Assignment 1: 3D Asset Collection

To start off the design process, research for the project was done by finding references for the main elements that will make up the final real-time 3D environment. These elements include the cycladic figurine, the altar it sits upon and the cave environment that it will be contained within.

Various image references were collected for the different elements of the environment. As shown above, the reference images have been grouped together based on what element of the environment they act as a reference for.

As the hero asset of the first assessment will be the cycladic figurine (and later the character model in assessment 2 that will extend upon this figurine) that mirrors the idol from Raiders Of The Lost Ark, a few images of Cycladic art were found and collated together. From these references, common characteristics in these cycladic figures could be found such as the fact that they all feature tall figures with lanky limbs posed with their arms folded, and long faces predominantly featuring a nose and not much else. Such features will be ones that I will have to consider when designing my own cycladic figurine; with the image I'm drawing the most inspiration from in this regard being the one in the top left corner. Along with that, I made note of two images in the bottom left corner that feature elongated heads, which appeal to me and will want to include in my own design. Lastly, while I could only find one image showing this, the top right image shows that these figurines can sometimes feature carvings/etchings which I also want to include in my own design to give it some more visual flare.

The next element I wanted to consider was the altar the statue would sit upon. At first I looked at some different stone pile altars shown in the top right corner but felt it was a bit too natural to feel like an altar. Next in the top left and in the middle row I looked at more man-made looking altars, however they felt a little too 'machined' for a cave environment. The story for this environment details that 'the cave appeared to have been built between out of handsmade stone bricks whereas other parts are constructed from stacked boulders', so I felt I should be looking for an in-between for what I had currently found so far. I found the design of the altar showed on the left of the middle row peculiar and did some further research to find out it was called a 'four-horned altar design'. I did some further research and found some other four-horned alters that fit the level of degradation and aesthetic I was aiming for. Furthermore, combining it with the stone pile altar, I feel it could help make the altar better fit the cave environment.

Real-Time 3D Environment Temple Ruins Chamber/Interior References - Mark Auman, 2023

While the cave environment is more of a focus for the second assessment, it was still important for me to gather references for it now as the scene constructed for the first assessment will contain elements such as the set dressing that will be expanded upon in the second assessment. Seeing as this project was inspired by Raiders Of The Lost Ark, the references I looked for included the chambers and interiors of temple ruins. I was drawn in particular to the images featuring foliage and overgrown branches/roots which have seemingly taken over the structures they are within. This is in line with the story presented for the project which states that 'the opening in the roof of the cave has allowed water, light and plants to get in and take root in the cave'. For my cave environment, I want to feature branches/roots throughout to help imply the sinister/supernatural nature of the statue located in the centre of room by having them grow towards it.

From the references gathered, I started drawing thumbnail sketches for these different elements of the environment to hone in on what I wanted the final project to look like.

KNB137 - Cycladic Statue Thumbnail Sketches - Mark Auman, 2023

Cycladic Statue & Character Model Rough Concept Sketches

KNB137 - Cycladic Statue & Character Model Rough Concept Sketch - Mark Auman, 2023

KNB137 - Cycladic Statue Concept Sketch (Coloured Rendition) - Mark Auman, 2023

KNB137 - Real-Time 3D Environment Concept Sketch - Mark Auman, 2023

KNB137 - Real-Time 3D Environment Concept Sketch (Coloured) - Mark Auman, 2023

KNB137 - Real-Time 3D Environment Concept Sketch (Coloured with Sky Light) - Mark Auman, 2023

After coming up with a design for the cycladic statue I moved onto conceptualising the 3D environment that it would be contained in. As discussed before, while the environment is more of the focus for assessment 2, conceptualising how it will look now will help inform the design choices I make in this first assessment and ensure the aesthetic remains consistent between the two assessment.  

Above I have provided the original scanned sketch, a coloured rendition of it, as well as a version that showcases what the lighting will potentially look like in the final scene. The project description states that the walls of the cave are composed of both handmade bricks and stacked boulders, so I did a mixture of both with the walls near the entrance to the room being made of bricks, while the outer walls are made of stacked boulders. To provide a sense of structural support, stone pillars and skirting line the walls.  At the centre of room sits the statue atop a four-horned altar that is partially encased in rock, as if to imply that the altar was carved out of this stone much like the rest of the room. This sits upon a circular stone brick platform which is situated upon a floor made of segmented pieces of stones. The lines of these stones line up with the  stone platform and naturally bring the viewer's eyes towards the statue in the centre of the image. Evenly spaced stone seats surround the statue as specified by the project description, and along with that plant roots sprout from the walls and are creeping towards the centre of the image, further bring attention to the statue. For additional set dressing, a lot of the environment is covered by moss, some of the brick wall is exposed to reveal dirt behind it giving context for where the roots are growing from. Taking a quick look through the Quixel Megascan library, I should be able to find assets that can be used to represent the environment quite accurately to how it is depicted in this concept sketch.

Shown below is a video showcasing the process of creating a coloured rendition of the sketch. I found this process to be very helpful in nailing what sort of colours I want for the environment and how the lighting will look in the scene.

KNB137 - Cycladic Statue Clean Turnaround Sheet - Mark Auman, 2023

KNB137 - Cycladic Statue Clean Turnaround Sheet (With Guides) - Mark Auman, 2023

With the concept sketch and turnaround sheet for my cycladic statue complete, I then started modelling it in Maya 2023.

I started off by adding in my clean turnaround sheets as image planes so I could use them as reference as I modelled. I then started modelling the base of the neck, and from there modelled the shape of the head based on the front view perspective. In the final image, I extruded the front and back of the head to somewhat match the extents of the head from the side view perspective. Through this, the base shape of the head was defined, however, the back of head was still flat and would need to be extended out to match the side view.

In the first image shown, I used a combination of extruding, loop cuts and vertex manipulation in order to make the head shape be more accurate to the turnaround sheet sketches. I also used bevelling to add some extra dimension to the head as well as define where the nose would be for later extrusion. In the second image, I then started modelling the nose, which I found to be somewhat of a challenge to get looking right from both perspectives as well as have it taper smoothly into the rest of the head, but I was able to get there in the end by making the faces near the top of the nose triangle and bring the brow area of the head forward a bit. Lastly, I extended the back portion of the head to match the side view sketch and from there, adjusted other areas of the head to make the face shape more natural; primarily the cheek area which has been brought back slightly in order to make this area of the face look more rounded. At this stage I also added in the two neck rings to the model, which was also hard to match to both perspective and in my opinion looks quite janky, however, I think this has worked out as it makes the statue feel more organic rather than it looking like it was machined.

In the next stage of modelling I decided to add some holding lines to model through loop cuts so that later when I smoothed it, edges that I wanted to more defined remained intact. I also added holding lines at the eye line of the model as I then manipulated and extruded the points and faces around this area in order to create the eyes and brow ridges above them. The second image displays a smooth preview of the model that shows that these holding lines do their job quite well, with the face being smooth while still maintaining it's definition where required. The final image shows the same smooth preview without the wireframe. It can be seen that there are some visible lines on the model due to the density of some of the holding lines, however, like the neck rings it somewhat adds to this idea that the statue is more hand-made coming from an ancient time. In the event that I do want these lines to be less visible, I can most likely smooth the model out a bit when sculpting to get rid of them.

Final model smoothed out - Mark Auman, 2023

With the final model complete, I needed to UV map the mesh to prepare for it sculpting and texturing.

UV Checker Texture - ByValle (16 June, 2021)

To check the current state of my model's UV's, I applied the above texture to the model.

As shown in the image above, the texture ended up being quite stretched. The neck area isn't too bad with the checker texture being somewhat recognisable there, however, the head area is very distorted. This is expected since as shown in the second image, the UV islands for the mesh haven't been cut up properly.

To create the UV map, I started off by created a camera-based UV map. After this point I would usually the 3D cut tool to select edges and create seams in the model and define areas that I would unfold to create UV islands, however, I have found this method to be somewhat time consuming, especially with a mesh containing various loop cuts such as this one. I instead followed a method shown in class wherein you select facing a certain direction and then using one of the orthographic cameras, use camera-based UV mapping in order to create the UV island for that section of the mesh. The above images show the process for the top of the statue's head.

I continued to use this method to UV map the rest of the model, with the result of which being shown above. The checker pattern is a lot more recognisable and is experiencing a lot less distortion. The second image shows the UV islands for the final UV map alongside the model, with the third image showing a clearer view of the UV Map

Now that the I have the final model completed and have UV mapped it, it was time to add further detail by sculpting the mesh. In order to do this , I brought the final model I made in Maya into Mudbox, where I could then use sculpting tools to work on the model further.

Before starting to sculpt, I subdivided the model three times in Mudbox in order to give myself enough polygons to work with when sculpting details into the mesh. It would also allow me to easily extract the sculpting work I do as a normal map texture which I can then apply to the original non-subdivided model which will help the mesh perform better when taken into Unreal Engine.

Once I subdivided the model, I then proceeded to sculpt the mesh. I decided to sculpt using different layers that each focused on separate parts of the sculpting process. For the first sculpting layer, I focused on smoothing out the mesh to get rid of some the dense lines present on the model that I have previously mentioned along with generally making the model feel more organic.

The next sculpting layer I focused on adding some texture to the mesh. Using the imprint tool along with a variety of different stamps, I was able to give the mesh this subtle rough stone-like texture.

On the final layer, I turned my attention to the side rear of the head where the carved pattern would be. Initially I was going to put this section on through texturing, but I felt it would be good to experiment with using sculpting to help this area be more visually distinct. I once again used the imprint tool with a stamp that featured straight lines in and angle it to create an imprint in the mesh similar to my original concept art. After some cleanup, I got the result above.

Happy with how the sculpted model looked, the last step was to extract the sculpted model as a normal map texture that can be applied to the original model. This is important because as shown in the image above, the subdivided model I sculpted on contains 354688 faces, which if I were to bring into Unreal Engine as is would be very performance inefficient. As an alternative, I can instead bake the details from this high-poly count model into a normal map texture, and then apply it to the original model which has a much lower polygon count.

Using the Extract Texture Map feature in Mudbox, I was able to extract a normal map from the sculpt model, save it as a PNG texture file, and apply it to the original model.

Normal map applied to copy of the original mesh (left) next to original mesh (right)

With sculpting done, the final thing I need to do with the model is texture it.

To prepare the model for texturing, I created an ID matte version of the model which will make the texturing process go a lot faster. In terms of base colour, the model from concept art is comprised of two colours, gold and blue, so by using ID mattes, I can define these areas beforehand so I can quickly fill just the areas with the textures/colours I need.

I then took the model into Substance 3D Painter where I would texture it.

To start off the texturing process I added a base gold texture to the whole model. I realised after this that I would only need to apply textures through ID Mattes to the neck rings and the carved rear section of the head, which was where I needed to add a blue-toned version of the gold texture. From there I did a bit of blending between the two sections of the statue to make the colouring feel more organic.

Once the base texturing was completed, I focused on adding some extra details to make the model mirror more closely to the original concept art. In the first image I focused on adding some patches of lighter gold that is less metallic to make the statue look more weathered.  In the next image, I added some darker patch of gold that were more metallic to provide even more detail and help better define features of the  model such as the eyes. For some final details, I added a few scratch marks to the model.

With the texturing done, I then exported the texturing done in Substance Painter as the four texture maps shown above.

Taking the model back into Maya and applying all the textures, the model appears as shown above on the left. Along with it is the base model (right) and the model with just the normal map (middle). 

Redoing the Sculpt Detail Application & Texturing

After completing the previous texturing process, I learned that while the method of extracting the sculpt details into a normal map and applying it to the lower poly mesh I used previously could work, it was not the most viable and easiest solution. After watching through the Week 3 Lecture, I learned you could instead bring your low poly model into Substance Painter, and then save your high poly model from Mudbox as its own model and apply it as a high definition mesh on the low poly model, which then applies the detail of the high poly model as texture maps onto the low poly model. Using this knowledge I restarted the texturing process as I felt doing it this way would result in a more streamlined process and higher quality final model.

With the high poly sculpt model baked into the low poly model, the base model for texturing already contains a lot of detail and will make it easier for me to more accurately texture and preview what the final model will look like.

The ID Matte Model I made earlier was also baked into the model to make the texturing process smoother.

I first applied a damaged gold smart material to the model and already it is looking quite reminiscent of the original concept art.

I then added the blue paint to the back of the head and the neck rings. I felt it looked a bit unnatural like that, so I added another painting layer and painted some splotches of paint on the edges of these sections to make it look as if they were hand painted.

I then worked on adding splotch like areas where the statue is not as metallic in order to convey that the statue has gone through some wear and damage

The next step I added some further touch ups to the model to make it look like it has more wear and damage, but used an assortment of different brushes such as blood splatters, bullet holes and scratch marks to provide more variety to the damage that the statue has sustained.

I felt while the statue did feel a lot more worn down, it still felt quite new, and wanted to make it a little more aged.  I added a clay like texture material to the sides of the model, to make it seem like the model is starting lose its lustre and further add to the wear of the model. I feel this works well, especially on the neck rings, and implies that this statue may be actually stone and has instead been immersed submerged in gold rather than be fully made of it.

I still felt that the statue needed something else to make it really feel it was worn down and damaged. In my concept art, there were some darker splotches throughout the face that I really liked, and wanted to somehow emulate that further. Taking into consideration the fact that the statue will be located in a cave and will supposedly have been there for some time, I used a dirt smart material  to add a final layer of grunge to the model.  I then added a mask and adjusted it so that it looked like the dirt was speckled throughout different points of the model.  The addition of these dirt details I feel brings the whole look of the model together, enhancing and providing clear context to why the previous clay layer is there. With that, the texturing is done. Redoing the model using this new method I feel has resulted in a much higher quality texturing job, resulting in the final model looking a lot better than if I had stuck with the method I had found on my own.

Creating an empty Unreal project and importing the model into a level, it can be seen that the model currently only has the basic AIStandard Surface material applied to it.

To apply the texturing done previously to the model in Unreal Engine, the texture maps shown previously are imported into the project, and the AIStandard Surface Material is opened up in Material Graph. Texture Sample nodes are created containing the different texture maps, which are then connected to the corresponding properties in the master node, which should give the model the textures it needs.

As shown above, the textures have been successfully applied to the model in Unreal. It does look quite strange, however, this is most likely due to the current lighting setup of the scene rather than any actual issues with texturing the model.

Inspecting the model in a separate scene which contains an HDRi skybox shows that the model textures have indeed been imported and applied properly. 

With the statue model imported into Unreal, it was now time to start building a diorama scene for it to be contained in, which will later be expanded into a full environment in assessment 2.

Unreal comes with a library of high quality megascan assets. Going through the library, I found a few assets I could make use of to build up some of the elements in my scene.

Putting together these different assets I found, I recreated the altar section of my sketch. As these are assets are sourced and I did not make them myself, it was not an exact 1 to 1 of what I drew in the concept art, although it is a good approximation.

The scene I had set up the assets in, however, already had a lighting setup prepared. As I wanted full control and understanding over the lighting in my scene,  I brought the assets into a new level which had no lighting. The above image is the scene on Unlit View Mode.

Using the environment light mixer, I added in a sky light, and then a directional light to add some lighting to the scene.

I then added a sky atmosphere and volumetric  clouds to help provide context to the lighting of the scene

As the final environment will be inside a cave, I encapsulated the diorama I had made inside a geometry box.

Using another geometry box slightly smaller in size and with its Brush Type set to Subtractive, I hollowed out in the interior of the previous geometry box so that there was a cave-like space inside for the diorama to sit in. No light is able to get in, however, so the above image is also shown in Unlit View Mode.

To allow light in, I needed to make a hole in the ceiling of this box I had made. To do this, I created a geometry cylinder, set it to subtractive mode, and placed it in the centre of the ceiling of the box, creatng a hole that is directly above the diorama.

Going back into Lit View Mode, the result of the last few things I did is that the inside of this box is now lit. Due to the direction of the directional light, the light isn't shining directly on the statue, so I may need to adjust it, however,  I can also experiment with placing the hole in different locations in the ceiling and making use of different angles for the directional lighting to see what will best work.

The above image showcases some experimentation of placing the hole off-centre and then angling the directional light to shine on the stature which was quite nice to play around with.

Referring back to my original concept, however, I felt having a centered light shining on the statue was still the best fit, as with the symmetrical design of the cave environment I had designed in my concept art, having the light off-centre would clash with the rest of the design. In addition to that, this direct over the top lighting adds this ominous feeling to the statue, contributing to the supernatural and alien-like feel I am aiming for.

I then added an exponential fog object and played around with the settings to make the light shaft coming from this hole more visible, matching more closely to the original concept art I had.

The rough cave geometry I had setup I felt was way too lit up by the directional light. To fix this, I then combined all the geometry meshes that make up the rough cave setup into a static mesh. Doing so made the environment feel a bit more accurately lit and from here I could focus better on adding additional lights to bring more attention to the statue at the centre of the scene. Along with that, I added some moss decals to the assets in the scene to more closely mirror the concept art.

To make the statue more visible, I added in some point lights around the statue to create a three-point lighting set up around it which allows the details of the statue to be more clearly viewed

It was at this point, however, that I noticed that my statue looked a bit off. Comparing it to how it looked in Substance Painter, the statue in Unreal looked way too metallic, even when taking into consideration the difference in environment. Analysing the material graph for the statue's material, I realised that the statue's occlusion roughness metallic (ORM) map was set to sRGB and was sampling the colour of the map. Instead, I disabled sRGB on the map and set the sampling type to linear colour. After doing this, the statue head looked a lot closer to how it did in Substance Painter as shown in the second image. I still felt it was still slightly off, so I experimented a bit more to get closer to the look I was aiming for. I knew at this point that the ORM map stored the different maps it contained in its different colour channels, and that it was sampling these maps linearly, so I reduced the saturation of the map from 1.0 to 0.875, which slightly reduced the metallicness of the model to a point that I was happy with as shown in the last image. This does result in the ridges on the top of head from the base model appear slightly, however, I feel it helps the statue feel more primitive and not as machined.

With these adjustments made, the statue is a lot easier to see in the scene, as shown in the image above. After inspecting my scene a bit more, I decided that the altar I had initially chosen did not fit the vision I was going for.  Once again referencing my original concept art of the 3D environment, I wanted the altar to look somewhat manmade in that the altar was carved out of stone. The altar shown above does somewhat fit that description, however, I feel it did not look organic enough and feel cohesive with the rest of the elements in the scene.  Therefore, I decided I needed to replace this altar with one that doesn't necessarily looks exactly looks like how it is depicted in the concept art, but one that better fits the vision I have and is more cohesive with the rest of the scene.

After looking through Quixel Bridge for a bit, I found a megascan model more suitable for the altar. Like the last altar model, this model was a pillar-like stone object, however, unlike the last model which I felt looked too refined to fit the rest of the scene cohesively, this new model's is more jagged and organic, retaining a more natural aesthetic that better implies that it was carved from a big block of stone.

The altar in the concept art was inspired by four-horned altar designs that I had found while doing reference research, and so to more closely mirror that, I found some more pillar-like stone models in Quixel Bridge and positioned them on the altar to give it some 'horns'.

I then added in some foliage and some more moss decals to the scene to provide some more set dressing and vibrance to the scene. I divert slightly from the concept art here, as it did not feature this much foliage, however, I feel the placement of this foliage help enhance the narrative of this statue being some sort of supernatural object. On the base floor around the first platform, I use vibrant and lush ferns for foliage, which have grown due to the light shining through the hole in the ceiling. As we get closer to the statue, however, I instead use dead weeds that resemble withered versions of the ferns. The implication of this is that the presence of the statue has caused the plant life around it to start decaying, helping convey the supernatural and sinister nature of the statue.

Shown below is a series of renders of the final diorama.

I have also put together a video showcasing and transitioning between the different renders.