#9 - Prop Blockouts
This week, I continued working on placeholder meshes for some of the most important assets in the environment. Getting these done and in-engine should begin to give an idea of how the final scene will be laid out, and highlight any flaws in the composition.
It is important to stress that all of these meshes are temporary, and most will go through a pass of sculpting and retopology. Right now, most of them only really exist as base-meshes for sculpting, and as such do not have optimised, game-ready topology.
The Bridge
The bridge was my main priority, as it is front and centre in the scene, leading the viewer's eye across the river to the front of the house.
To begin, I exported the mesh created using UE5's built-in modelling tools (seen in the paint-over shots), this gave me the exact scale and rough shape I needed.
The result was this:
I like the uniqueness and silhouette of all the different, separate bricks, but I know that UV'ing and texturing all these unique bricks will take far too much time, not to mention texture space.
Before I come to sculpt it, I think I'll probably break it down to maybe 3 or 4 bricks, sculpt and texture those, then duplicate them and re-create a similar effect to what I have right now.
For the main body of the bridge, I'll probably create a tiling texture to apply, then go back in and cut and extrude some of the detail out, rather than trying to create a unique sculpt for such a large object.
I'm also not so sure about the rounded rocks clipping into the sides of the bridge. It was an interesting stylised detail I'd seen in some of my reference images, but I'm not sure how well I've re-created it.
It's quite hard to tell how it's going to look once I have the surface detail sculpted in behind it and have a texture applied. It may blend in better at that stage, but it's definitely one that will have to be revisited to some degree.
As the bridge is such a focal point, I am conflicted on keeping the amount of modelled detail in this shot. I think the silhouettes of the individual cobbles break up the surface nicely, so a handful of optimised cobble meshes with nice baked detail from a trim sheet feels like a good approach that shouldn't be too performance-heavy.
Walls and Fences
Walls and fences seemed an obvious, simple, and useful addition to my kit.
Many of my references include old dry stone walls, as well as wooden fences, often serving as leading lines to draw the eye to the important parts of the composition.
Wooden fences seemed the easiest to put together:
My plan with these is to sculpt one post, and one board, and reuse these over and over. With these two pieces, it should be possible to create an in-engine tool that creates and places dynamic variant fences along a spline.
A similar approach should be possible with the stone walls, although I have a lot more sculpting and retopology to do with these:
The walls are made up of three components: Cap, Top, and Base.
The second image shows the effect of duplicating the top and base pieces along a line, with a cap on either end.
I think the tiling works pretty well here as is, but it could be improved by removing some of the more unique details.
Obviously as it is, there is way too much geometry, including a lot of faces between bricks that can't even be seen.
My plan here is to take the mesh into ZBrush, sculpt the bricks into more appealing shapes, then attempt to merge the bricks into a much more optimised single mesh, using a normal map to bring back much of the surface detail.
Again, like the fences, I should be able to create a tool to lay these walls out along a spline in UE5.
Windmill
As a mesh that is only really seen at a distance, I knew the windmill didn't have to be elaborately detailed, but it was imperative that it had a unique and interesting shape that would stand out well against the skyline.
This windmill is based heavily off of one of my references, seen here: https://www.artstation.com/artwork/VgNgqg
The final version will bring in more of the details from the reference like ropes dangling from the wooden posts that jut from it, more detailed window frames, detailed wood elements, and interesting stonework around the base of the tower.
I also plan to add a really simple spinning animation to the blades for the final shot. This should add some additional motion to the scene, and also give the viewer a feel for the weather conditions in the scene.
Tree
I made a very (very) quick tree in SpeedTree:
I hate it, it's terrible, but I didn't want to dedicate any time to it at this stage for a few reasons:- I want to sculpt the trees for the final version, rather than just generate them.
- I want to experiment with some different ways of doing the foliage, but I'm not quite at that stage yet.
- I also didn't want to spend time making any textures at this stage, as they'd most likely need replacing anyway.
- I just needed a simple placeholder to scatter around in-engine.
I hate it, it's terrible, but I didn't want to dedicate any time to it at this stage for a few reasons:
- I want to sculpt the trees for the final version, rather than just generate them.
- I want to experiment with some different ways of doing the foliage, but I'm not quite at that stage yet.
- I also didn't want to spend time making any textures at this stage, as they'd most likely need replacing anyway.
- I just needed a simple placeholder to scatter around in-engine.
Rocks
Moving swiftly on to the comfort of these lumpy grey blobs:
These rocks are clearly very generic, and also very poorly optimised. Again, they are just vaguely rock-shaped base-meshes for me to sculpt up nicely later on.
All of these rocks in their current state are very generic and uninteresting, but (hopefully) won't be in the final scene.
Having some basic rocks to scatter around the scene should help fill up some of the otherwise blank spaces well. Rocks should work well to hide some of the low-resolution landscape geometry, as well as hiding the seams where meshes intersect with the landscape.
The low resolution landscape is most obvious along the riverbank, and the early screenshots make it apparent that this area needs special attention.
To this end, I have created some similarly generic meshes that I intend to place and stretch along the riverbank:
These blobs are hideous as they are, but I'm hoping that in the final scene, with nice sculpting, texturing, shaders, and foliage, they'll blend in really nicely with both the landscape, and each other.
One of the techniques I hope to employ to achieve this is Tri-Planar mapping. Using this technique, a tiling texture can be projected and made to tile across all of these meshes seamlessly.
This, combined with another technique called World-Aligned Blend will allow me to dynamically project a tiling rock material to the front of the mesh, and a grass material to the top.
You can read more about these techniques here:
Comments
Post a Comment