[TODO] Play around with VIIPER to possibly replace ViGEmBus use

[Ryochan7]

I wrote off VIIPER as a viable alternative to ViGEmBus beforehand due to issues with the required usbip-win2 driver. The driver runs horribly on my system by default and introduces significant system slowdown just having it in the background without any virtual devices created. Finally got around to downloading the source code and building it myself to play around with compiler settings; Visual Studio default settings for Release builds of C/C++ projects are bad. Merely changing build settings is enough to make the drivers much more viable. Not sure about usable yet as I have not tried to install SISR again.

https://github.com/Alia5/VIIPER

[Ryochan7]

tbf to usbip-win2, ViGEmBus also did not meet my initial requirements until I got involved. Had to change the driver and library (C and .NET versions) to make it perform on par with ScpVBus; even later I found a way to make ScpVBus outperform again. The library changes (used in DS4Windows) were never introduced in the vanilla library because Nef did not want them so vanilla ViGEmBus (and consequently VirtualPad) never met my standards.

[Ryochan7]

Started with SC2SmoothBrainifier as a basis for experimenting. Got the program working after fixing the initial puck controller check. Tested various changes with the drivers. That led to experimenting with how to change libVIIPER as that seemed like a choke point.

Not sure the toolkit used to build the official library release but it fails in a couple of spots on my end. It might be worth cleaning up and submitting a PR. Anyway, some very basic support for libVIIPER was added to this program locally.

I was not too thrilled with the initial performance so that led to having to read through the source code. I have never used the language Go and it has been difficult to find where types are defined. I keep getting tripped up by the syntax too. Some progress has been made but I don't know how much further I can take this.

https://github.com/Mennenth/SC2SmoothBrainifier

[Ryochan7]

Not sure why but the docs seem to be wrong about the default write buffer timeout when it comes to the library. Explicitly setting a value made a significant difference. Performance now seems to be on par with ViGEmBus if not slightly better. I have not enabled a rumble feedback callback but I would hope that would not affect input state reports.

Still need to play around with it more. The current progress is a bit more promising.

[Trihy]

Did you took a look at hidmaestro? Seems promising.

Good to see you back.

[Ryochan7]

I have never heard of the hidmaestro driver before. I will have to try it out later but I am wondering how the driver gets around to making multiple virtual devices while being a user-mode driver; from what I have seen, a KMDF driver is required to build a bus driver to make that work for virtual devices.

VIIPER seems to be the major hype driver right now thanks to the SISR project. So far, I have run into a few small hurdles (CPU usage because it is polling based) but it mostly works. Force feedback also works. I am not sure I will end up moving over to using VIIPER at this time but it is an interesting test. It would be nice to have a maintained driver and I failed at getting something going.

[Trihy]

Vigem is a dead end. There is no interest at all to fix the bsod problems. Not sure why.

I also read the dev had a talk with nefarious about doing something together, but I guess it didn't went too far and started hidmaestro.

So we will have to move on. Let's see how it goes. Interesting times for virtual pad emulation.

[Ryochan7]

I don't know how much money Nef gets from DSX but I would assume that is a big reason he would not want to collab with anyone other than Paliverse. The community version of VirtualPad (without DualSense emulation) that was supposedly being worked out got memory holed pretty quick.

I tried PadForge and the HidMaestro driver and I did not like either. PadForge could be interesting as it seems like a successor to the older Universal Control Remapper program. However, the app runs slow, the mappings to Xbox 360 from DualSense do not work (wrong stick values, wrong trigger values, B binding does not work for some reason) and input latency is worse than Steam Input. It is not a completely objective comparison but PadForge feels slower than SC2SmoothBrainifier with a puck connected Steam Controlller despite SC2SmoothBrainifier being vibe coded.

Other than being able to digitally sign a few releases of this program, I wasted money getting an LLC registered. It is up for renewal next month but I might just opt to dissolve the business. I definitely do not have the money to renew my EV certificate for another year. I was hoping for some more autonomy for my software.

[Ryochan7]

Finally figured out a way to force flushing so VIIPER can mimic autosubmitreport (or whatever it is called in ViGEmBus). I think the buffered report writer might still be needed for other routines but at least now the flush timeout can be set longer and not have such an impact on button responsiveness. CPU usage is also reduced.

[Ryochan7]

I do not think I can spend any more time on this. I ended up getting virtual DS4 support enabled as well. There are at least a couple of issues I should file with the VIIPER project.

I have put the prototype code out there. Not sure if anything will ever come of this.

https://github.com/Ryochan7/DS4MapperTest/tree/viiper_testing

[Ryochan7]

To answer my own question from before, HidMaestro does not seem to allow dynamic plugging in of output devices. Any change in output device configuration requires the HidMaestro driver to be restarted; PadForge takes care of writing the current device configuration to the registry and restarting HidMaestro. That is how the program gets around not being able to create a bus driver in UMDF. A bit more inconvenient but it works.

[Trihy]

But maybe a final non bsod build of vigem was worth it to being released (from nefarious). But I guess was not planed at all.

Nice review about hidmaestro and viiper. Hid gets frequent updates and devs seem open for ideas.

I was looking for a vigem replacement for touchmote but since it's stuck at FW4.81 I'm not sure if I could use any of those. Was about to read the docs after finishing some mods to tmote.

[Ryochan7]

libVIIPER should be viable for a .NET Framework project. It is a C library (converted from Go). There is no C# wrapper so it needs to be created in a project.

HidMasetro would be harder to try to incorporate. I do not believe there is a wrapper so the client app has to create the HID descriptors for the virtual devices.

[Ryochan7]

I stand corrected. https://github.com/hifihedgehog/HIDMaestro/tree/master/sdk/HIDMaestro.Core

[Trihy]

Need to read the docs for that (viiper and hidmaestro) Been making some experiments adding motion support, DSU server, etc. The motion dongle is a pain to work on. Next step will be probably replacing vigem.

BTW read somewhere you were who added DSU support for DS4Win. Wish I knew that earlier : D. Been scratching my head to match wiimote accel and gyro orientations among which units and scale to use. I found a program called PadTest_1011 which helped with that.