Monday 7 December 2015

Final Shop Front Blog

I created a Floor Mess Mesh that would be scattered on the ground in engine using the foliage tools. I created this by projection mapping. Later I added some nuts and bolts to the mesh.  As this asset is extremely small and will always sit quite far away from the player’s perspective, I have simply used the texture sheet for the metal panels on this asset, but I have altered the size of the texture in engine to 128 for this asset.



Average normal

I used the Average Normals script throughout this project. This is an amazing piece of script that saves allot of time with baking. Where you might bake a nice clean curve onto an edge using high to low poly baking. With Average Normals by simply placing a single smoothing group over any edge that you don’t want to have a harsh edge, and then running this script you will often get the same result. Running this scrip prior to baking will also yield better results.

Propulasion

I modelled some additional assets and continued to dress the scene.



Inside the base of the pylon an engine is being used to power a winch.



 Texturing

Due to the highly stylised nature of this brief I was going use flat colours to save time. But the second I started applying my fill layers I realised that in order to get the level of dirt, rust and general wear that I wanted, that this would not be sufficient. I immediately started using all the maps to create high quality textures. I still tried to keep a level of stylisation with the colours, colour saturation and with the forms I was adding with normal maps.

I textured every asset in Substance painter.




I colour picked all the colours from a certain set of mad max imagery.








This is my layer stack in Substance painter. Here I was using fill layers, and then painting on to the masks to remove and expose or paint and hide the below layer. This work flow is good for layered wear. Here I’m painting allot of painted metal panels. If I remove the paint I expose the rust underneath. This works as in this environment any exposed metal would be rusty. If I choose that a particular area needs some clean metal, I then can remove the rust to expose more pure metal. I also colour picked a layer to alter the hue of the metal.

Below I have created a design for the scorpions Eyes, this design includes a wrench.


All the textures that I created in substance painter will eventually be 1024s, it’s standard practice to work at double res which is what I have been doing.  In substance painter though there is actually a huge amount of information lost when you finally bake out the textures. This is why when I eventually bake out the textures from substance I’ll bake them out at the highest res possible, which is currently 4k.

I have found that when working at 2k, baking out at 4k and then rescaling to 2k that you still end up with a texture with a lower quality than what substance was initially displaying at 2k. Baking out at an even higher res could be more appropriate for 2K. For this project baking out at 4k to then be downscaled right down to 1k should be fine.



Sand

I created the below Sand Texture. I’ve tried to capture how sand can catch light and glitter.




I’ve done this by adding an additional roughness and metalness map. I’ve also used the normal map twice, it’s at its default tiling and also it tiled it at a much lower frequency to create some larger forms.


I also used vertex painting during this project to create these oil stains.



I also tried to add a third texture for tire marks. But this was unsuccessful due to the vertex density of this mesh. This is a low poly project, it’s only 15k for the whole seen.


I have already Turbo Smoothed the base mesh to create the oil stains. In order to have the control needed to create the tyre marks that I want to create I will need to add several more iterations to the Turbo Smooth, this would put this project over budget.




 Here I'm trying another vertex painting technique. 

This is the finale texture set up.

Decals

I have also created a set of decals. In order to reduce the size of this engine file, instead of creating 9 32 bit textures, I created 1 32 bit and then 8 16 bit masks. The black splodges on the below texture sheet are actually where the Scorpion decals are.




Friday Finale Feed Back

Rust texture needs to be redone as it currently looks like spray on rust.

General material definition is bad.

For sale signs are confusing as two different signs will be by the same thing.

The structure looks to flimsy. Should have been constructed much more solidly.

Sand at first glance is good, but after inspection it’s a bit weird, snow like.

Scorpion Decal feels like a toy.

All the maps are really low res.

Wheel Texture is bad.

Many of the metal panels can be viewed directly from the side, as they are planes this causes them to disappear.


Del unrelated FMP feed back

If I want to do vehicles for my FMP I should designate half the time period to the environment. This is because the vehicle modelling segment of the games industry is also occupied by professionals from the vehicle industry and their skills are extremely high. If I want to create vehicles to show interest in this area that is fine but don’t rely on it.

The amount of work that we should be producing for FMP is extremely objective. One train of thought suggests that we should be creating large amounts of something e.g. only creating 3 characters for FMP is not enough and does not show the ability to work fast so 5 would be more appropriate. Conversely we have been told that if you were planning to do 3 characters, then you should probably cut this down to 2. This is the quantity V quality argument.

It was also suggested that if you are creating 2 or more of something, then you should create two vastly different things.  

My current FMP Idea

Currently I’m playing with creating 3 vehicles. One highly polished photorealistic model, and then two different lower poly stylised vehicles. I would then create an environment to house these assets. I also think it would help if these vehicles were drivable, to show understanding of how to actually create a riggable game ready asset. I would personally rather not spend 10 of the 20 weeks on the environment, 10 weeks on 3 vehicles would probably have to be changed to 2 vehicles.

Back to Friday work

I tinkered with all the textures to try and create more texture variation, and make them more realistic. The issues with my textures is that the transitions between different texture types are extremely soft. This is due to the extremely low Flow rate that I was using in substance throughout the texturing process of all my assets. This has resulted in extremely flat and seemingly unvaried textures, most areas are a muddy combination of many material types.

Increasing the contrast of various layers has already added allot more variation to some of the assets as its added areas of clean metal that weren’t visible before.

The main problem I have had during this project is that I have made everything far too low poly. At the start of this project I thought that creating a vehicle and an environment would be a real push, as a result I completely over estimated how low I would have to go.

So right now at this end point I have modelled a vehicle and some additional assets, and I am 4k off the tri budget. Because of this I have gone through and turbo smoothed allot of the rounder assets, to help hide their low geometry. This though has now made these meshes inconsistent with vertex density of other meshes, but I have the budget and may as well make the level look as good as I can in the short amount of time I have left.





This image shows some of the changes that I have made









































No comments:

Post a Comment