bad and/or gst-plugins-ugly package. I'm unfamiliar with Solus so I'm not sure how you might install them. I also thought GE packaged it's own GStreamer plugins but perhaps not. I remember having to install one of those packages on Arch but a few months ago it worked out of the box with a new-ish GE release.

You might want to reach out to someone on the Proton-GE Discord server if it still doesn't work, as issues with that release are handled by GE. Hopefully you get it sorted!

I'm not sure it has anything to do with what plugin is installed on my system, since it seems GStreamer 1.8.4 is shipped with proton.

Here is the log with proton-6.3.8 when my system is trying to play the movie (we can see that GStreamer is loaded from proton and not from my system) :
steam-1113560-relevant-part.log

These are parts that I think are relevant :
1/ The video seems to be recognised as MIME type x-ms-asf : GST_EVENT gstevent.c:847:gst_event_new_caps: creating caps event video/x-ms-asf

2/ The plugin specialised in media converting is well charged from proton : GST_PLUGIN_LOADING gstplugin.c:915:_priv_gst_plugin_load_file_for_registry: plugin "${HOME}/.local/share/Steam/steamapps/common/Proton 6.3/dist/lib64/gstreamer-1.0/libprotonmediaconverter.so" loaded

3/ No plugin is found to decode the stream :

decodebin gstdecodebin2.c:4719:gst_decode_bin_expose:<decodebin0> error: no suitable plugins found:
Couldn't set protonvideoconverter0 to READY

0x7f1ad8003140 [INFO]  GST_ERROR_SYSTEM gstelement.c:2234:gst_element_message_full_with_details:<decodebin0>posting message: Your GStreamer installation is missing a plug-in.
winegstreamer error: decodebin0: Your GStreamer installation is missing a plug-in.
winegstreamer error: decodebin0: ../src-gst_base/gst/playback/gstdecodebin2.c(4719): gst_decode_bin_expose (): /GstBin:bin0/GstDecodeBin:decodebin0:
no suitable plugins found:
Couldn't set protonvideoconverter0 to READY

|ERROR] WINE wg_parser.c:2287:decodebin_parser_init_gst: Failed to play stream
warn:mfplat:media_source_constructor Failed to construct MFMediaSource, hr 0x80004005.

I've made some more test with proton-experimental and the error messages are the same, nevertheless I put the log here.
steam-1113560-proton-experimental.log

I have different error messages with Proton-GE-6.21-2 (though the game still freeze),
I don't have tons of warning with wrong ELF Class when loading gstreamer plugins.

The video format is well identified as wmv2 :

asfdemux gstasfdemux.c:2948:gst_asf_demux_add_video_stream: Adding video stream #0, id 1, codec WMV2 (0x32564d57)

GST_EVENT gstevent.c:847:gst_event_new_caps: creating caps event video/x-wmv, wmvversion=(int)2, width=(int)1920, height=(int)1080, codec_data=(buffer)e7ffb480, format=(string)WMV2

But for an unknown reason the data structure seems incorrectly filled :

[WARN] structure gststructure.c:1023:gst_structure_set_field: Trying to set NULL string on field 'language-code' on taglist. Please file a bug.

Or is it something else ?
[WARN] asfdemux gstasfdemux.c:927:asf_demux_peek_object:<asfdemux0> Unknown object 00000000-00000000-00000000-00000000

Anyway I put here the log.

Thank you
steam-1113560-proton-GE-621-2.log

I've finally figured it out, after increasing log verbosity using GST_DEBUG environment variable, I saw that GStreamer couldn't read the first 64 bytes of the videos. So i though that maybe the videos were corrupted, after launching a check via Steam, the videos were redownloaded and now the videos are well played. I finnaly passed to Chapter 2, hourra !

So from what I understood, this game can only be played using a proton-GE build since the Gstreamer plugin needed to read the WVM 2 format is only available in the "ugly" plugins. But for licencing issue these plugins cannot be included in "Valve" proton, am i right ?

@alex9k1 Kwin's desktop effects should be disabled while playing, otherwise you'll get stutter and increased latency pretty much everywhere.

That said, the game is known not to play nice with any sort of GPU power management because it synchronizes with the GPU multiple times per frame (which is something no game should ever do, but oh well), so having too much GPU power on something like an RX 6000 GPU will cause issues because it doesn't clock up properly. On my system I need to do echo high > /sys/class/drm/card0/device/power_dpm_force_performance_level in order to be able to play the game at 60 FPS.

Just tried that on my RX 6600, clock

speed was at 2,6GHz constant, and NieR was running at ~46fps

the green line looks suspicious to me, I tried all that was mentioned here but was not able to get stable 60FPS with my RX 6600 in this specific game.

After some tinkering and reading the Proton readme I was able to resolve the FPS issue by simply adding PROTON_USE_WINED3D=1 %command% to my launch options, so I believe this is some kind of dxvk related thing after all with my specific gpu maybe, with opengl it works very good, it might help to try this out if someone else has this weird stuck framerate at around 46.

I can't reproduce the problem here.

Turns out that there is a game bug that also happens on Windows on AMD GPUs when Resizeable BAR is enabled that causes low performance. I added a workaround to DXVK.

Wow, just as a heads up and maybe it helps someone else: the bleeding-edge branch of Proton Experimental has fixed the fps drop issue completely on my 980Ti ! I can finally play the game on linux at the intended speed.

I also tested the new Proton 7.0-1 but the issue is back again unfortunately. So, the issue is present for the normal Proton Experimental without bleeding-edge branch, 6.3-8, 7.0-1, all Glorious Eggroll's proton versions.

I'm not sure what on the bleeding-edge branch has caused this change.

Turns out that there is a game bug that also happens on Windows on AMD GPUs when Resizeable BAR is enabled that causes low performance. I added a workaround to DXVK.

Oh dear, I love you so much for figuring this out @doitsujin . That's exactly the workaround that fixes the issue for me on all proton versions it seems!

I dropped a dxvk.conf next to the game's executable with d3d11.apitraceMode = True and can confirm the fps issue goes away regardless of the proton version. Thanks!

(Still... I'm intrigued by the fact that the bleeding-edge branch has some sauce that made it work too)

@MagZu If the game crashes on WIndows when using DXVK, and works when you aren't using DXVK, then it might be worthwhile opening an issue. If by WIndows, it's a WIndows 7 install, then it'll probably crash even without DXVK since the version of Media Foundation shipped with WIndows 7 doesn't have the necessary functionality either. Minimum supported version is Windows 8, as stated by the API documentation

In Windows 7, the function MFCreateDXGIDeviceManager is implemented in the dll mshtmlmedia.dll and the game will play videos if this dll is found.

Can someone test this game in Windows 7 mode in Proton? As in, doing winecfg in the prefix and changing the Windows version to 7. The videos will be black but should play if you add mshtmlmedia to system32 and syswow64.

Changing the Windows version can change the behavior of the game so it's worth experimenting to see if it has any benefits (performance, bugs, etc.).

Games window is offset in a really odd way when using fullscreen:
(screenshot spans my 3 monitors, the edges of the other 2 windows are the monitor edges)

You can see my background bleeding through on the top and right sides of the game window.

Using Fedora 38, Mesa 23.0.3 with a AMD Radeon 6700 XT, and tried on Proton versions 7.0-6, 8.0-2, and Proton Experimental. Same issue appeared in all versions.
My Steam System Info is here: https://gist.github.com/CorruptComputer/f164c2d48606808dbfbc07ee8810efd5
It is out of date because the new Steam UI doesn't let you copy the text from this window:

• GPU: RX 6500 XT
• Driver/LLVM version: Mesa 22.2.5
• Kernel version: 5.15.0-71

The cutscene crash is back in current proton versions : intro video at title scene + cutscene after grimoire noire fight + cutscene after wendy fight + cutscene at shadowlord castle + cutscene towards the end of route E. All those except the intro video are blocking game progression.

Tested on :

• Proton 8.0-2
• Proton-experimental
• GE-Proton8-3
• GE-Proton7-55

For anyone having issues with pre-rendered cutscenes not displaying correctly (like showing a test screen for a few seconds instead of the actual cutscene) with the newest GE-Proton version:

Either input this as a startup command:
PROTON_AUDIO_CONVERT=0 PROTON_AUDIO_CONVERT_BIN=0 PROTON_VIDEO_CONVERT=0 PROTON_DEMUX=0 %command%
or
use GE-Proton 9-13 or older

Edit:
Pre-rendered cutscenes can also be played with official Proton when shader pre-caching is enabled. Do note however, that at the time of writing this, the pre-rendered cutscenes only work flawlessly on stable Proton 9.0-3. Using Proton Experimental (even bleeding-edge) has a video and audio desync problem.

@uyopi the issue should be solved with proton bleeding edge https://github.com/ValveSoftware/Proton/wiki/Proton-Versions#proton-bleeding-edge