Assignment 2: Real-Time 3D Environment

As this assessment essentially expands on the first assessment, some of research planning and conceptualisation for the environment and characters that will be developed in this assessment have already been created, which are showcased below.

Since developing the project in assessment 1, however, the ideas and concepts for the project have evolved, and so I thought it would be best to do some more research, planning and conceptualisation of how certain elements will look like in the final scene. In particular, I would like to find some more references for the environment, as I did divert a bit away from the original concept art I developed in assessment 1, as well look do some additional concept sketches of the cycladic character model that will feature in the scene. 

Further Planning and Conceptualisation of Cycladic Character Model

In assessment 2, it is required that the final environment features 2-3 characters in it to help convey a narrative. Before getting to that, I thought it be best to further plan and conceptualise the look of these cycladic character. In assessment 1, while planning and conceptualising the look of the head idol, I took some references from cycladic art to help inspire my concept sketches of the head idol, and as part of that, I also did a rough concept of what  the body of this head idol would look like; forming the first concept of what the cylcadic character in the final scene would look like.

In the first assessment I discussed how I added carvings to the side of the head's rear as well as some neck rings painted blue to add some visual interest to the model. I also discussed that I felt it made the idol feel it had some sort of cultural significance, with it resembling some sort of deity that an ancient civilisation worshipped. I have added similar rings to the arms of the character model as well as added a carved section to have a visually interesting way of separating the torso from the legs, but feel I should look at some further references to see if I there is anything else I should add to the model to create further visual interest to the model.

Shown above is a concept sketch for the character model that will feature in the environment. As stated before, it doesn't really diverge from that initial concept art I made while making the idol head concept art since I like the simplicity of the body.  I have made the body slightly slimmer, however, so that in proportion the head seems abnormally large, making it seem like the head does not belong to the body; further adding to the alien and sinister aura the idol head is supposed to have.

The other main element in consideration for this assessment is the environment in which the idol head and characters will be contained within. Like the character model, I had already come up with an initial concept art for how this will look in the first assessment.

While this concept art did provide great direction in the first assessment, I feel my ideas and vision for the final environment have evolved slightly. For one, the walls showcased in this concept art I feel look either too primitive (the outer two walls) or too man-made (the inner three walls). In assessment one, I discussed how I changed up the design of the altar as I felt that while the initial model I chose for the altar resembeld the concept art quite well, it felt too machined to feel cohesive with the other elements in the scene. Similarly, I feel the walls in the final environment should be designed to feel more consistent with the elements of the scene I have presented in the first scene.

Using the references previously discussed, I got to work on a refined concept for my environment. Rather than start from scratch, I used my initial concept art as a base, erasing elements I didn't like previously and replacing it using ideas and concepts I had gathered from the references. The most obvious change is with the walls of the environment, which have been changed from the rocks and man-made bricks to more rough and primitive stone bricks. the colour of these bricks have also been changed to have a warmer colour, which I feel better reflects the atmosphere of the environment and the references it is inspired by. The ground is now covered with some vegetation that surrounds the altar and the chairs surrounding it as well as the environment just generally being covered in more moss; further implying its age and state of abandonment.  The roots have also been reduced in size a bit due to concerns of being able to find suitable assets to represent them in the final 3D environment, but they now grow from both the ground and ceiling, and still all point towards the idol head in the centre of the room. The door into the chamber has also been changed to be built from a more crude rock formation and beyond that is a dirt floor tunnel with a stone staircase at its end; further implying that the  chamber is somewhere underground. Lastly, I have added in some instances of my character model concept sketch to get a gauge for how it would look in the environment and get some ideas of how they should be posed and animated. The character on the left has its arm stretched out to the idol, perhaps in an attempt to reach for it or just to point at it, while the character on the right is collapsed on its knees perhaps in awe or horror of the idol.

Here is a timelapse video showing the process of refining the original concept art, to incorporate ideas and references I gathered for this assessment.

With the conceptualisation of the environment and character model done, it is time to start preparing these elements. The first I wanted to tackle was modelling the character. Initially, I was intending to model the character by using the statue head I made for assessment one as a base, and then extending on from it to create the rest of the body. A base body was provided by the teaching team, however, and after watching the lecture discussing character model topology along with taking into consideration the time I had to create my model, I felt it was best I use the base body. There was still some work to be done, as using the base body would require me to remove the original head of the model, and put my own head onto it.

I first imported my head model into a scene with the base body model, lined the two models up and then cut off some faces off both to prepare them for attachment.

With the head and body prepared for attachment, I then proceeded to move the vertices  on the edge of the head's opening to somewhat line up with the vertices on the edge of the body's neck. I had soft selection enabled so that neighbouring vertices would move along with the targeted vertex based on their distance from it, resulting in the shape of the statue being preserved and still looking quite natural.

After going around the edge of the head's neck base and adjusting the vertices that compose it, I was able to line up these vertices to match the vertices of the body's neck. Next up I will use target welding in order to attach the head to the rest of the body. Shown in the images above, however, the head contains a lot more edge loops, which means as it currently stands, there will be some vertices on the head that won't be able to target welded to vertices on the body.

To fix this issue, I deleted some edge loops from the middle of head as currently  it is quite dense in this area due to these being added as holding lines for the model when it was being smoothed in the first assessment. I was concerned removing these edge loops would result in the shape of the head being changed, but upon deleting them, I found that the head shape didn't change much; perhaps this was because the model was already smoothed, so these lines did not have much purpoe anymore. 

Even though I did delete the edge loops properly, after closely inspecting the model for any topology issues, I found some stray vertices next to the middle edge loop of the head.

To fix this, I target welded the majority of these stray vertices to the closest middle edge loop vertex as they were often quite close.

There were some stray vertices that I could not get rid of which were on the nose. Deleting them resulted in some weird shading occuring on the nose that I was not able to fix. To circumvent this, I added some extra vertices to the bottom of the nose and connected them up so they formed quads, and then connected the stray vertices to these newly added vertices so that they also formed quads. I then had to create some more vertices and connect them to create quads to at the top of the nose bridge to cap off this section of the mesh.

After fixing up all the stray vertices, I noticed that one of the faces on the neck was missing/distorted. It turned out I had somehow accidentally target welded the vertices of the face to one of the vertices in the bridge of the nose; most likely while I was fixing up the stray vertices on then nose.

To fix this, I deleted the half of the head with the distorted face and then mirrored the remaining half to remake the other half.

With the head mesh prepared and the issues with it resolved, I combined it with the body mesh and began target welding vertices from the head to the body.

When I got around to the rear of the head, I found that I still had more vertices on the head than the body. The resulting mismatch meant that I had to create some new vertices and edges on the body in order to properly weld the vertices together, which I was able to do.

With the model complete I then moved onto UV mapping it. When I checked the initial UV map, it only contained data for the head, and that half of the UV maps were gone. This was most likely a result of deleting half of the face and mirroring the remaining half. This isn't too much of an issue as the UV map data for the mirrored half of face is stored in the same UV shells, but for texturing purposes I will want the different sides of the face to have separate UV data, so I will have to redo the UV maps for the face as well as add UV map data for the body, seeing as that was removed in the process of combining the two meshes.

UV mapping the head was easy considering I had done it for assessment one. I used the same orthographic perspective camera-based technique to create the different UV shells the head was composed of.

I then moved onto UV mapping the body, first creating a camera-based UV map and then splitting the body into different UV shells using  seams.

Most of the model ended up looking quite good, however I did notice some warping on the inner leg, so I added some more seams to section of the legs from the torso in order to get rid of it.

A lot of fixes and details were added to the head in assessment one through sculpting, so the same was to be done with the full character model. When I tried importing the character model, however, it turns out there were some issues with the model that I needed to fix before I could start sculpting.

I decided to continue to see if MudBox would tell me more about these errors and there ended up being a red point highlighted on the model. This ended up being a stray vertice on the body which I was able to get rid of.

I was still receiving some incomplete UV set errors in Mudbox so there was still something wrong with the model. It turned out that some of the UVs were reversed or overlapping. I deduced that this was most likely due to the camera-based technique not working as well with the full character body model as it did with the head. Since I already had the UV islands cut up, I remedied this issue by unfolding the UV maps which got rid of the reversed and overlapping UVs.

With the UV map issues taken care of I began to sculpt the model. I first focused on redoing the sculpting of the head, trying to closely mimic the same process I used to sculpt the idol head. I wasn't too bothered with getting it exactly the same, as I felt it being slightly different would be better than it looking identical.

I then brough the model into Substance Painter, applying the same method of baking the details of the high poly model into the low poly model in order to retain my sculpting work. The resulting texture work did end up with some loss of depth with the arm rings, however, I was fine with this as given the body is supposed to be worn down and malformed, it could be implied that the arm rings would be in a less pristine state than the neck rings, so I could work with this.

I then proceed to focus on doing the texturing of the head, applying the same techniques I did in the first assessment.

Before moving onto texturing the bulk of the body, I added a smart dirt layer which applied to the whole body and applied blue paint to the arm rings and the crotch area to match the concept art.

For the body, the intent was for it to look almost like a rusted metal body. I couldn't initially find a material that would best suit this look, but found a material called 'Medium Acoraspora Lichen' which texture wise had the appearance I was looking for. I adjusted this material to better suit the colours shown in my concept art and then applied this material to the body of my model. From there, I made iterations on this texture, first applying it to the blue paint areas of body to make the body texturing more cohesive, then to the face to also make it more cohesive, and then little touch-ups here and there. Lastly, I added one last layer of this material that was slightly greener in order to better match the concept art. With that complete, the model was fully textured.

With the character model modelled and textured, and the environment conceptualised it is time to start putting together the scene. Before fully putting it together though, The environment needs to be 'whiteboxed' in order to get a sense of how the environment will look like in 3D, and if it is of an appropriate scale. To get started, I created a duplicate of my diorama scene from assessment 1 and deleted the containing box, leaving behind just the statue diorama and the rocks forming the hole above it, which I left to get a sense of how high the environment should be.

I then began to mockup the environment, starting with the chairs. I used white cubes to represent the chairs for the moment. I used a chair megascan to get the scale right, making these cubes larger than the chairs as I felt that the character models would be slightly larger than an average human and so the chairs would need to be larger too. Speaking of which, I imported my character model to get a sense of its scale and found that it was quite large.

I also imported the texture maps of the character model to test that they worked and and as shown above they were able to be imported and applied successfully.

After scaling down the character model. I duplicated it and placed it in the scene to get a sense of how the two characters would look in the final scene.

I then placed some pillars around the scene to help define the size of the environment, later changing them to more cylindrical objects as I felt this would better mimic the concept art and make mocking up the rest of the environment easier.

I then created walls that connect between the pillars. To speed up the process, I created walls for one section of the environment and then copied and pasted that around to create the full temple-like structure.

Here is what the base structure looks like with the walls and pillars fully implemented

I then placed the camera inside the scene and moved it around and found that I was able to comfortable place it within the environment and capture all the elements required. I did feel however that the walls were to short and the pillars were not thick enough to feature prominently in the scene.

To remedy this, I increased the size of the pillars along with raising the height of the walls, which I made the environment have a more appropriate scale to the relative to the character models and other elements in the scene.

Next I created the pillar caps to add more depth to the pillars as well as prepare for the pillars that will be featured in the roof of the environment.

I then created pillars that were on a 45 degree angle that extended out from the top pillar caps and went towards the centre of the scene.

I then added a roof to the environment and then readded the hole in the roof, adjust the rocks so they would better fit this new roof.

This is what the scene looks like from an inside view. Already we are getting similar lighting to how it looked in the diorama with the addition of an environment. Lastly, I just need to add walls to the sides of the roof.

Before adding the side walls, however, I noticed that some of the pillars were not lined up to the corners or the middle of the sides of the roof, which would make using modular walls for the roof side walls hard to do. To fix this, I changed some of the angles the pillars face towards in order for them to line up better with the shape of the roof.

I then added the walls to the sides of the roof. I was unable to get these walls to fit exactly to the holes in the side roof, so there is a bit of overlap and clipping occurring, but due to it just being a white box, I was fine with this for now.

With those walls added in, here is the final mockup of the scene with lighting on to see how the scene will be lit up. Additional lights could be added, however, for the moment, the key light helps to highlight the statue at the centre of the scene, which is what I want.

Single Wall Module

To make the later development of this environment smoother, I tried to break the white boxed environment down into further elements. I noticed that one section of the walls could be repeated to create the majority of the environment, so a few variations of this section could be made to comprise the environment while still providing a feeling of variety and uniqueness to different parts of the environment. Most notably, I will need a variation that features a door, as one of the walls in the concept art features a door

Double Wall Corner Module

I found a similar section for the diagonal corners of the environment. As I mentioned before, the white boxed walls could not be modified too greatly to fit within the pillars, but the general idea still comes across, and in the final module, I will be able to make the assets that compose the module fit the desired specifications. A few variations of this segment of the environment could also be made to comprise the environment while still providing a feeling of variety and uniqueness to different parts of the environment.

Chair Module

The chairs will also be a module in the environment, with multiple of  it being placed in the scene. It will need to be developed to mirror how they look in the concept art, as currently it is just represented as a cube (with a normal chair next to it for scale).

Roof Module

With the environment white boxed, the last elements to plan was the character models and how they would be placed into the scene. I had already placed instances of the character in the scene to get a sense of scale for the scene and for camera shot reference, but now I needed to find animations and apply them to these models in order to help convey the story of the environment. To do this, I need to rig the character model to be able to animate/apply animations to it. I first prepared the model for rigging by ensuring it lined up nicely with the axes.

I then used Quick Rig to generate some guides that will form the skeleton rig of the model.

These quick rig guides were not entirely accurate, so I spent some time adjusting them, ensuring that they lined up from a front, top and side perspective.

With the quick rig guides adjusted, I then generated the mesh's skeleton rig and skinned it in order to link it to the mesh.

With this completed, the mesh was now attached to its rigged and could be moved using it.

I then posed the model to see if there were any areas where the model stretched unnaturally. For the most part, the model in most areas reacted well to the poses I put it in, so not much needed to be fixed.

I did find, however, that raising the arms caused part of the chest to expand and stretch in a strange way, which was not ideal.

In Paint Weight mode, I examined the left arm and found the following. The paint weigh of the left shoulder and left arm influenced a significant portion of the chest area, which was likely the cause of it stretching and expanding whenever the arm was raised.

To fix this, I reduced and smoothed out these paint weight areas so they would have less or no influence on the chest area. 

I then mirrored the paint weights from the left side of the model in order to have these changes applied to the right hand side of the model so that I wouldn't have to do it manually and to make sure the model's weights remained symmetrical.

With those paint weight changes applied, it can be seen that when the arm is raised, the chest does not stretch or expand as much, making for a much more natural looking pose.

I did a few more poses just to ensure that everything was alright with the model and from this experimentation I found that I was happy with how the model had been rigged.

To apply animations to my character model, I brought the rigged model into Mixamo, where I then proceeded to find animations I could apply to them. To be in line with the original concept art, I wanted the animations to convey that the characters were either in worship of the model or fearful of it. Shown above are three animations I found that I felt fit this criteria, with the first one potentially having the character kneeling down and worshipping the statue, the second one could be in fear of it or in horror to see the state of their body, while the third one could be in awe or inspecting the statue.

With the animations picked, I then imported the rigged model into the unreal project.

I applied the same character model texture to it to ensure that still worked despite the extra rigging work done to this version of the model.

I then brought in the different models with the animations applied to them, all of which were successfully able to be imported in with no issues.

I tested that the textures could be applied to them as well as see that their animations could play with no issues as shown above.

Removing the static character model meshes, I then placed the three animated models around the altar to get a clearer idea of how they looked in the environment.

Already I was starting to get a good sense of how the final scene should look like. These animations help contribute to conveying the supernatural and sinister nature of the statue. All of them seem in awe of it, however, they also convey some sense of fear in some way. The character on the left seems to worship the model, but it's inability to look directly at it implies it also fears it. The character in the middle is able to look at it directly, but it's erratic movement implies that it is somewhat in shock and awe at being able to directly witness the statue. The character on the right carefully inspects the statue, with it's hands slightly out towards from it, implying it is somewhat cautious of the statue.  In combination with the lighting, this really helps to enhance this sinister feeling that the statue is supposed to convey, and once the rest of the environment is developed, it will make the focal elements shown here more immersive.

With all of the main elements of the environments now planned, it is time to start developing it, starting with further developing the white box modules developed earlier. I first started by looking through Quixel Bridge for megascan assets I could use to replace some of the white box module geometry I had created, and found this Roman Column model that looked similar to the pillars I had drawn in my concept art.

Replacing the white box cylinders with the pillars, I found that the roman columns fit quite well into the module. The column comes with no capital but does come with a base, so I used the base of the roof support pillar as the capital of the wall pillar which ended up meshing together quite well.  The colour of the pillar I felt were a little too plain and will need to be adjusted to better fit the aesthetic of the general environment, but that will be done once I have sourced other assets to compose the different modules.

For the walls, I originally was going to compose the walls out of various square-shaped stones to build up the wall. While browsing the Quixel Bridge megascan library, however, I found this Japanese Mossy Stone Wall, which is composed of various stones that are somewhat square in shape, and so I felt it captured the same sort of aesthetic I was going for with my environment walls. Making use of this megascan asset will allow me to more quickly develop my base modules and from there create variations of them.

Using multiple of these mossy wall models, I created the wall segments of the single wall module. The overall result looks quite good, albeit a little repetitive.

To alleviate this, I rotated some of the mossy wall modules to make the texture of the walls not look as repetitive. While it still looks a bit repetitive, this should be enough for now, as later on I will be creating variations of this module that will help the walls of this module look more distinct.

I felt the wall module still required a bit more detail. In the original concept art, the walls had skirting on the floor and at the point where the base wall connects to the roof wall segment. I found this japaenese wood carving beam, and felt that fit the aesthetic I was going for with this element.

I then added that to the base single wall module, and with that it was complete.

I then applied a similar workflow when developing the base double wall module. I was able to make the mossy walls fit within the tapering areas created by the pillars by scaling them and the resulting wall still looked nice. For the second wall segment, I wasn't able to scale the mossy walls to nicely replace the white box geometry located here. I ended up having to source a variant of the Japanese Mossy Stone Wall that was of different proportions, and used that to create the second wall segment of the module. Lastly, I created the skirting for the module similar to the single wall module.

The next module to develop was the roof module. To prepare it, I ensured that the rock formations forming the hole all neatly connected to each other in a circle.

I then went to find models that could be used to compose the rest of the roof module. I didn't want to use the mossy walls model I had used previously, but did want whatever I used for the roof to remain consistent in aesthetic to them. I was able to find this Japanese Shrine Stone Floor model that looked similar in appearance to the mossy walls, and decided to use this to compose the roof module.

Using multiple instances of  this model, I was able to construct the same octagonal shape for the roof. It was a little bit more rough, although I feel this makes it feel more organic. The use of multiple versions of this model overlapping to create this shape also does create some weird texture overlap, however, it isn't too noticeable, and the addition of some branches later on should help to hide this. The roof will not be a major focus of the scene , so this shouldn't be too much of a concern.

Finally, I sourced some assets to make the seat. I was able to find this Roman Capital model that could act as the base of the seat, while the Roman Stone Ruin I had used for the altar could be used to create the arm rests of the chair.

Putting these assets together, I was able to put together a chair that resembled closely to its depiction in the original concept art. There is some difference in the textures between the the two models, but now that all the base modules are done, I can make some slight edits to the materials of the models that compose them in order to make all them more cohesive.

Taking a look back at the modules that will compose the environment, it can be seen that for the walls, the environment can be entirely composed of two wall modules, which can be done by making putting together four single wall modules and four double wall modules into a circular formation. If we were to compose the environment out of just four of each of these base modules, the environment would look too 'samey' and repetitive, and so some variants of the base single wall and double wall modules need to be created.

Single Wall Module Variants

For the first single wall module variant, I decided to create a variant that had a door. The original concept art depicts one of the walls behind the statue having a doorway made out of rocks leading to a staircase. For the doorway, I found a Japanese Gravestone asset in Quixel Bridge and used three of them to create it. To ensure that it was to scale, I placed an instance of my character model to ensure that the doorway was of an appropriate size relative to it. Lastly, beyond this module would be a corridor with a staircase at the end of it, but for now I will develop that separately, as the focus for now is on creating variants of the single wall module.

For the second variant of the wall, I added some debris to the front of the wall to give it some visual distinctness from the base single wall module. While it is only quite a simple change, the purpose of these variants is to make the walls feel a bit less repetitive and more organic, but at the same time it should not take the attention away from the main focus of the scene; which is the statue that will be at the centre of the scene, and the character models that surround it.

As the final environment will be composed of four single wall modules, I only needed to make one more variant.  For this variant, I experimented with adding various tree branches and tree roots to match some of the walls in the concept art that had roots sprouting out of them. On a recommendation I received in the workshops, I also added fern plants to the wall. Using these single wall module variants, I can create parts of the final environment that are similar in structure and maintain a cohesive base design, while also adding some variety and visual interest to the walls in order to stop them from looking repetitive.

Double Wall Module Variants

For the first double wall module variant, I added some rock formations in front of the wall. Again, the intention of these variants is not to make them the focus of the scene, but to break up the monotony of just using the base modules, so I am only making subtle additions and changes here. It's quite hard to see the rocks in the unlit view, so I have provided a view of this module from the detailing lighting view (as it is even harder to see in lit view given the current lighting setup).

For the second double wall variant, I experimented with it in the same I did with the third single wall module variant. I added  a variety of branches and roots to the module and then added some further rocks/debris at the bottom of the walls of the module.

For the final double wall module variant I expanded on the same work done on the previous variant. I added some more tree roots to the bottom of the module, as well as some large rock formations which are all centred around the centre of the module. 

With all the environment modules and variants created, it was time to start to put together the final environment.

Before starting to put together the roof and the walls of the environment, there was an element of the environment I had somewhat neglected up until this point, which was the floor. I wasn't really sure how I wanted to it to look, as in the concept art it is just depicted as a stone floor segmented into large panels that radiate out from the central altar. I wasn't really sure if this fit the overall aesthetic of the environment in relation to how the modules developed work. So for the moment, I used the same Shrine Stone Floor asset used in the roof module to create a rudimentary floor for the environment, but I may replace this later with a more detailed floor once I have put the rest of the environment together.

To start off, I first brought in the altar diorama I had developed for assessment one into the scene and placed it at the centre of the scene.

I then worked on aligning the roof module and one of the single wall modules so that they were at the right height in order to be connected. From here, I can start bringing in the other modules I developed in order to put together the environment.

I added in the other three single wall module variants to each side of the environment. From here, it can be seen that in between each of these single wall modules, a double wall module can be added in to complete that segment of the environment. It should be noted that the two single wall modules closest to the camera are facing away, and due to backface culling, we are able to see through them, however, when within the environment, they will display properly.

Taking the base double wall module, I inserted it in between two of the single wall modules, showing that the different modules do indeed fit well together.

I then added additional double wall modules of different variants to compose the rest of the environment.

With the wall modules enclosing the statue added in, the base geometry of the environment is complete.  It can be seen, however, that there is an empty hole where the door variant of the single wall module is, and so some more of the environment will still need to be developed to provide an appropriate context for its existence. In addition to that, it can be seen in the image that light is bleeding into the environment despite the presence of the roof. This is most likely due to the back face culling on the wall model assets, and so a fix for this will need to be implemented as well.

First up, I added a corridor that connects to the doorway variant single wall module.  Rather than make a completely new model for this segement of the environment, this corridor is just made up of three single wall modules. 

In the concept art, this corridor featured stairs that go upwards. I was able to find this Japanese Shrine Stairs asset that fit the same aesthetic as other elements used in the scene, and decided to use it for the environment.

I then inserted the stair asset into the corridor and it fit quite well.

To fix the light bleeding issue, I encased the environment in a geometry box. The environment does feature a hole that light is supposed to come through, so I added a hole in the geometry box to allow for light to reach it.

The result of this is that light no longer bleeds into the scene, with the only light coming through the hole.

I do want the rest of the environment to be somewhat visible, so I added an additional light near the hole which is used to light up the surrounding environment.

One element I did want to redo after doing environment composition was the floor. As I stated in the previous section, I rudimentarily put the floor together by putting together a couple Japanese Shrine Floor assets together. While I do like the depth it provides the ground of the environment, it does seem somewhat repetitive and the seams between the assets that compose it are quite visible (especially in lit view). Therefore, I felt it was best if I created a floor for the environment that was a little more 'hand-crafted'.

To start off, I created a plane that was around the same size as the floor module.

The material I would be applying to the plane was a blend material. A blend material is material that can combine and blend up to 3 materials, and can be further edited through the use of vertex painting. The first step of this process was selecting materials that the blend material would be composed of. For the base material, I chose this mossy stone floor material that I felt was similar in aesthetic to the japenese shrine floor asset and the rest of the environment. As the cave environment has been supposedly abandoned for some time, the middle layer I chose to add was a forest mulch material, which acts as a layer of dirt/grunge on the stone floor. Lastly, for the top layer I chose a wild ferns and grass material that can be laid on top of the forest mulch layer to imply subtle grown on this dirt, enhancing the idea that this place has been abandoned for some time.

With these materials selected, I then proceeded to create my Ground Blended Material that I would then apply to the plane I had made earlier.

When I applied the ground blended material to the plane, however, I found that the texture was quite large on the plane, and that I was unable to do any vertex painting.

This ended up being due to the base Unreal plane not containing enough vertices to properly do vertex painting. To solve this, I created my own plane in Maya and subdivided it a few times to create a plane that would allow me to do vertex painting. I then imported this into Unreal to start vertex painting it.

After importing the high vertex plane into Unreal, I stil found that the texture was still quite large on the plane, so this issue was not with the vertex density of the plane.

After tinkering with the settings of the ground blended material, I found I could reduce the size of the texture on the plane by increasing the tiling x and tiling y parameters of the different materials. After messing around with the settings for a bit, I found tiling parameter values of 7.5 to be what I was looking for.

With the textures of the blended material displaying properly on the high vertex plane, I could finally start vertex painting it to provide it more detail.

I first painted in some patches of the forest mulch, with most of it extended out from the edges, with some odd patches here and there .

On top of those dirt patches, I then added some of the wild grass and fern material to imply the growth of plant matter, which I felt looked quite good.

Ground Plane Experimentation and Lighting Adjustment

To see how it would look, I placed the ground plane into the environment. The floor was a bit hard to see at first, so I also adjusted the lighting in the scene slightly so that the the light would scatter into the environment there and therefore make the floor and the rest of the scene more visibly clear. I quite like the look of the floor here, and felt that the lighting helped to accentuate the normals of the different materials, which were hard to see previously as I had been working in unlit view. I do feel, however, that more dirt and grass could be added.

For set dressing, the only thing I really needed to add from the concept art were some fern plants that surround the chairs around the altar. From here, I also made adjustments to the positions of the chairs and the characters present in the scene. Further adjustments were made to the lighting to make it feel more natural but at the same time warm as this environment is supposed to be contained within a jungle. For the moment I am quite happy with what is currently there, as it doesn't distract too much from the main focal point of the scene, but instead draws attention towards it.

Before going further into the set dressing work done, I want to bring attention to the changes of the lighting setup. In the old lighting setup, there was a lot of focus on the focal point of the scene, however, at the same time it made the rest of the scene quite dark by comparison. In some respects I felt that this was unrealistic, as I felt the light would scatter around the environment a bit more even though it was coming out of a hole. So in the new lighting setup, I adjusted some lighting settings, as well as added an extra point light, and the resulting lighting feels a lot more realistic, natural and cinematic.

While this is not strictly set dressing, I added another doorway variant single wall module to the other side of the corridor. I added this because with my final environment shots in mind, I felt utilising the corridor in one of the shots would be nice considering I went through the effort of building it.

Previously, however, due to the backface culling of the models that comprise the module, you can currently see through it and into the rest of the environment.

Adding in the extra wall ensures that this does not occur as shown in the above image. With this module I added, I now have a functional doorway that works from both sides. Using the image above for reference, I feel my first shot will have the camera going through the doorway towards the statue, before rotating the camera and facing the statue front on.

In the workshops, I got feedback that the placement of my ferns felt quite strategic, a sentiment I agreed with. I was advised to make it feel more natural, I could add some plant decals to help the placement of the ferns feel more organic. I found this Green Weed Vines decal that I felt could help to 'root' the ferns into the ground, providing it with a better basis for their existence in the scene. I also found an Asphalt Pothole and Rock In Ground decal that I could place around the plants to provide the ground plane a bit more variety in texture around the area that the plant is located.

Already the addition of these decals helps to better ground the fern assets. I especially like the weeds in the water, implying that the ferns are growing due to the water nearby.

To add some further variety to the ground around the plants, I also found a moss patch decal that I would also place around the plants.

The addition of the moss definitely helps to provide variety and context to the plants placement in the scene.

While placing some moss decals, I noticed that they could be applied to the chairs. The chairs in the concept art do feature some moss and plant matter on them, and as they currently still looked quite clean. I decided to add the moss to the chairs, experimenting with different placements to make each of the chairs look more unique.

While going around placing decals, I found that the horns of the altar had changed colour. Previously in assessment one, I had modified the model's material to better match the colour of the altar. This had changed, however, and I realised it was because the arm rests of the chairs in the scene used the same model, and I had modified it earlier to match the base of the seat better. 

To make it so that I could modify the material of the horns while ensuring that the material of the arm rests of the chair do not change, I created an instance of the material and then adjusted the properties of this new material instance so that it would match the altar better.

After applying this new material instance to the altar horns. They now match in material to the altar much better. At the same time, it can be seen that the arm rests of the chairs in the background have retained the same material they had before.

Finally I duplicated and flipped the roof to give the cave hole some more dimension, and added some scaled up fern plants to help imply and outside environment.

After showing my work in the workshop for feedback,  I was told that a lot of my environment work would go unnoticed with how dark the scene was. I was advised to add some additional lights diegetically into the scene to help brighten the scene, as well as provide a feeling of depth to the scene. I found this oil lamp quixel asset I felt would fit quite nicely in the environment. I then downloaded a fire particle asset pack which contained a candle-like flame visual effect, and combined the two to make my own asset that I could use as lighting objects in my scene.

With the main light of the scene increased slightly to provide a bit more visual clarity, the addition of these oil lamps helps brighten the scene by illuminating the walls of the environment, which are usually within darkness, which has the added benefit of making it easier to grasp the depth of the scene.

To add more life to the scene, it was recommended to make the foliage look like it was moving slightly due to wind by manipulating their material shaders. When I inspected the material graph for the foliage material, I found that there were some nodes that were contained within a section labelled 'Wind'. By default it was set to false so I set it to true it see if it would give me the wind movement effect I was trying to achieve.

After adjusting some of the nodes in the Wind section, I got the following effect on my foliage. The effect is subtle, but I feel that is somewhat realistic given the only opening to let wind in to the scene is the hole at the top of the scene, and it still helps to bring more life to the scene.

Even with the characters, oil lamps and foliage, the scene still felt quite static. The issue was that the movement of these objects are quite subtle/minute, and so it is hard to notice these movements, even less so when the camera is moving. I was advised to add something like water dripping from the ceiling or insects that flew around the scene. I initially wanted to added water dripping from the ceiling, however, the asset I found Unreal Marketplace was hard to modify, and in the end I felt it distracted too much from the core elements of the scene. I instead found this Particle and Wind Control System asset, which contained a firefly particle system in it. I quite liked this and experimented using it in the scene.

With the environment complete, I decided to name it The Tomb of Malum, with malum meaning evil in latin; further contributing to the sinister nature of the idol. I put together a sequence of shots to showcase the final environment which are shown below.

And last but not least, here is a video showcasing the final environment.