FMP - Week 18 The ends drawing near

   Last week I mentioned a sudden drop in the frame rate of the level. An unexplainable one at that. When I came to open the engine file this week, and expecting it to drop again. It didn't, and I haven't had any problems at all in that respect. So I just ended up leaving it, as there's nothing I can do until it happens again.

   Over the weekend I spent more time working on the LOD's for my assets. I soon realised that I should have put more geometry into some of them, as I don't feel I can scale down a lot of them to the point of making an efficient LOD without ruining the silhouette. This isn't really a problem, more of a future reference point for further projects. The LOD's aren't a hard process, they're just time consuming, but I did set aside a week for optimisation, and between this and packaging the textures, I think I'll be fine on the time I have left.

As me and Sharn are switching the engine file between us, it gets to a point where I've added models and textures, among other bits like switching tiny things around, no big deal really. But after a while these things add up. So I spent a little time adding the tweaks I'd made, back in and altering other little bits too. Just general house keeping bits too, such as naming conventions and foldering assets. I did a test package of the engine too, to see how it runs as a finished application. I'm happy to report, it runs perfectly well. Not only does it run well, we've gained about 10 fps!! Happy days!

I'd been waiting for a clear day to get to work on the landscape I had left. The one which spans from the main door of the wall, and down the side leading to the opening which the ship sits in. In my head I wanted to have a base of dirt with grass, pebbles and puddles mixed in. Towards the door I intended to have either the same again, if the ship zone worked well with this combination. Or something along the lines of tarmac, or a series of stone tiles, but that might get too repetitive with the market zone having a similar set up.

With the landscape tool unreal recommends only having up to 4 different textures to blend together. But from what I found online, you can have double that, and maybe more. To do this its as simple as clicking on the texture sheet and changing the "sampler source" to "Shared: Wrap". Its not recommended to go crazy with the landscape textures. But the 7 I'm using should be fine.

Landscape material.

As a test I used the grass I made in the style matrix project and laid it out over the landscape for more interest. I intend to make some new and improved grass before the end of the project. But this will do for now. The test grass is too dark and a bit too clumpy. So when I make the new grass, it'll be lighter and more spread out. For now the grass isn't casting shadows. I'm sure its just a tick box somewhere. But I cant seem to find it at the moment, but I will return to this in the future. As for now I've got bigger fish to fry.

I went back and added more of the macro variation to the majority of my tilables too. Its amazing how much you can improve your scene by doing this quick and easy process. The key to this is to use tilable textures for the macro variation. I ended up using a few non tilables because of the detailing on them. but I'll probably go back and change them when I have time. Just because of a few seems I can see on the variation pass.

Before macro.

After macro.

I felt the environment needed a bit more movement in it. So for the "BAR" sign outside the bar. I added a flicker too one of the letters. This added that extra bit of depth and interest to the market area. Along the lines of lights, I added more to the scene in general, so its a bit more true to life now, rather than extremely harsh shadows.

With the end drawing near and degree show shortly following. We've been having a few chats with Chris and Craig about what'll be happening. So far we know theirs going to be food, booze and industry people coming in to see our magnificent work. We've been discussing the seating arrangements and what we need to have done. We intend to have a playable level each (Myself and Sharn) Along with a fly through each. Additionally to this I'll probably be making some changes to the engine file to add more interest for the player.

Texturing also continued this week. But im nearly finished!! I textured the mechanics. My favourite building currently, as a i lathered it in graffiti. With a little help from Jake's book on the topic. I took inspiration from a few of the artists. A french man named Boxcar Jeim AKA Jeim, and an Argentinian called Julian Pablo Manzelli AKA Chu. Their Graffiti consists of wacky cartoon characters. So I thought it'd be a nice fitting for a humorous background art to sit in this bleak scruffy looking world.

Boxcar Jeim.

Chu.

Mechanics with my graffiti.

FMP Week 17 - Plugins, engine problems and LOD's

   These past few weeks have been pretty consistent texturing progress. one of the things I moved onto this week and originally had no idea how to make, was the chain link fence. I knew It had to be a tillable, but I wasn't sure how to go about making the texture.
   Little did I know that there's a plugin for 3ds max that is purely for making wire fences such as this. It gives you a variety of parameters such as the size of the wire loops, how many twists occur etc. I keep getting told that by doing something in a more efficient way it isn't necessarily cheating. But I cant help but feel like I'm cutting a corner by using this. The alternative would have taken me quite a few hours and I would have had to do quite a lot of tweaking before reaching something I was happy with. Where as this whole process, of making the mesh and projection mapping, to lining up the texture sheet to be a tillable, followed by making the texture itself in engine. Took me all of about half an hour I think. Much more efficient than about 4+ hours of work for the same result.

Not bad for half an hour.

I worked with decals in engine this week too. Which aren't exactly practical, efficient or working properly. Initially I had the problem of them working fine-ish, in the light. But in the shadows they virtually disappear.

Standard decal.

Decal with emissive. 

This is obviously a huge improvement, but it only works for coloured decals. It wont work in dark areas where the light flickers, because it'll be illuminated all the time. And because of this, grime is out of the question due to these limitations. This is one of the things I'm constrained too by the engine version, but even if I used an up to date version, I believe they're still not 100% fixed.

LOD's are a useful way to reduce the draw on the engine if you're having trouble with the frame rate. So I intended to make multiple for each asset. The process of making them wasn't long at all really, and it made me realise I need to put more geometry into my assets in the future. Some of the assets I made didn't need an LOD, because they were as low as I could go without disturbing the smoothing groupings.  So a few of my assets don't have them, where as some have multiple. I had a feeling I could package the LOD's in with the original mesh. The same as you'd do with the collision. But when I tried to do that, they didn't combine. The alternative was to import them seperatly. Which doesn't take long at all really. Its just a bit annoying when the feature should work.

A selection of LOD's next to their respective meshes.

Last off I had a few random problems with a drop in FPS too. Completely unexplained. I didn't know if we'd reached some kind of limit on the engine, or if that was even a thing. So as a test I decided to remove Sharn's side of the wall in a copy of the engine. To see if anything would happen. Unfortunately nothing happened and it still dropped considerably. More on that next week.

FMP Week 16 - More texuring

  As we keep a constant flow of critique between the team, Jake mentioned the level looked a little dull and brown this week, so a bit of brightening up was in order. The idea was to have the majority of the level looking fairly muted with areas of pop. I think the reason my environment is looking this way currently is because I haven't fully imported a lot of the brighter assets textures yet. But despite this we also felt the walls could do with a few more colours added in. some muted blues/ turquoises were what we decided on. It really pushed the colours and added more interest to the level overall.

TOO BLAND A BROWN!

A few colour changes to the plaster.

More texturing cloth was done this week too, like I said in my previous post I've got a good system down I think for a comfortable cloth material. So this didn't take me too long and looks like a good cloth material in my opinion. I also took to texturing the series of signs I have scattered around the market area of the scene. These vary in texture shape size and eye draw. The more basic stalls have hand painted/ botched together signs which have little appeal aside from the obvious name/product they're advertising. Where as others have emmisive borders or text which obviously draw your eye more. These are owned by the higher quality shops/businesses such as the bar and gun shop. Which clearly have more money to throw around.

More cloth for the scene.

Variety of signs.

The main thing this week I gained was the knowledge of colour manipulation in UE4 via a blueprint system. Last week I was shown how to change the colour via an instanced material. I knew Mike Pickton had a better and more practical solution to this problem, so I asked him. He explained that you can do it specific to each mesh actor in the scene using a fairly simple blueprint in which you set a parameter. Which appears in the panel at the side settings of the asset to change the colour. The main thing that I had to do, was set a mask to the sections of the mesh which I didn't want the colour change to occur. Along with setting the section I did want to change, to a central grey colour, so the colour occurred naturally and didn't clash or mix with the initial colour.

Any colour you want!!! 

Mike said this parameter is a good way to add variation but don't use it for as asset you use a huge amount or on a large variety of assets either, because having a large variety of meshes with this on them can cause a drain on the engine. So use it fairly sparingly.

I had a weird issue this week too, with the differentiation between vertex painting a texture on a mesh and using the landscape editor. Even though I'm using the same material on both, I had a clear difference between the two. It turns out I had a value of "0" set on the specular channel on only one of the materials.

Before.

After. (And with a few tweaks to the main floor contrast)

But, oh and its a big one... that didn't fix the problem entirely. Unfortunatly. I'm still fairly confused on the mater. I feel like it has to be a difference I have no control over between the landscape editor and the process in which vertex painting is applied. Either way I have no idea of the solution. Its not to much of a problem, but its a thing that could push the environment that bit more.

Along with helping me with a few problems in engine Mike did a great talk this week about his time at ubisoft. The talk included general tips and tricks about environmental art, lighting and other placement and draw that the engine has to deal with when rendering the scene. It was a great insight into what I should be thinking about when placing assets in a large scene/environment. one thing I found particularly interesting, was when he said about how the engine draws assets in sections of the map. And if you only have one instance of an asset in that area it has as much draw as another section with that same asset instanced multiple times.