Difference between rendering a Container and rendering a RenderTexture

[caymanbruce]

I am confused with the two main methods to animate objects in pixi.js.

The first one is to build a sprite and put it in a PIXI.Container. I can then add this container as a child of the PIXI stage. Every tick of the animation I just need to update the sprite's position and PIXI will handle the rest for me.

I can also build a sprite or a displayObject and create a RenderTexture which is about the same size of what I want to render. Then I use 

renderer.render(DisplayObject, RenderTexture); 

to render it on the screen. This works fine too. I still need to update the position on every tick so the displayObject will animate.

But what is the difference between using a Container and a RenderTexture? There must be a reason we have two options to render something. What use cases are these two methods best fitted in?

[xerver]

RenderTexture renders everything in a container's children tree to a single texture. If you have a lot of sprites, which are all static, you may not want to have to iterate a tree with all those sprites, update all their transform matrices and render them all. Instead you can render the entire tree to a texture then render that once with a single sprite.

If you just have one sprite running around, then rendering to a renderTexture first, then using that in another sprite and rendering that is going to be slower.

[caymanbruce]

3 hours ago, xerver said:

RenderTexture renders everything in a container's children tree to a single texture. If you have a lot of sprites, which are all static, you may not want to have to iterate a tree with all those sprites, update all their transform matrices and render them all. Instead you can render the entire tree to a texture then render that once with a single sprite.

If you just have one sprite running around, then rendering to a renderTexture first, then using that in another sprite and rendering that is going to be slower.

Thanks that sums it up. 

How about mixed situation? Such as a MiniMap.  I have read some discussion about writing MiniMap in Phaser using RenderTexture. I used to do that with simple sprite and container. A MiniMap has a lot of static objects, in the mean time it also shows local player movements in a constrained space. Let's make it simple so that I only use a dot to represent the local player. So there will be one moving sprite, but maybe hundreds of static sprites on the map. In this case, performance-wise, is it better to use a container or a RenderTexture?

[xerver]

8 hours ago, caymanbruce said:

Thanks that sums it up. 

How about mixed situation? Such as a MiniMap.  I have read some discussion about writing MiniMap in Phaser using RenderTexture. I used to do that with simple sprite and container. A MiniMap has a lot of static objects, in the mean time it also shows local player movements in a constrained space. Let's make it simple so that I only use a dot to represent the local player. So there will be one moving sprite, but maybe hundreds of static sprites on the map. In this case, performance-wise, is it better to use a container or a RenderTexture?

It is (always) better to measure and see