Android/Mobile Game - LWRP or Built-in rendering?

Hello everyone,
We are going to start a new mobile game project next month, and as listed in Unity 2019.1 features, LWRP is now production ready.

Our game will feature a lot of objects per scene, detailed characters (only up to 2 maximum in the screen at the same time), fair amount of effects and no shadows are required.

Our question is, in this case, what will be better to use in your opinion, Built in or LWRP render pipeline?

 

I'm relatively new to Unity development and therefore I'm also not an expert on the LWRP, but I can provide a few observations from my relatively limited experience and also a few links to useful info.

I've been using the LWRP for a relatively complex Android game for the last month or so. After upgrading to LWRP, I have noticed a performance boost when profiling, however I also made other optimizations at the same time so I can't speak much about the performance benefits solely due to LWRP, and of course the impact would very much depend on the specifics of your game.

The way I think about it is the LWRP is a stripped down, clean, customizable renderer while the built-in renderer is more feature-rich and mature, but potentially more bloated? here is a feature comparison from the dev team.

I've seen the built-in renderer referred to as 'legacy' by some, and the fact that new features like Shader Graph are only supported on SRPs implies to me that at some point (probably a few years or more) the built-in render pipeline will be deprecated, but that is just my uneducated hunch.

One further observation, as the SRPs are relatively new, LWRP only came out of preview in 2019.1, a lot of 3rd party assets don't support it yet. I expect support will follow for many packages in the coming months.

Here is a talk from a recent conference: Introduction to the Lightweight Render Pipeline - LA Unite 2018

There is a flowchart in here to help you decide which pipeline to pick: https://docs.unity3d.com/Manual/BestPracticeLightingPipelines.html

 

I'm about to attempt my first conversion from built-in to lightweight this week, in order to push the quality of our graphics a little further on the Oculus quest. I've already done tons of optimization, with great results, but am checking out LWRP anyway in hopes to get a little further.

I went through the comparison of features from this unity blog article, and I found that I was already doing with built-in most of the things that Lightweight does by default. (Unity says "The rendering features of the LW RP are mainly a subset of the built-in renderer.", which makes sense. Many optimization are already able to be achieved without LWRP.)

The one notable feature (from my moderate understanding) that is only available in LWRP is Single-pass Forward rendering - Sort of a combination of the advantages of deferred and forward, single pass forward rendering applies all lights in a single pass like deferred, but without using the GBuffer like Deferred normally uses - so it reduces passes and avoids the memory and read/write cost of a GBuffer. However, it only applies shadows for one light. This is a much more performant rendering path than either forward or deferred, and is not available in Built-in, so this alone might make it worthwhile to switch to LWRP.


Of course, if you haven't already done loads of optimization steps, it seems LWRP will do many of them for you. Might as well skip all that trouble!

 

Hi
Please clear my ambiguity
I developing an android game that has not any realtime light and all of them was baked .
Does it have any advantage that I bother myself to use LWRP and adapt my resources to it ?