Week 17: From Engine Polish and FPS Problems

Slow Progress - Container Walls

This week just felt like I managed to accomplish nothing at all, however looking back at it, I must admit that I did actually get quite a few things done. It's just not as obvious as it usually is. I left of last week with a change in idea and how I assembled the level and lit it up with torches. The main problem I encountered was an FPS issue at this point, which in return would take me a long time to fix. While it was mainly the PC that was the problem, I wanted the level to run smoothly even on weaker machines, so I decided to take some time and optimise my torches so that they would create less FPS-lag. I started by looking into the particles itself and trying to mess around with the settings so that it would spawn less particles (even though I only had quite a few to begin with), but at the same time making sure it still looks the same. Not much changed after quite a bit of messing around. I only got a very small boost in the FPS.

I then decided to step back and work on materials. A member of my group was supplying me with textures for the walls and it was my job to put them into engine and create the materials. I wanted to make sure, that the materials were customisable, to enable us to mess around with them and do quick changes as necessary. One example for this was that we wanted for the rust on the wall to be optional, so I used Linear Interpolates to blend textures together. In the end we decided to not use this feature though, as a level with rust all around seemed to give off a much better feel. We went through quite a lot of work in getting the rust to look believable, which was a lot harder than we first imagined. One of our first tries resulted in this:

Rusty Container Wall - Take 1.

As you can see the rust just looked a little out of place, which we later found out was caused by the fact that the rest of the wall was just too clean. Our original plan was to just have some walls with rust and some without, but in the end it just looked better with rust all around. We also decided that while it was nice to be able to customise the colour within my wall material, it created quite a few problems with certain colours just not looking great at all, so in the end we just stuck to making the walls white, with a slight greyish tone. We also added a lot of dirt to them, to make them look more worn off. In the end we came to this:

Final Rusty Container Wall.

Wall Texture.

... which I was quite happy with. We encountered a slight problem with seams appearing on certain parts of the texture, but we solved this by creating a patch of dirt and using it as a decal to cover up the seams, which worked like a charm. In the picture above you can also see the ground texture, which was also created by my group member. For the ceiling we decided to use the same material as the walls, as container ceilings usually have the same metal walls on the ceiling, too.

Solving the FPS Issue

Or at least finding a work around... I took quite some time this week on trying to get the level to run smoothly. I tried even creating LODs for the particles on my torches, which on the faster PCs boosted the FPS from 40 to 60 on a scene, however on other PCs the FPS were still just under 10. So in the end after trying nearly everything I decided to get rid of the lights that were being emitted by the particles and instead put a point light into the torch blueprint. I then used the blueprints timeline function to give the light a dynamic intensity and radius to create the torch flickering effect. In the end it didn't look as pretty as it did before, but at least now the FPS were at a more acceptable 24 instead of 9.

Torch Blueprint (Dynamic Light)

Here is also a picture of the final torch:

Torch.

I decided to end the tweaking at this stage though and will now focus on putting in all the assets into the scene. My group members have been hard at work over the past week and all the assets are finally done (I even managed to do a few myself!), however they still need to be unwrapped and textured, which is what we as a group will focus on for the upcoming week. Hopefully, we will be able to get everything done though, since most of the scripting and tweaking within the engine is pretty much done, thanks to my efforts from this week. This should save us some time and once the textures are done we just need to put them into the engine and everything should work just fine. Since there is a bit of confusion as to when our hand-in actually is, we are aiming to complete this project by Friday (instead of Monday), so hopefully in the next post I will be able to show off the rest of the level itself.

Stay tuned~