Well that's a lot of hats !
So why not finish the project but making it super generic? Strip all star wars terms out. Then when the game is finished, allow for modders to make a conversion mod that will reinsert the star wars material? That way he can finish it and we all can get what we want and no one has to give up on their dreams.
When are you guys going to make God of War 5?
Abel Dopazo talked about the process of creating his marvelous environment scene based on Cambodian temple Angkor Wat in UE4.
This scene is inspired by the real-world location of the temple Angkor Wat, in Cambodia. The main interesting feature being ancient carvings on the stone pieces, overgrown by huge roots since it’s in the middle of a jungle. I decided that it was recognizable enough for most people to see it as grounded in reality, but still having an almost fantasy-like vibe.
It was thought out as a playground for me to try different techniques that would improve workflow when tackling scenes that had a lot of elements while retaining a sensible performance for real-time rendering.
I thought of using a fantasy concept that takes that idea to a more stylised extreme, and cherry-pick the elements that would allow me to achieve a very specific aesthetic that has the best of both worlds: believability along with flashy hints of fantasy here and there. This is the main concept I am using, by concept artist Veikka Somerma:
Things that I am taking from this is primarily the use of pinks to contrast the general greenish tint that the ambient light from the jungle gives to the grey stone pieces, moss, etc. It will be a nice color triad if we include the ground, which will be an almost reddish colored soil:
I started off by gathering as much reference as I could, trying to recognize patterns on the temple pieces that could make modularity easier, even if there are going to be destroyed versions of these assets. I decided how I would break the different pieces for an efficient workflow that allowed me to have enough variety for the assets to be believable but at the same time, be able to quickly put together parts of the temple and modify the layout in case I wanted changes in design and/or composition.
So I decided to have modular pieces, some of which would have destroyed versions of them, then another version that even though its based on a modular part, has unique features so that I could have interaction between objects (for instance, a roof piece broken where a heavy root is on top of it). Here is an example of cross-referencing and pre-planning for this stage:
So, first things first. Scale. I quickly put together placeholders of the pieces and the main roots to try and get proper scale before jumping into anything else. When I had some base pieces to work with, I played with them to create a composition that I believe is interesting, having the eye lead throughout the image, flowing along direction lines created by the roots and the path I created going up in the distance, using the big arch to frame that part so it acts as a secondary composition on its own.
Notice how the focal points are balanced left to right. I used that framing technique to create a foreground, middleground and background separation for the eye to read easily.
In later iterations, I lowered the camera and moved it a bit to enhance this sense of depth, have a foreground that is closer to the camera, and pointing it up ever so slightly to lead you to feel that you are looking up to something tall and imposing.
Once the scale of the assets was established I moved on to create the high poly versions of them that I would later decimate, and clean up and UV in Modo.
I started off with the big root set as it is a focal point. I chose to have a baked 0 to 1 normal and AO maps, then create a tiled Substance Designer material to overlay on top of these.
I wasn’t too worried about copying the real roots, but having a close look would make people instantly ground it in reality.
Here are more examples of the modular pieces that I sculpted:
I would later separate this piece into 3 different trims that share the material for optimization and speed purposes.
The high poly sculpts seem to miss micro details and they look stylised: this is due to the fact that I am not worried about perfect photorealism. After some tests, that is the amount of detail that after a bake I considered perfect since I will use shared detail normal materials on objects made out of the same material, to help to tie them up in a believable way.
I started creating secondary shapes in the form of smaller vines, broken pieces and rubble, and further developing the look of the scene. Also starting experimenting with some of the intricate carvings, which I might use to sculpt in ZBrush, and grab the heightmap and mask to stamp it in Painter.
Several of these pieces were modeled and duplicated around, so I could then bake them as alpha textures with normal information I could stamp assets with.
Moving to UE
At this point, I had enough elements to bring the scene to Unreal and start setting things up, and whenever I had a newer version of any asset I could simply update it.
After some progress, I did a paint over in Photoshop to try and see what direction I wanted the general mood to go:
I started modeling and sculpting my own set of leaves that I would use with masked images found online to texture in Designer, make cut-outs and arrange them in several ways to create clusters and variants that I could paint in Unreal but also hand place where I need more control.
Additionally, I used cut-outs of full trees as cards in the background. Some of them are combined with leaf planes (or clusters) to create a sense of real volume.
Thanks to learning Substance Designer I could sketch a fluffy moss material that looked the way I wanted from the distance a player would normally look at it. Also, it is using a fuzzy shader and it is world aligned so that I can apply an instance of the material to any mesh and intersect it with roof tiles to make a seamless transition that looks like accumulation in those areas. This is the base from which I will create several clumps, masked and with a slight offset from the main mesh to achieve the fluff.
First vegetation pass and most of the assets placed. Testing some normal decals for the tire tracks. Will make particle emitter for the dust and another for the tire spitting pebbles.
Working on the big arch at the moment after exploring different ideas. Currently sculpting the damage from several roots cracking the stone slates on the arch. After that I will move on to the stairs and rubble, then Substances for the stone, vertex painted lichen, and variations of the reddish soil.
Sculpting the big arch destroyed stone pieces, making sure that the roots interact with them making them crack.
I then worked on setting up materials, experimenting with a couple of smart materials in Painter to speed up the texturing pass, as well as to quickly create variations in terms of moss masks, dirt, lichen etc. The usage of smart mats allows for a consistent look across all the stone pieces while still gives me that art input to paint in an out specific details I want to bring out.
It eventually evolved into something a little bit darker like my initial reference, since I decided to leave the stylization for things such as the roots and trees design (proportions, color, placement).
Also, since the stone material across assets is pretty uniform looking (in the reference as well), I thought of breaking it up using variations over the texture, particularly on the moss mask and dirt, then heavy grass, plant, and leaf placement to get rid of this uniform look. The color of lichen really works nicely as an accent to the less saturated base values so I think I’ll try to make use of it more since in real life it does appear constantly and covering very large areas at times.
I created a tool in Substance Designer that would allow me to create decals for moss, different types of lichen and fungi. At the texel density of the scene having scanned these wouldn’t have made a lot of difference quality wise, and the power of procedural iteration meant lots of variations really quickly.
Lighting & Color
Apart from that, I reworked the lighting to match my paintover a bit better, started playing with simple particles and subtle color balance since the main values for the textures are there already, so I can get a better feeling of the mood.
Excusing the random block out placement of a lot of vegetation, I did test a first grass blade pass to see how it works in between the newly created slates and rubble.
Also, the big roots have a 0 to 1 simple diffuse map on top of the tiling textures to break it up and add uniqueness to them with paler points, more brownish towards the bottom, etc. I used a world based tint to break color up on vegetation assets and ground.
Definitely will need to tweak the shapes of some of the roots and transitions between them so they seem more natural.
After playing with different lighting scenarios and ideas storytelling-wise, and receiving feedback, I decided to go in a different direction. The former light was muting all the colors and crunching the values together in the darker-mid range. Almost sunset time would mean enhanced the red part of my color palette, along with interesting lines created by treetop shadows.
Furthermore, the motorcycle narrative was a bit distracting from the composition flow and was too overpowering with detail. So I kept it simple and added more elements in the background to help a more gradual read from the foreground, instead of stopping halfway.
Once at this point, I painted a screenshot in Photoshop to try and see what the new direction is and what the next steps should be. As you can see, overgrown vegetation helps with the breakup and makes the structure much more interesting. On the big root shapes, even the moss and tint wasn’t helping the plain values and was looking dull, so I decided to go for making use of tertiary shapes in the form of thinner vines. These could also be used as points to place hanging ivy and more, which added to the visual interest.
Colors are more vibrant now, and to help with these accents I toyed with the idea of prayer flags hanging around, these created additional lines that could support composition, added a storytelling component, and helped break the grey, brown and green of the whole place.
Finishing the project
As a note, even though the place is heavily inspired by Cambodia, these flags are not necessarily from that particular place, but it is part of the personal stylization I am going for.
This was the first pass after finishing the concept, vegetation still not in. A lot more of fluffy moss, especially in between modules. Particularly useful in between crevices and on uniform surfaces that might look too boring.
After adding some vegetation and correcting the thickness of the vines.
Adjusted light for mood purposes, added birds hanging around and flying, and modeled a low poly temple to add in the background as the arch frames it perfectly and makes for a last attention spot in the image read.
Also added an extra layer on the ground material, that includes chunky roots poking in and out of the terrain, which is a simple material made in Designer, with opacity mask to show only the roots, then vertex blended in Unreal.
Final image! It has come a long way and I definitely learned a lot during the process.