Rendering of first person weapon model

Hey,

can anybody explain how the rendering of the weapon model works?

I expected to find a second camera that overlays the rest of the scene to avoid weapons to stick through other geometry (walls, characters, etc.) when player moves too near to any of those.

What I did find is some flag called "_FpsModeFov" inside Lit.shader, for example.
However, I cannot find any usage of it. So what's going on here?

 

Because it not use second camera It's all - great opportunities which give us SRP

 

Tiny Gun?

 

Yes, I know that it's not using a second camera. But how is it done with SRP then?
My search ended at the mentioned variable / flag. I don't know where to look further to understand it.

 

Do you mind to point out where is this function running at? since HDRP code are quite big, i just cannot find the correct files

 

Thanks! This helps me a lot.

 

Hi Peter - is it possible to not shrink the weapon? because for example sometimes we need to have an accurate interaction (sword hitting shield) in first person and we want the shield to render on top of the enemy sword and of course not clip with geometry. Too difficult?

Small doesn't really work in cases like that sadly, since these things can penetrate up to the camera near clip in theory and won't be kept at bay at capsule bounds...

 

@petera_unity Oh nvm i found them

 

Curious why you guys don't just put it at the end of the render queue, that's what I've always done to avoid this issue. That way you can keep everything scaled correctly, not use double cameras (wasteful), and it always draws over everything.

 

Like he said, the shader is used to 'reproject' the mesh using a custom projection matrix with a different field of view from the camera and not to avoid sticking through geometry.

 

Yah I got that, I just didn't think the two were mutually exclusive. Custom FOV, then have it render AFTER everything else.

 

That's normally what people do isn't? even though it still using custom projection matrix, the Shooter Hand and the world interaction hand have different FOV.

 

I cannot unsee it now I've seen it - thanks a lot

Do you have a recommendation for the clipping issue for melee combat FPS games? not worried about the FOV because we don't have fov adjustment for melee FPS and shrinking them just breaks the whole thing when they interact. It would be better to just say "OK you're on top!"

 

Hackish way? separate the collider and the renderer. Just make sure the animation are synched

 

Well the hack doesn't actually work. The physics aren't the problem. the sword from the enemy penetrates (clips) through the shield and so on. We simply want to make sure the shield renders on top of most things but not all (UI and particles for example).

Yep that's with a tiny 0.2 scale shield right in my eye.

 

Ah good point, especially the shield clipping and the fast moving animation part.

 

I'll give it a go! but I'm sorry for dragging the thread a bit tangent

 

How about doing number 2 but with custom projection matrix and not tiny gun? since it can be done in Legacy pipeline

 

if you take a look in this video from 3:17 to 3:50, it really seems like the first person gun+hands mesh is full-sized. Really not sure how they handle this. In some cases the gun is really really huge and would definitely clip through walls all the time if it was rendered without any special tricks

EDIT: actually.... I noticed at 0:13 that the arms+gun mesh is rendered on top of the BFG shot. Seems like they might just be rendering it on top of everything else

 

My instinct would be a simple stencil pass or something.... Seb mentioned somewhere you could probably just easily reserve a range of depth. I don't know how I'd do either to be honest, in HDRP

In my case I'm just going to have to let the clipping happen, or perhaps render something to texture then composite it.

 

In URP, Stencil pass seems the way to go. It is efficient, theres a tutorial to set it up in ~15min and it just works.
Now after switching from URP to HDRP, the HDRP-Asset / scriptable render pipeline asset seems to be missing so theres no place to set this up?

I don't really want tiny guns because

• it is harder to work with them (zooming, unzooming the camera whenever you work with them)
• incorrect shadows
• way harder to set up projectiles, think a rocket launcher (or bazooka) where you see the rocket at all times
• harder setup of physical interactions like knife kills and other melee weapons
• having to adjust the near clip plane of the camera, limiting the far clip plane or increasing possible Z-fighting issues
• and probably a lot of other issues of which I will find out later in development .. scary

Using camera stacking with two cameras is bad for performance and you can't use fog in your post processing / sky and fog volume - or the fog of the weapon camera will be rendered on top of the game camera.

So whats the way to go now? Is there any option similar to the URP version with the stencil pass? How to solve this very common problem in HDRP right now?

PS: I tried a bunch of things already, looked at different forum posts and none seems to have a good solution (like the one in URP), thats why I am asking now.
Theres a youtube channel, using a Custom Pass Volume (with DrawRenderersCustomPass), but setting this up for every single material on my weapon and arm models is just insane ..

 

rly? did they fixed problem with shadows (when turn on cascades)?

 

I guess so - I saw some issues with shadows, I could see them throught my opaque weapon but this went away (or I stopped noticing it? idk, but I think it got fixed), now I am working with HDRP, so I can't check it anymore.

Edit: Now I am using camera stacking on HDRP, which decreases performance and you have to turn off the fog on your weapon camera (advanced settings .. ) but everything works now as expected, except for Gizmos/Handles - they are drawn with the wrong camera in gameView.

Edit2: Camera Stacking was a performance killer, so I switched to using the shader instead, scaling the weapon and well, actually, weapon clipping is one of the least important things as indie dev. Considering proper AI, fluid movement, non-jittery elevators, good pacing, consitent and logical world building, good performance, nice graphics, understandable and easy-to-use UI - weapon clipping is really just a small detail and I decided to stop caring about it, so I can focus on making a good game instead.

 

Where does one adjust the render fov for the weapon?

 

There is a parameter on its shader for that....

 

What I meant was, where do I change the value for the parameter, so that the the value that I want, is the one that gets used for that object, when the game creates a new instance of said object? I can play around with the parameter's default value inside the shader all I want, once the material has been created, changing the default value won't alter the value in the material. On top of that, if there's a prefab or scriptable object holder a value that gets applied to the material (in code), then the value that's set on the material won't matter either.

 

Yes? Thats why it is called a default value. If you want the same FOV for all objects using the shader, just don't expose the value --> make it a Vector1 Node, instead of a property but then you can't change it at runtime.
If you want to change it at runtime and you want to change it for all GameObjects using the same material you have to use MeshRenderer.sharedMaterial API - and then set the value via sharedMaterial.SetFloat() API .

This will set it for one material, if you have more than one material and want to set the FOV for all of them, you have to create a List of all these materials and then iterate throught them, to set the FOV for every single material.
Pseudocode:

foreach(Material mat in myFovMats) mat.SetFloat("_FpsModeFov", 40);

You just have to be aware that if you change a material Asset and exit play-mode, all changes will stay because you changed an Asset.

 

We just wrote our blog for solving a similar problem described in this post. This is our blog: https://www.intetic.com/blogs/unity-hdrp-render-object-on-top-of-everything/

Basically, we are using two Custom Passes that we are overriding Material on a specific layer, so it's not tied to a Material, but to the object that belongs to a specific layer.

Hope we helped devs that have a similar situation.

 

Wow, I haven't looked at this for a while but I see my meaning never got through. Somewhere, there's an prefab or a scriptable object, with a setting that holds a value that I can change without having to make any alterations to the shader. By changing the value in that location, I can change the FOV that get's used in the weapon within the game. Yet, there are multiple levels of that value with different priorities, so where is the one that is intended to be set when someone wants to set the FOV? Why write code that overwrites a feature that does the very thing I want to do, when I can just use said feature. I'm not asking how to brute force an override. I'm asking if anyone knows the location in which they put the setting that does end up getting used for the weapon's FOV.

Well, at least that's what I was looking for back when I was going through the FPS Sample. At this point, that information is no longer useful to me.

 

The feature is on the weapons shader. Find a weapon in the game and look at its material. On the shader there is something called FOV overrride(or something like that).