KNB137 AS2 Real-Time Project

Planning

Moodboards

The setting for my environment is a temple residing in an East-Asian inspired fantasy world, with the idol being a sacred artifact that is worhsipped by the common people. While gathering reference images and inspiration, I looked into pop media with environments that are inspired by ancient East-Asia such as Genshin Impact, Avatar the Last Airbender and Grandmaster of Demonic Cultivation to identify the textures that were featured and general atmosphere that were created. Through my research, I found that archaic constructing materials such as (lime)stone, wood and the natural mountainscape were commonly used to build these environments as the natural landscapes were often incorporated into the infrastructure. I found that the lighting also tend to be quite dark with dim lights due to the limited lighting from a closed off temple and to purposefully create a dramatic mood.

For the characters, I wanted to base my models on the three wise monkeys, an iconic set of caricatures that has its roots from East-Asian culture. I thought this concept would be perfect for my story context since they are guardians that reside in the temple to protect the idol from evil. Since there are three monkeys representing each concept (see no evil, hear no evil, speak no evil), I will make three models and differentiate them by creating unique textures that reflect the idea each monkey represent. I also wanted to incorporate a bit of the cloud design into the details from the previous assessment, but whether I incorporate it or not will depend on the amount of time I have.

After doing a general research into ancient East-Asia architecture, I did a more in-depth research into the temple interiors, cave environments, as well as common structures and items that appear in temples. For the assets, I wanted to include a set of worshipping items such as an incense burner, cups and seats to convey the ritualistic traditions surrounding the artifact. I also thought of incorporating high pillars with heavily detailed carvings and mythical guardian statues (as seen in the asset moodboard) to tie the Chinese temple look together. However, since these assets could potentially detract from the focal subject, I will find alternatives with potentially less detail to ensure the idol still remains as the centre of attention.

For the temple interior, I found that my envionment references often had elevated ceilings that highlight the grandness of the place. The temples also had an overall rounded or square structure that brings focus to the centre piece, with elevated elements such as platforms and stairs in the background bringing variety to the ground structure and section off main areas. I will use these same elements to carefully curate a composition that will emphasise the idol in the scene. The dramatic nature of the scene is reinforced by the limited light sources, such as lanterns, fires or natural sunlight that peek through from the opening in the roof. To emulate the same dramatic atmosphere, I'm planning to use through an opening in the ceiling as my main light source, with dim lantern lights littered across the scene to light up smaller areas.

Sketching

For the shape of the temple, I was inspired by the bagua map (eight trigrams), a Chinese symbol that represent the fundamental forces of reality. While I mostly used this shape because I thought the structure would be interesting and made it easy to make duplicable wall modules, I also found that the ideas of divinity associated with bagua was relevant to the heavenly nature of the artifact.

Initially, I used a basic circle structure for the temple, but later changed it to the octagon shape of the bagua. I also played around with the placement of the long pillars that extend to the ceilings; while they were initially connected to the outer walls to section off different wall modules, I later moved them towards the inner circular area to section off the centre area and prevent it from looking too empty.

Extra assets such as guardian statues and incense burner were placed at the entrance to highlight the importance of the idol and indicate that the temple was used as a place of worship. The elevated platform and stairs were also added later to make the artifact seem more sacred and the centre altar more distinct from the ground. Around the scene are lanterns that surround the circular centre to help recreate the dim lighting seen in the reference images.

Modelling

Character

Using the 4 to 2 topology method, I started connecting all the extra hanging edge loops to the neck. Here, I had to be strategic about where I place cuts in the topology as not all of the gaps come in pairs (as seen in first image) which is the most convenient shape for this simplification method. Before doing this, I first deleted as many edge loops from the head as I can to make sure the number of loops matches up with the neck. It is important that I delete the edge loops that aren't connected or don't define the face as to not break the topology.

In particular, I struggled with this one extra gap that was left hanging as the simplification method usually requires two or more gaps to work. To resolve this, I initially attempted to connect it to the neighbouring 4 to 2 topology in order to connect the loops. Realising that this would create even more edge loops in the topology, I decided to take another approach and delete the edge loop causing the singular gap instead, as it wasn't one that defined the face. Then, on the other side, I connected  merged the edges together to form the 4 to 2 topology again. The uneven edge loops made the simplification process challenging as it required me to be strategic about the cuts and merges I make to reduce the amount of new edge loops that would be created. From this, I learnt that the most efficient method of simplification was to subtract as much unnecessary topolgy as I can before adding and connecting any new edges and repeat this process until the topology is at its simplest form.

Here, I encountered an issue where instead of the idol having too many edge loops, it was the neck instead. To solve this, I first target wielded the vertices together, then divided the two rectangles into three triangles instead to ensure that all the vertices are connected. This way, no n-gons would be created.

After merging all of the edge loops together, I then smoothed out the seams of the neck using the smooth brush from the sculpting tool set. This is to ensure that the neck is rid of the distorted and flat surfaces caused by connecting the head and neck together. When the neck has a smooth curvature, I moved onto the tidying up the topology and looking for any places that could be simplified further.

To simplify the topology further, I added edge loops in between the 4 to 2 topologies and target wielded vertices together to reduce the number of horizontal edge loops. I was surprised by how simple the topology was able to become after using this method. There were also certain areas where no new edge loops were required to target wield and reduce the edge loops (see last two images above). From this I learnt various simplification methods and continued to carry on the subtracting before adding philosophy mentioned earlier.

With the other 4 to 2 topologies, I continued using the simplification methods I have learnt to make the topology more simple. This involved target wielding extra vertices together as well as creating and deleting edges with the multi-cut tool to figure out the best way to connect existing vertices. I found that the process became a lot easier as I repeated it, and I became more intuitive about where to wield and where to cut to get the most efficient topology.

Lastly, to clean up the mesh, I straightened out the edge loops of the neck as much as I could before moving onto the rigging. The topology simplification process had a bit of a learning curve for me  initially, but it became easier to do as I repeated it. Through this process, I was able to gain knowledge on how to strategically add and subtract from the topology using modelling tools in order to get the most simplified topology.

UV Mapping

When unwrapping the UV of my character, I first selected all of the mesh and used the automatic projection to get the shells. However, realising that I could use the unwrapped UV I had done for the idol previously, I reverted this change and used automatic projection on the neck and the body only instead.

To unwrap the legs, I started by cutting down the side of the legs and separating the front and back. I wanted to hide the seam in a less visible area so that the texture isn't cut off abruptly and to make sure the texture wraps evenly around each leg. To do this, I placed a cut in the middle of the inside of the legs as well as down the middle of the crotch, then separated the four shells. Then, I flipped and sewed the corresponding UV shells to the legs and unwrapped them.  For the body, I cut down the middle of the shoulder and torso, separating the front and back from each other. As with the arms and hands, they were cut and separated from the main body as it would stretch out the pixels.

Using the same process as the leg unwrapping, I cut a line down the middle of the right arm continuing the middle line from the shoulder and unwrapped it. However, realising that the seam would be visible at the top, I sewed the cut back and instead continued the cut from the side of the body as it would be in a less visible area. After learning where to cut the UV, I repeated the same process for the left arm and unwrapped the UV successfully.

The hand was more challenging to unwrap as it experienced more pixel stretching issues. I started by separated the thumb UV from the hand and cutting a seam that follows the cut from the arm before unfolding it. With the hand, I cut along the side and unfolded it. However, I was uncertain about whether this was the best way to unwrap the UV and later experimented with different ways of cutting and unfolding the hand and thumb UVs to find the best layout.

With the feet UV, I also experimented with separating the UVs into different parts during the unwrapping process. I started by cutting out the sole of the feet and cutting the back of the feet to unwrap the main UV. However, since there were still pixel stretching issues (as well as the UV shape making it difficult for compacting), I decided to separate the UV even further into three parts: the sole, the toes and the heel. My concern at this point would be that the texture would not wrap around the feet well as there was a cut on the sides of the feet. Thus, I ended up separating the top of the feet, the sole and the sides into three UVs. While I was unsure if this final outcome was the best way to cut the UV, the pixels did stretch evenly across the final shapes.

Here, I experimented with different ways of cutting the hand seam to find the best way to unwrap the UV. I first reverted the UV for both the hand and thumb, unfolded the uncut UV and determined that the side of the hand and thumb must be cut to relieve the stretching at the tip of the fingers. However, instead of placing cuts unknowingly, I decided to cut along the sides of the hand and stopped the cut at the thumb to ensure the texture would wrap around the more visible/outer areas of the hand better. Continuing the cut on the side of the hand, I also placed a cut along the inner side of the thumb as it would be less visible. This process was quite challenging as I had to place cuts and unfold more strategically to make sure the texture will be wrapped around nicely in the more visible areas while also making sure the pixels won't stretch.

While trying to clean up the UV and look for any mistakes, I found that there were a few edges that I was unable to sew together despite it seeming like it was already connected. I realised that this may be due to me not merging the edges together properly during the modelling process, which meant two separate edges still remain at these cut areas. To solve this, I selected said area and used the merge tool to make the edges close to each other connected.

After unwrapping and cleaning up all of the UVs, I compacted them all into one UV shell, the size of the UVs determined by their importance (whether they were the focal point) and proportional sizes. This UV unwrapping process was more difficult than before as I had to be strategic about where I cut and stitched different UVs together to ensure the textures will be wrapped around smoothly in the more visible areas. Despite the few bumps I ran into, I was able to learn how to identify the best areas to cut and unfold to get the most effective UVs through this process.

Rigging

Skeletal Mesh

With the skeletal mesh, I used the quick rig tool and moved the individual reference points to match up with the model. I also used the mirroring tool to ensure that both sides of the rig are the same. The process was straightforward as I didn't run into any technical issues. I only needed to make small adjustments for the elbow rig to ensure it bended the model properly since the mesh is a lot thinner in that area.

Skin Weights

With the skin weights, there were also no major problems with the painting process. The main steps involved flooding certain areas with 1 to ensure the joint had full control of it (e.g. flooding the whole hand at the arm, flooding the top of the head at the neck) and erasing the joint areas with less value to ensure that the surface is not distorted during animation. With the latter, I had to create a gradient with the brush to ensure that the way the surfaces bend around the torso is not stiff and stretch weirdly.

While the rest of the skin weight painting process was fairly straightforward, I ran into a bit of trouble at the shoulder as I noticed the side of the chest was bending weirdly when the model was imported into Maximo for animation. Here, I had to erase the skin weight at the shoulder using the same method as I did the torso to make sure that the arm and shoulder surface doesn't seem stiff when bent. I was unsure if my final result was the most effective, but the bending at the side of the chest looked less obtrusive than before.

Texturing

For the textures, I started off by using a wood base since I wanted the monkey guardians to look like statues that came to life, and adjusted the colour based on my reference image. Then, using a brush, I added the white areas around the face and stomach and airbrushed the edges to make it look more blended in and make the brush strokes less obvious. Throughout the process, I also made sure to use the mirror tool to make both sides the same.

Afterwards, I added details on the face, hand and feet to make the features more defined. With the hand and feet, I first added a dark grey to separate them from the body and imitate the monkey caricatures I found in my references. Then, I added black lines to define the fingers and toes and adjusted the height value of the brush to make it look embossed. Finally, I airbrushed a darker grey at the tip of the hands to add depth. While not pictured here, I also added ears using the dark grey and white on the sides of the head to make it more obvious that the character is meant to resemble a monkey. Lastly, I added the facial features using a similar red to the idol and painted it with increased height value to make them look like they were painted on.

To add more unique textures, I added in short strokes of lines around the neck, arm and top of the face to create the illusion of fur. Here, I also used a brush with increased height value to make them look embossed. After finishing these fur texture, I erased the edges using an airbrush to make them look like a gradient that fades in, and to make it look less like the fur texture abruptly stops in the middle. Additionally, I also erased the edges of the white on the face and stomach using the short fur brushstrokes to tie the design together.

As the last bit of details, I decided to use the same gold-like colour to paint over the lines on the hand and feet, as it gives better contrast against the dark grey and creates a more coheisve colour palette with the rest of the textures.

The texturing process was pretty straightfoward as my previous knowledge and skills helped me achieve the effects that I wanted. My background in illustration was also helpful in helping me decide the colours since I understand how to combine certain colours to make the palette more cohesive or contrast against each other. Ultimately, I decided against creating different textures for each of the monkey guardians as it would be too time consuming. Despite this, I was still very happy with the final outcome.

Environment

Blocking

Before building the environment in Unreal, it is important to plan and block out the environment first toidentify what modules can be duplicated and make the assembling process more efficient. This is also to ensure the proportions are correct and make it easier to line up objects according to the plan I sketched out. Using Maya, I began to model the environment using simple geometry as stand in for different assets.

The first step to creating the environment was creating the base shape of the floor, as my environment has an unique floor shape. First, I created a square plane and divided it equally using edge loops. Then, to get the right number of edges needed for an octagon, I wielded the edges of the corners together. I was unsure why the octagon shape looked off, and realised after looking at the reference images that the proportions were off. To finish off the base shape modelling, I made the middle areas wider by scaling the edges outwards.

Next, I added in a circular plane in the middle of the floor to make the central platform for the altar to be placed on. To make the walls, I added inner edge loops so that they can be extruded. After extruding out the walls, I deleted the inner edge loops at the bottom of the floor as they were unnecessary. For the ceiling, I extended the edges from the top of the walls and connected the vertices together. However, this meant having to manually adjust the vertices to make sure the topology lined up with the floor. If I were able to redo this model, I would instead duplicate the floor for the ceiling and connect the edges together to form the wall as it would ensure that the topology is all mirrored.

After creating the room, I moved onto making the environments assets and adjustments to the layout. I added the stand-ins and the incense burner and and a set of guardian statues at the entrance to make it look like a place of worship and indicate that the idol is heavily protected. With the pillars, I experimented with their positions to see what would make the best layout. I tried moving them to the wall to act as separators between the wall modules, but later determined that they would look better closer as individual pillars standing in the middle rather than being attached to the walls as it acted as a good way to separate the central platform from the outer area. 

Next, I added an extra platform and ramps to elevate the altar further. Realising that the room would be too small for the camera to capture renders in later and for me to add extra assets in, I decided to also expand the rooms size. Finally, I added in some partitions connecting the pillars together to close off the central area even more and added stairs all around the platform to make the altar even more elevated. This is to make sure that the idol remains the focal point in the environment since I noticed it would be too close to the guardian statues in terms of height with the previous platform set up.

As I wanted lighting to fall through the ceiling and hit the central platform to highlight the idol, I decided to make an octagon shaped hole in the roof to make it seem like it's part of the temple structure, rather than the natural cave environment from the previous assessment. To make the hole, I used the multi-cut tool to make cuts and ensured that edges are all straight. Then, I simply deleted the surfaces in the middle. Later, I also decided to delete the partition that closes off the central platform from the outer area as it made the composition look too cluttered.

After completing the general layout of the environment, I decided to cut out the parts of the environment that can be duplicated to increase the effiency of my workflow by employing modular kits. To do this, I first started by expanding the size of the room according to tutor feedback to ensure there will be enough space for the camera to render the composition later in Unreal Engine. Then, I rescaled the room to ensure it will be proportional to the character and idol model later in Unreal Engine.

When the rescaling process has been completed, I started cutting out the modules that can be repeated. Initially, I was going to cut the room  in four quarters in a cross shape vertically and horizontally. However, realising that this would but decided to do it diagonally to ensure that the assets will also be duplicated in a set. After cutting the module into a quarter, I merged everything into one mesh and deleted the type by history to ensure everything will be transformed consistently. Then, after making sure it's sitting in the world centre, I applied freeze transformation. This makes it easier for importing as the guide will automatically be in the centre of the scene and be proportional. I also had to move pivot point to the corner of the environment guide in Maya first before importing, as it would cause problems with the position it's imported in in Unreal.

Unreal Engine

After importing the character model and the environment guide, I placed the character into the scene and rescaled the guide to ensure that the proportions are to scale. Then, I started placing in Quixel Bridge assets for the extra props. For the long pillars, I duplicated the stone pillar asset I found on Quixel and attached it to the ends of each one vertically. While I would've liked an asset that didn't have extra surfaces sticking out at the bottom, this was the closest one I could find to the pillars I wanted. Although I wasn't able to make this change here, I would have rotated some of the pillars to make the duplication less obvious since I only duplicated them and attached them to the ends of each other.

For the stairs, I created it by duplicating a stair and stair wall asset and assembling them to make one big stair. However, I realised that because of the way the original asset is, one side of the asset is completely see through. To resolve this, I used the mirror tool to make sure the side with the visible textures are facing the outside. Realising that the inner area of the stairs would also be see through, I decided to duplicate and mirror the walls away to ensure no empty surfaces remain.

To make the stair into a duplicable module, I grouped the selection and used the merge actor tool to make it into one mesh. By making the module into a merged actor, it made it easier to duplicate and rescale the size and width of the whole stair, since each individual item would be transformed separately otherwise. Then, I dragged the new instances of the merged actor into the scene and rescaled it to fit the central platform.

After looking at the platform layout, I later decided to remove the four diagonal stairs as it was clustering the platform too much and made it easier to spot that the stairs are replicated assets. I found that during the assembling stage of environment building, I relied more on my intuition in composition to determine how the layout should look as it was easier for me to understand how things should be laid out once I actually see it.  This also compensated for the poor planning I did for the environment, since I only had a rough idea in mind and didn't think too much about the type of textures and additional assets I wanted outside of the key ones I had already drawn in my initial sketches (pillar, incense burner, guardian statue etc).

Following the same modular making process, I decided to make the trigram pattern found on the Bagua symbol as patterning for the floor using a stone pillar asset. I did this by duplicating  the pillar and attaching them to the ends first. This time, I decided to rotate the pillar around so that the broken end of the pillar face each other. Then, using this two pillar module, I formed the horizontal line. With the diagonal line, I used one module, rotated it, then mirrored it on the other side to ensure that the pattern is equal. After making one trigram, I merged the mesh into a merged actor and duplicated more trigrams that way.

To create more variations in the textures and modules around the set, I added an upside lantern to imitate ceiling lighting. For the walls, I added a horizontal carving strip and a tree ornament carving in the centre. These details made the set less plain and more like an intricately built temple.

Using the geometry actors in Unreal, I built the walls of the sector following the guide as close as possible to ensure that the sector module would all match up later. Since my environment has a specific architecture and isn't a natural organic environment, I had to be especially careful about making sure my modules will line up with each other. With the ceiling module, I used the cube brush to create different planes and made a static mesh.

To make the different sections of the ground more distinct, I decided to use the plywood texture from earlier on the trigram to make it stand out from the scene. I also applied the same texture to the altar that the idol sits on to make it draw more attention in the scene. Additionally, I decided to add a round ground section with marble textures in the centre to make the platform stand out more and separate the inner area from the outer. This idea was inspired by the bagua pattern in my references, as it had a round circle in the centre of the trigrams. The marble texture I chose specifically uses a bright teal and muddy green colour as the bright value gives good contrast against the platform and is also cohesive with the existing colour palette I had already created.

When duplicating the modules for the environment, I discovered that the gaps didn't quite line up properly. The trigram pattern in particular had noticeable clashes due to the way they were positioned. I was unsure if I had made a mistake while making the environment module despite following the guide I had made. Thus, I decided to go back and try to remake the module to see if there's a better way of constructing them to make them line up properly.

With the remake, I tried making two modules instead of the three in the original guide. Using the geometry actors, I used a square plane and used the subtractive setting to cut out the pizza shape. While the wall module creation didn't have any problems, I realised that I couldn't use this method to create the ceiling since it would have gaps in between the modules. I decided to instead make the ceiling with one singular plane and use the cylinder geometry to cut out the shape in the middle. Before I did this, I duplicated the wall modules I'd made to see if it was actually better than the original modules I had.

When making these wall modules, I realised that the horizontal wood carving looked weird being cut off on an empty wall next to it. I took this chance to create a new wall module without the wooden ornament to add variation in the wall textures, making it look more cohesive. After duplicating my modules and forming the octagon shape, I realised that the end result ended up being uneven. This also applies to the trigram pattern on the ground. While I was unsure why things didn't line up properly, I went back to making module according to original guide since it seemed more reliable than the new modules I made & I was running under a time constraint at this point.

Lastly, to introduce some asymmetry to the scene, I added some fallen debris using rock and stone pillar assets littered around the scene. This made the temple seem more weathered due to age and indicate that it isn't at its peak condition because of natural environment wearing it down over time. I found that my improvisation skills and ability to adapt to changing circumstances was a great aid in helping me solve issues that came up in the module building process. While I was able to avoid some of these issues, I felt that I would've benefited from a more constructed planning to ensure that no similar issue will come up in the future when I make environment modules.

Lighting

For the lighting, I wanted to follow a similar colour palette to the last assessment and used a yellow and green palette. As I wanted to make the temple look like it's sitting in a mountainous and foresty area, I made the sky light slightly green to indicate the greenery being reflected in the light. I also wanted to add a bit of volumetric fog to make the light from the ceiling look softer, making the idol look like it's basking in heavenly light. After adjusting these light settings, I was able to come up with the final result pictured on the right. However, I was still unsatisfied with this since it didn't have the mysterious dim lighting that I wanted.

Before making further adjustments to the environment lighting, I added in the orange lighting inside the ceiling lanterns around the scenes to imitate flames. I used orange specifically as I knew that orange and green/teal makes good contrasting colours. The intense orange also adds to the ominous atmosphere that I wanted to create. After finishing the lantern lighting and adjusting the volumetric fog, I made the sky more cloudy. This was because I wanted it to look like a bad omen is coming, hence why the guardians are praying to the idol. Cloudy weather is often used to indicate the worsening of a situation in cinema and environmental storytelling, hence the use of it here. 

After adjusting the clouds, I made slight adjustments to the volumetric fog and lighting contrasts again as needed. I also added light inside the lanterns on the ground at first to make the scene more illuminated. As I want the lantern lighting to be the brightest lights in the scene, I darkened the sky even further, making the atmosphere more ominous and foreboding. However, this also made it a little hard to see the centre altar due to the dimmed sky light. To resolve this, I decided to use post process volume to adjust the colours and contrasts and added extra lights to light up the altar.

When reopening the file again, my scene didn't have the same lighting that I had saved from before. Because of this, I ended up having to readjust the lighting again to get the dim lighting that I want. The result was darker than I had expected but was close enough to the previous version. Following this, I used post process volume to increase the exposure and contrast of the scene, making slight adjustments to the colour temperatures. Then, to light up the altar better, I added a slightly green-ish point light above the platform and increased the radius to make it look more like natural light being flooded in.

Lastly, to make the hole look less geometric and unnatural, I decided to make these stone gate modules to surround the opening in the ceiling. Using the modular creation process I've been employing throughout the environment building, I made a stone gate and duplicated it around each edge of the octagon shape. While the gaps didn't connect properly, it was good enough to create a shadow silhouette. However, I was unsure of how effective these gates were in making the ceiling opening less geometric as it looked similar to how it looked before I added in the stone gates, most likely due to the direction that the lighting is shining in from.

Animation & Camera

When importing the Maximo animation, I ran into a few issues as I couldn't find the original rig used for Maximo. The tutor showed a way around this, which was to import the skeleton of one of the animations and create the animation by dragging them onto characters during camera renders. To prevent the same issues from happening in the future, I will make sure to keep a copy of the character file I used for Maximo to avoid complications. For the character animations, I looped the standing and nervous playing throughout the whole scene as I didn't have time to make any more complicated animation & I didn't know how to make them transition smothly. However, with the praying on ground animation, I had to be strategic about where I place cuts and loops since the character starts from a standing position and ends up a kneeling position, meaning the animation doesn't loop smoothly.

When rendering the sequence shots, I experienced difficulty with using the cameras for each shot as the shots would disappear after exiting to the main sequence. Instead, I had to use one main camera to get the shots as I couldn't figure out how to resolve this technical issue. Hence, I had to be more careful about where I place my keyframes as to ensure the camera doesn't move weirdly when transitioning to the next shot.

By keyframing the focal length and aperture of the camera, I was able to create camera blur that made the environmental shots look like exposition shots introducing the idol and characters. I realised that when rendering these shots that the room could have been expanded to give the camera more space, since the new pillar positions ended up blocking the camera in certain positions.

I was unable to render the final sequence in Unreal Engine for some reason, most likely due to using one main camera for the sequence rather than individual cameras inside the shots. Instead of getting the high quality render, I had to screen record the sequence as an alternative. I also realised too late that I forgot to turn off sRGB on the texture material for the idol when rendering the animation sequence and made sure to turn it off for the final renders.

Final Renders

Through this assessment, I understood the importance of planning when making duplicable modules. While my improvisational skills helped me navigate around technical difficulties, it is still important to plan ahead when making duplicable modules to ensure the workflow is the most efficient it can be. Despite the technical road blocks I ran into, I was still fairly satisfied with the final outcome as it conveyed the narrative I wanted to tell through different elements of environmental storytelling and mise en scene.