Question Confused about RenderTextures and Camera.Render()

I am working on a tool that creates icons of GameObjects.

A basic rundown of what it does:
It uses the scene view to align the snapshot.
A temporary camera is created and its position/rotation is set to the same as the scene view camera
I do this to change a few settings on the camera (I dont want to mess with the scene view camera)

After calling

Render()

on the camera and setting

RenderTexture.active

, I get exactly what I need when I call

Texture2D snap = new Texture2D( m_SelectedIconSize, m_SelectedIconSize, TextureFormat.RGBA64, false );

snap.ReadPixels( _snapArea, 0, 0, false );

Is setting the active render texture the only way to get a camera render ?

It feels weird to save the current render texture, set the active one manually, read the pixels and set it back to the saved one.

I tried doing the same with

Graphics.CopyTexture( _cam.targetTexture, 0, 0, (int)_snapArea.x, (int)_snapArea.y, (int)_snapArea.width, (int)_snapArea.height, snap, 0, 0, 0, 0 );

but the result is always grey.
I guess thats because it doesnt actually render the camera to its render texture.

Frankly, I dont get it

 

No... as per the docs:

https://docs.unity3d.com/ScriptReference/RenderTexture-active.html

I quote:

"if all you need is to make a Camera render into a texture then use Camera.targetTexture instead."

 

Ok I found this.

I use

Texture2D.ReadPixels()

and

Texture2D.EncodeToPNG()

which both run on the CPU.

Graphics.CopyTexture()

runs on the GPU.

But just looking at the

AsyncGPUReadback.RequestIntoNativeArray

page makes me wanna leave it as it is, because I have a feeling the code looks even more hacky afterwards.

Maybe I will try to make my own

ReadPixels()

.