Plugins + Scripts VAMMoan 2 (teaser)

[hazmhox]

Side note: I'm what you can call a "tech artist" or a "front end coder". I'm okish with code, and I can manipulate systems quite easily. But I'm not a "low level" guy. Anything very low touching to memory management and all that crap is above my paygrade.

I'm very careful with what I do, test a lot, compare perfs, be careful to optimization, and tend to read a lot to understand the best practices. But I'm probably doing some things wrong in some areas.

But in general, I think I tend to release at least "ok tier" plugins that won't kill your VAM (if you use VAM properly).

 

[TBD]

Well yes, I know how to mitigate or work around these issues. As a content creator however, I don't expect users to be able to as well and want to make their experience as robust as possible.
Even if I add a big red warning message to my hub post/update, the vast majority of users won't see or read that. They'll either run into issues, be disappointed and delete the resource, or bother me with support requests.

My personal "play" environment of vam can handle hours of uptime and lots of scene changes without hiccups.
The only issues I have are with some creators' CUAs breaking things during deconstruction. E.g. some of the custom shader code in assets can brick skin renders until a hard reset and I know of some realfeel assets, which take 20+ minutes to unload. In all cases, some errors and exceptions appear in vam's debug log (not message log), which asset creators either don't notice or don't care about.

 

[TBD]

Side note: I'm what you can call a "tech artist" or a "front end coder". I'm okish with code, and I can manipulate systems quite easily. But I'm not a "low level" guy.

Side note: I'm a software engineer slash systems architect and I'm used to taking things apart whenever their internal behaviour does something unexpected.
I could never create features like vammoan2's audio, or any CUA/clothing/younameit for that matter, but I can offer my help in terms of software architecture and low level .net stuff.
Both skillsets are required when you're shooting for perfection so please feel free to "use" me.

 

[hazmhox]

Side note: I'm a software engineer slash systems architect and I'm used to taking things apart whenever their internal behaviour does something unexpected.
I could never create features like vammoan2's audio, or any CUA/clothing/younameit for that matter, but I can offer my help in terms of software architecture and low level .net stuff.
Both skillsets are required when you're shooting for perfection so please feel free to "use" me.

Cool! Duely noted : D

The only issues I have are with some creators' CUAs breaking things during deconstruction. E.g. some of the custom shader code in assets can brick skin renders until a hard reset and I know of some realfeel assets, which take 20+ minutes to unload. In all cases, some errors and exceptions appear in vam's debug log (not message log), which asset creators either don't notice or don't care about.

That is very weird! That said, I'm not surprised

I've always bitched about the lack of understanding of Unity's ( or even video game at a larger scale ) pipeline and people not caring much about the end result (or trying to learn the ropes). I remember years ago, there was a trend on scanned assets for several months. Like stupidly bad assets/topology... and people shooting like 20+ million polys rooms/enviros. The result was so bad in VAM that I always facepalmed about the fact that there was no way it wasn't the case on their end. The moment you slapped a light on that, the shadow casting was so stupid the you dropped to 15fps on desktop... yet, they shared that crap.

Also, the "use" of things (whatever thing) without paying too much attention. For instance I noticed people using VAMifier on my enviros. Which is pointless since they are all made of custom shaders which are ignored by the plugin. Yet, the plugin is there, scan for the whole hierarchy to simply ignore everything.
This is some useless overhead in a scene which as zero impact on the final aspect. And yet you can find instances of people saying "I think it looks better!" when the log (or simply just the visuals : 'D) clearly shows nothing has changed

But what you say is true, I do the exact same, I try to mitigate as much as I can... but over the years I realized it's impossible due to the very heterogenous nature of a scene in VAM. The amount of assets/plugins can be insane and also scripted in a non very optimized manner.

So the best I can do is give knowledge by documenting my plugins or giving advices through guides. At some point, the best intentions you and I (or some other advanced creators) have on our plugins, will not fix the entire ecosystem which is by nature, kind of unstable.

 

[TBD]

Yeah, all we can do is keep on struggling.

Like stupidly bad assets/topology... and people shooting like 20+ million polys rooms/enviros.

Haha, I remember a bedroom asset not long ago, which ran fine only after I deleted the bed cover (or avoided looking at it). I don't know if it was scanned but the poly count was through the roof.

 

[Thavus]

I'm not doing any cache, it's pure assetbundle load and storing references in Dictionaries.

One thing to get used to when starting VAM (especially if you code/do big scenes), is to hit hard reset pretty much all the time. The assetbundle system is handle by Unity and the way VAM handles it is quite strange... I've had a shit ton of problems with assetbundles/assets from assetbundles being locked into memory for whatever reason.

To updated plugins in your scene, use that method. It shouldn't behave differently than your previous version when loading it.
Also, very quickly, I understood VAM is quite bad at managing memory. More or less, I never play/work on stuffs "back to back".

Code a plugin with assetbundle?

• Make my default working scene
• Save, test plugin.
• Modify code, reload plugin
• Save scene, hard reset, reload scene
• Rinse and repeat

Playing a big scene from a cool creator?

• Load scene once, let everything cache
• Hard reset
• Reload scene and play
• After scene playback, hard reset

Most of the problems I often see/do support for on the forums, are people playing an hour or more, loading dozens of scene back to back which makes the game perform worse and worse. Also, plugins can contribute to memory issues, as there are people probably not optimizing enough or forgetting about releasing assetbundles properly.

With the hard reset method, you can play/work/code on VAM for 4,5,6 hours or more without ever having issues or a crash to desktop.
If you don't, you will have problems.

You will eventually even witness some weird stuffs like your scene behaving super weird on animations or characters, and simply exiting the game and restarting it will fix the problem.

I never found reasons or explainations for this, having a very "tidy" way of working with VAM makes everything work nicely. If you don't, you will probably end up smashing your head against the wall (like I did several times over the years before understanding the phenomenon ^^)

Sounds like VAM has issues with memory leaks. Unity has a built in Memory Profiler which can be utilized to identify and fix such leaks. The developer of VAM would need to use that to figure where these problems are. Maybe they are already aware, but are busy working on the next version of VAM though.

 

[hazmhox]

Sounds like VAM has issues with memory leaks. Unity has a built in Memory Profiler which can be utilized to identify and fix such leaks. The developer of VAM would need to use that to figure where these problems are. Maybe they are already aware, but are busy working on the next version of VAM though.

Yeah I know I'm working with meshed

As I was saying '[...]the entire ecosystem which is by nature, kind of unstable.'

The sandbox nature of VAM + the ability to code yourself at a fairly low level (compared to pretty much any game with modding capabilities nowadays) + the ton of content made by people who don't really have an idea of what they're doing = problems.

No matter how hard you'd hit the profiler. You can never control someone loading 1500 shaders and 3500 materials in an assetbundle, and how they're gonna deal with all that data. This is probably a problem VAM will always have sadly.

 

[TBD]

Maybe something like this would be the best approach, both in terms of robustness and performance.

private static readonly HashSet<string> _allowedColliders = new HashSet<string>
{
  //list the correct collider names for e.g. hands, paddles, vrcontrollers here
  "some collider"
};

private static bool ShouldTriggerSpank(ContactPoint contactPoint) => _allowedColliders.Contains(contactPoint.otherCollider.name);

What do you think of this approach, hazmhox? I could implement that if you like. We'd just have to discuss which colliders to accept.

 

[TBD]

A temporary workaround could be to increase the required velocity to 1 or 1.5.
I would like to use vamm2's spanking sounds in a pending update to The Grind of your Life but the random spank playback during sex makes it awkward.

 

[hazmhox]

What do you think of this approach, hazmhox? I could implement that if you like. We'd just have to discuss which colliders to accept.

It seems nice, but do you think it could handle CUAs ?
I'm seeing miles away people would expect any CUA to work with it potentially.

Edit: your scene looks nice btw, I should try it

 

[TBD]

Yeah, CUA colliders can be named pretty much anything. Handling atom types differently isn't crazy expensive. If we don't want a configuration UI for the filter logic, we need to come up with a good default behaviour.
For my use case, something simple like filtering out all person colliders but hands would suffice. For a more holistic approach, we would need to come up with a table of valid/invalid atom types and collider names. E.g. persons' hands, vr controllers, paddles, all CUAs (?) but not things like slates, geometry objects or CollisionTriggers.

PS: thanks, you should

 

[hazmhox]

Yeah, CUA colliders can be named pretty much anything.

Yeah lol I know ^^
I was more refering to a "global" filter to authorize pretty much any CUA without discrimination. I know that no one will ever read or try to normalize their asset creation based on a plugin (for instance, specific naming). I did that for the throwable script, I don't think I've seen a single asset using it.

Handling atom types differently isn't crazy expensive. If we don't want a configuration UI for the filter logic, we need to come up with a good default behaviour.

That would be nice!

For a more holistic approach, we would need to come up with a table of valid/invalid atom types and collider names. E.g. persons' hands, vr controllers, paddles, all CUAs (?) but not things like slates, geometry objects or CollisionTriggers.

Why not! what do you propose?

PS: thanks, you should

I will !

 

[VamEssentials]

This is really awesome! I finally got around to testing it out today. Here are some feedback/enhancement requests...

Increase Character Limit from 32 to 150
I got around to testing this out today and 2 challenges I had were when I imported in some audio files I have, the 32 character limit blocked almost all of them. I tested it on AudioMate and they worked on that. From all the audio assetbundles I have seen a lot of them will use full sentences to describe the audio. One audio file I have is a sentence with 104 characters and its 22 words with number 1 at end. That's probably one of the longest. I'd suggest maybe increasing it to 150. One issue with audio/assetbundles is people naming audio files 1, 2, 3, 4, 5 and you can't really tell what is what to easily filter out what you want or don't want. So it would encourage more descriptive lines. As of right now I would have to go through and rename a lot of audio files to use it with this so I'd likely stick with AudioMate based on this.

Add Auto Play/Continuous Play Option on Folders like AudioMate
It would also be nice to have an option to auto play/play continuous on the folder versus a trigger. AudioMate has that option in their settings. Only downside to theirs is no gap interval option that I am aware of so it's back to back, when it could be maybe spaced out to play next in 5 seconds/10 seconds etc. which would seem more natural I think.

 

[hazmhox]

Increase Character Limit from 32 to 150

Windows has a max path length of 256. I won't push the character limit at 150 as I created that limitation because I immediately had issue by simply grabbing an example voice file from Freesound which had a huge name while I was deving the plugin.

For instance, my path for a scene goes in general like this : _VaM\Saves\scene\hazmhox\ReadOnlyMemories\assets\audio\VA\Selene this is without counting the full path. The full path is 112 chars long. If I put a 150 letters long filename, including ext, it's 266 chars, which will create problems in the explorer and potentially file system commands.

I'm fail-proofing the system for people not aware enough of the file system limitations. Because the moment you go "no one is gonna hit that". The next thing you know, someone reports you a critical error you don't understand the origin, when it was simply a path issue.

I can give you 48 (or maybe 64), I'll do some tests. But that's it.

If you're planning on seriously doing VA, I will recommend to work as we do in the industry with VA file. Short prefixes, identifiers and potentially hashes. For instance : Char_DialogID_LineID ( ex : Rebecca_DialogA_Line01 ). And it could be even shorter as you could sort the folder with the character name since you have only one folder allowed per char : )

Add Auto Play/Continuous Play Option on Folders like AudioMate

VAMM2 VA being a single folder per char. How would a continuous play option even work? If you have let's say, 10 dialogs in a scene, the system would spit all the dialogs in a row (which means 10 times "amount of lines per dialog").

I'd suggest to combine your files into one single file (even if you have multiple lines in reality) if you want a continuous playback. On top of that you'll have more control over the pace.


Overall, I will dive more into VA features IF people start using it and bring scenes with the current existing features. Which is not the case at the moment... I haven't seen a single scene using proper VA implementation with VAMM2.

 

[TBD]

what do you propose?

atom type	rigidbody name	collider name
Paddle	object	Collider
Paddle	object	Collider1
VRController	HandPhysical	*
VRController	L_Finger_Index_A	L_Finger_Index_A
VRController	L_Finger_Index_B	L_Finger_Index_B
VRController	L_Finger_Index_C	L_Finger_Index_C
VRController	L_Finger_Middle_A	L_Finger_Middle_A
VRController	L_Finger_Middle_B	L_Finger_Middle_B
VRController	L_Finger_Middle_C	L_Finger_Middle_C
VRController	L_Finger_Pinky_A	L_Finger_Pinky_A
VRController	L_Finger_Pinky_B	L_Finger_Pinky_B
VRController	L_Finger_Pinky_C	L_Finger_Pinky_C
VRController	L_Finger_Ring_A	L_Finger_Ring_A
VRController	L_Finger_Ring_B	L_Finger_Ring_B
VRController	L_Finger_Ring_C	L_Finger_Ring_C
VRController	L_Finger_Thumb_A	L_Finger_Thumb_A
VRController	L_Finger_Thumb_B	L_Finger_Thumb_B
VRController	L_Finger_Thumb_C	L_Finger_Thumb_C
VRController	R_Finger_Index_A	R_Finger_Index_A
VRController	R_Finger_Index_B	R_Finger_Index_B
VRController	R_Finger_Index_C	R_Finger_Index_C
VRController	R_Finger_Middle_A	R_Finger_Middle_A
VRController	R_Finger_Middle_B	R_Finger_Middle_B
VRController	R_Finger_Middle_C	R_Finger_Middle_C
VRController	R_Finger_Pinky_A	R_Finger_Pinky_A
VRController	R_Finger_Pinky_B	R_Finger_Pinky_B
VRController	R_Finger_Pinky_C	R_Finger_Pinky_C
VRController	R_Finger_Ring_A	R_Finger_Ring_A
VRController	R_Finger_Ring_B	R_Finger_Ring_B
VRController	R_Finger_Ring_C	R_Finger_Ring_C
VRController	R_Finger_Thumb_A	R_Finger_Thumb_A
VRController	R_Finger_Thumb_B	R_Finger_Thumb_B
VRController	R_Finger_Thumb_C	R_Finger_Thumb_C
CustomUnityAsset	*	*
Dildo	b1	*
Dildo	b2	*
Dildo	b3	*
AptBook01	object	Import
AptBook02	object	Import
CyberpunkKeyboard	object	Import
CyberpunkLaptop	object	Laptop_Base
Person	lThumb1	_Collider
Person	lThumb2	_Collider
Person	lThumb3	_Collider
Person	lIndex1	_Collider
Person	lIndex2	_Collider
Person	lIndex3	_Collider
Person	lMid1	_Collider
Person	lMid2	_Collider
Person	lMid3	_Collider
Person	lRing1	_Collider
Person	lRing2	_Collider
Person	lRing3	_Collider
Person	lPinky1	_Collider
Person	lPinky2	_Collider
Person	lPinky3	_Collider
Person	lCarpal1	_Collider
Person	lCarpal2	_Collider
Person	lFoot	_Collider
Person	lBigToe	_Collider
Person	lToe	_Collider
Person	lSmallToe1	_Collider
Person	lSmallToe2	_Collider
Person	lSmallToe3	_Collider
Person	lSmallToe4	_Collider
Person	rThumb1	_Collider
Person	rThumb2	_Collider
Person	rThumb3	_Collider
Person	rIndex1	_Collider
Person	rIndex2	_Collider
Person	rIndex3	_Collider
Person	rMid1	_Collider
Person	rMid2	_Collider
Person	rMid3	_Collider
Person	rRing1	_Collider
Person	rRing2	_Collider
Person	rRing3	_Collider
Person	rPinky1	_Collider
Person	rPinky2	_Collider
Person	rPinky3	_Collider
Person	rCarpal1	_Collider
Person	rCarpal2	_Collider
Person	rFoot	_Collider
Person	rBigToe	_Collider
Person	rToe	_Collider
Person	rSmallToe1	_Collider
Person	rSmallToe2	_Collider
Person	rSmallToe3	_Collider
Person	rSmallToe4	_Collider

 

[TBD]

CUAs are debatable. Environments may cause unintended spanks. On the other hand, it's often best to turn collisions for CUA environments off and use slates and cubes instead.

 

[hazmhox]

CUAs are debatable. Environments may cause unintended spanks. On the other hand, it's often best to turn collisions for CUA environments off and use slates and cubes instead.

This is a great idea. I mean... this actually what we do in general using and creating simpler colliders. Having this as a recommandation (parent a simple box) is a nice idea!

 

[XcOm03au]

dang, I am trying to get the random function to work, cant seem to work it out :|
I mean the random one liner works, no problems, am hoping it would work with custom dialog I have made. As in select from a pool of customer dialog?

can anyone dumb it down plox (as in what steps to get it to basically work)

 

[hazmhox]

dang, I am trying to get the random function to work, cant seem to work it out :|
I mean the random one liner works, no problems, am hoping it would work with custom dialog I have made. As in select from a pool of customer dialog?

can anyone dumb it down plox (as in what steps to get it to basically work)

1. Create voice lines. Save them as ogg files
2. Name them GROUP_GroupName_something (ie: GROUP_Dialog01_Line01.ogg)
3. Put them in a folder, select the folder for the VA
4. Use the action to playback the VA (any action with "VA" in it, like Schedule VA or Play VA Immediate... )

 

[VamEssentials]

One issue with audio in VAM is you set your settings, then you go to the next scene and you have to set your settings again. One thought I had today is wouldn't Vammoan be better as a session plugin in VAM2? I tend to use the same voice settings or try to get them the same each time. Even if someone has there own settings, I usually change them to settings I prefer. If you could set those in session plugin settings where it works on every scene or you can load your default preferences, I think that would make it more usable across multiple scenes. Just a thought. Not sure what the downsides to something like that would be, but it seems like it would make more sense for it to be a session plugin. Especially for those using the same audio across multiple scenes from a pool. Versus having to map it each time from scene to scene.

 

[hazmhox]

You're very weird sometimes Haha
Like, if you did not have thought through what you currently have in mind.
I'm saying that for just a section of what you're suggesting here :

or you can load your default preferences

... because at that moment, you exactly know what a plugin like VAMM2 in session mode would do: actually not know what character you're refering to, so you would have to load a setting everytime.

Even further than that... you would need as much VAMM2 in the session, that you have characters in the scene, which is different from one scene to the other. So depending on the scene you load, you would need to load a specific session preset.

It makes as much sense as suggesting to have a session plugin for your characters, clothes, assets... or even Timeline!

It's a "character based" system. You would end up having to do the exact same things than loading a plugin preset or clothings preset in any scene. And the only "downside" you're talking about here, is the long process of a "three clicks action" to load a plugin preset you could do for your characters if you are using the exact same settings everytime.

And if you really think hard about this suggestion, let's say you were right... this would mean pretty much, that any plugin you enjoy for your characters, should be a session plugin.

Plugins presets exists for a reason, use them : )

 

[TBD]

This is a great idea. I mean... this actually what we do in general using and creating simpler colliders. Having this as a recommandation (parent a simple box) is a nice idea!

Please give it another round of testing.
PS: The impact velocity measurement is still really, really inconsistent. You get vastly different results depending on glute soft physics on/off and also if the object you spank with has physics disabled. You can test this with paddles or books. I'm not sure why that is or what to do about it.

 

[hazmhox]

Please give it another round of testing.
PS: The impact velocity measurement is still really, really inconsistent. You get vastly different results depending on glute soft physics on/off and also if the object you spank with has physics disabled. You can test this with paddles or books. I'm not sure why that is or what to do about it.

The whole relativeVelocity thing drove me nuts, so much that at some point, I assumed it was a Unity thing.
And if you look for velocity discussion on Unity, you get a lot of result on the inconsistency of velocity. I ended up in threads talking about the fact that any velocity computations should be done in FixedUpdate... but, it's useless since in VAMM's case since it uses the collision event.

I don't understand how this is so random. And even further, we're not even talking about bullet physics with stupid speeds... it's a frikkin' slap, and it behaves on a basic linear animation like the velocity could vary 10 fold every time ( from a 0.1 to a 1 ). It's a situation where I'm like... "yup, whatever"... and try to compensate with timeouts or alternate methods.

Thank you! I'll test it and implement it : )

 

[XcOm03au]

ah OK, thanks, I will go ahead and try this. I have been saving them as regular wav files from xvasynth

1. Create voice lines. Save them as ogg files
2. Name them GROUP_GroupName_something (ie: GROUP_Dialog01_Line01.ogg)
3. Put them in a folder, select the folder for the VA
4. Use the action to playback the VA (any action with "VA" in it, like Schedule VA or Play VA Immediate... )

 

[hazmhox]

ah OK, thanks, I will go ahead and try this. I have been saving them as regular wav files from xvasynth

Wav are ok : )
Wav or OGG are recommended. Just don't use mp3