Why does "Output Particle xx" context work like "Update Particle" ?

I mean updating my particles in" Update Particle" context. It makes sense.
But I can also use update particles attribute nodes in "Output Particle xx" context ?
And if you create a default VFX graph. You can see that Set Size over Life and Set Color over Life in the Output context.
I mean why not just let the output context only handle the output settings.
It's confusing me and my co-workers.

 

It might be confusing at the beginning, but it has purpose, so let me explain:
Update context and Output context does not work in the same way - if you update particle in update context you work on real particle data and it will be persisted, while update in output context is lost, it exists only for duration of rendering (current frame).
Most of the time it doesn't matter, however if you make something more complex it's very handy, let me give you few examples:

• Single system can have multiple outputs, for example 3. In this case you can update particle once in update and render it 3 times from the same data, otherwise you would need to calculate it 3 times what is worse for performance and not desired - you don't want to simulate physics 3 times.
  
• Performance aside, you can set scale in update and then set (or multiply / add) it again in output based on that value, so for example particle rendered in second context is twice as big. You can do anything you want, like displace particle or change color.
• Even if there is only one output you can do things you couldn't do without that system, for example kill particle. You can set particle

  alive = false

  in output, so it won't be rendered, but only for the current frame, because you work on copy of real particle data. If you did it in the update, then it would die forever.

As you can see it does not matter for most of systems, but it's important to at least understand the difference between output and update context, so that you don't look for strange bugs in future.

 

You can have multiple update and output contexts, the choice is yours where you put what. If you want you could just remove the update and go straight from init to output. For some complex graphs it's more performant and easier to manage if you do general stuff in a common update, then further modify it in the output context.

 

Thank all you guys ! I get it now.

 

Just to clarify, the top right output node is the invalid one? Otherwise it'd be fine? Apologies for the mistake, just thew a "complicated" setup together to demonstrate and I had just downgraded to 2020 for Entities 0.50