Film Room Project: Evaluation

Before coming together as a group, the personal films I would have liked to have worked on included: The Trophy Room from Highlander (1986), Seth Brundles apartment from The Fly (1986) and any of set from Bunrakku (2010) as portrayed from the stills below.

The trophy room – Highlander 1986

Brundles apartment - The fly (1986)

A common set on Bunrakku (2010)

What stood out to me about the Highlander room was how prop heavy it was, each with their own different and unique textures which my group and I could possibly study and showcase within our level. In addition the circular format of the room would allow one to be more immersed in our level and props than a larger emptier room. The idea was to use said room as a literal trophy room of our skills in 3d and engine.

Brundles room in The fly (1986) was chosen for a similar reason. The room at first glance appears messy allowing us to fill our scene with a range of props, each with differing materials.

Moreover I felt that we could use the image marked red as a potential focal point to further develop our modelling skills as well as showcase our pbr skills. In the film the item is seen glowing and emitting smoke. Such a prop also allows the opportunities to study and showcase skill in emissive textures and also particle effects.

In contrast, Bunrakku (2010) was chosen because it offered the opportunity to play with lighting due to the heavily stylised nature of the film. The film itself is made to be a reminiscent of Japanese puppet theatre. Other opportunities offered by Bunrakku included experimenting with particle/ weather effects in engine to convey some of the moods and atmosphere featured in the film. However, the room posed problems as the simplistic nature of the models featured within the scene would have made the level look low poly and would not sell our 3d skills.

My Group and I selected Jay Gatsby’s terrace from The Great Gatsby (presented above). I was initially apprehensive as unlike the others I had not seen the film prior to undertaking this project. Additionally, it did not fall into the genre of films I would usually watch. Having said that they were similarities to the Highlander room. For example, Gatsby’s terrace features many props, each had its own unique textures. It offered the same opportunity as in the Highlander room to showcase our skills in PBR as well as to practice and sharpen our lighting skills just as would in the Bunrakku scene. The Jay Gatsby’s terrace provided a further opportunity to work with particle effects such as candle fire and the chance to learn how to balance different light sources and emissives such as the window, chandelier, candlelight and sunlight. In its own unique way Jay Gatsby’s terrace mirrored the opportunities identified in my previous choices. We envisaged that many of the assets in the room could be copied across and save us time to focus on engine problems.

During the production process I was tasked to model and texture the following:

• An Angel Bust

• A set of candlesticks

• A Circular Window

• Wall mounted lanterns

• Banister

• Tapestries

• Floor tiles

Each model presented its own set of challenges. The Angel bust statue was initially meant to be modelled in max up to a certain point in which it would be taken to Z brush which I would then sculpt over.

What I did not anticipate was the amount of time needed to complete the task and proficient skills in Z brush. I initially thought my confidence and skills in anatomy would enable me to complete the task with ease. Unfortunately I found getting to grips with Z brush’s interface very difficult. It became apparent at my first attempt at sculpting human features that I was not making any progress with Z brush.

My first attempt – I had difficulties sculpturing the human features even on a non-existent animal it with Z brush, illustrating that perhaps Z brush was not the option for me

This led me to use a mix of box and strip based modelling on the Angel asset.

This was not without a challenge. The challenges encountered included making the features more human like and in particularly getting the proportion and perspective right. For example, on my initial take, the face looked too boxy and inorganic and the proportion were out of perspective. To resolve these, I applied the relax tool to the bust and worked with smoothing groups to round off sharper areas. Resolving the bad perspective of areas on the model was addressed by working in 4 viewports: top, front, left and perspective. I anticipated that taking this approach would allow me to see any mistakes I was making first hand and correct them as I did the modelling.

Moreover I deleted one half of the model and modelled under the symmetry modifier to ensure that no asymmetrical problems came from the modelling process. This modifier was used regularly when appropriate to make sure that symmetrical models were symmetrical. This approach helped me with the unwrapping and texturing process.

I took advantage of the symmetry modifiers use of copying an unwrap, by modelling and unwrapping one half or one element of a model and used symmetry to duplicate the unwrap and model to speed up the process. This gave me more UV space for textures. Other methods utilised include overlaying similar Uvs on top of each other and slightly adjusting the verts onto one another to enable the Uvs to match.

This also allowed me to be more efficient with texture space especially when dealing with aspects of the model which should not really have that much variation in texture and could be overlaid each other.

I encountered a number of problems when packing my Uvs. Prior to this project my knowledge and skill relating to packing was limited. I therefore sought tutorial support and spoke to others. The feedback was helpful; my knowledge, understanding and skills in packing greatly improved. I found I could pack my Uvs based on the largest surface taking up the most amount of space for the texture sheet as well as packing and breaking my shapes around each other.

Feedback from tutor

My confidence and skills in unwrapping has improved. These are shown in the new and old UVS presented above. I have gained a good understanding about key concepts rather than my earlier focus only on processes.

Compared to my experience during modelling, the texturing and unwrapping stage came with ease due to my knowledge and understanding of concepts and techniques needed to complete the task. My reflection on modelling is that I had slowed down my workflow by overestimating my skill and knowledge based on the idea that I had been on the on course for a year but I have not actually modelled a years’ worth of assets. This project has pushed me to re-evaluate my skill and search for alternative methods of modelling such as working in 4 viewports to ensure there are no mistakes in my assets.

For textures I used the following programs

• UE4

• Photoshop

• Crazy Bump

• 3DS max

The more I textured the more I realised how efficient and helpful it is to produce textures in Photoshop and preview them in 3ds max. I had found that in several cases due to the nature of the unwrap seams were overt in models.

As seen in the above image, the initial seams were overt. Using the clone brush tool in Photoshop, I resolved this by cloning pixels in the key areas, which made the seam look less obvious.

Additionally, generating textures was a very iterative process. This involved packing, remodelling and re texturing my models. There were times when the textures were too noisy and had to be downscaled in Photoshop via overlaying opacity or adjusting normals by playing with the levels of my height map in Photoshop and converting it into a normal in crazy bump.

The above is an example of the iterative process I undertook with texturing. Some problems encountered included making a texture, applying it onto a model only to find that it did not read as well as I had hoped. On example being the wax texture on my candle was too saturated. This made my asset look stylised.

Looking back the project assisted me to view 3d and 2d as a related process. Working in both 2d and 3d have sharpened my outcome in texturing and modelling. My current textures are stronger than the previous.

In addition, my contributions to the Group included overseeing the administrative tasks such as drawing up a timetable, asset list and a Group blogs to track our progress. To ensure effective management of our work, I set up a Dropbox account with folders and naming conventions in them. The idea was to allow Group member more confident in UE4 to import assets. This enabled effective editing and changes to be made by all members albeit with some difficulties where some members renamed files with incorrectly which had difficulties with the importation of assets.

We utilised each other’s feedback to further develop our work. We helped each other with the iteration process during implementation of assets in the engine. We imported and textured our assets in engine and within the level and find problems that needed resolving and worked together to address these. One example is my Angel asset, this proved to be too big in engine.

This was a common practise across many of my own models as no scale size had been previously established. Looking back it would have helped to have a coherent system in place at the start because we spent too much time focusing on resizing assets rather we could have devoted more time in asset refinement and textures.

In addition, although we delegated tasks among ourselves however I felt that some of work could have been shared more equally. To give one example problems such as off proportional sizes were corrected by the member responsible for the majority of the engine tasks. This problem looking back should have been solved among ourselves as we gave this person more work to deal with, when they already had a lot to work with.

Although we were fairly organised by setting ourselves deadlines, some members failed to meet the deadlines. Others preferred working from home. These created problems relating to importing assets and having group debriefs on progress.

A key learning point for me and something I would like to take forward is to more open and schedule in the time table regular meetings to revisit the initial plan. Another relates to creating a space to share any concerns relating to the project and find solutions on how to proceed.

Above is a still of the final shot compared against our reference. One thing we managed to achieve was a sense of likeness within the scene. I feel that we managed to achieve our goals in conveying a contrasting range of materials in our scene. We a met our other goal which was to balance the many light sources within our scene. One notices how there are no oversaturated lights which may cause visual discomfort within the scene. However the final scene is not without its problems. At first glance one could argue we managed to achieve a likeness but it is not a mirror copy. Unlike the film scene itself our production is much thicker in dimension and several of the assets are either too small compared to the reference of have slight differences to them such as my angel assets wings peaking outwards. These issues were a result of not spending enough time working out measurements prior to modelling. This caused differences in sizes and proportion which additionally took up time that could have been spent refining lighting and the colour grading of the final scene. On inspection one notices that the film still features a much warmer palette compared to our production.

Although not as accurate to the still, such a quick job improved the likeness with such ease illustrating if we had more time we could have refined the colour grading filter and the overall presentation of our scene.

Team Manifesto:

This still was chosen as it is aesthetically pleasing and offered many opportunities to showcase our skill in PBR. It has many assets including unique material properties. It offered the opportunities to show our skills in lighting as well as sharpen our knowledge and skills on how to balance the different lighting sources. Particularly it would challenge us as a Group not only to learn how to recreate these different light sources but also balance and their properties in the strength of light emitted.

We also took the view that many of the assets could be duplicated giving us more time to focus on PBR, lighting and other stretch goals including flame particle effects and animated curtains. Although these stretch goals were not fully met, the fame particle effect was replaced with a flipbook texture to ensure the scene itself was efficient. Additionally the animated curtains was not pursued because that the windows in our scene were closed which meant no wind would be blowing.

In terms of planning we created an asset list with a tri budget delegated to each assets at the start of the project. To ensure that we were all on task we split up roles among ourselves and took advantage of Google Doc’s live update feature to make sure all changes to the tri budget say members going over was notified in due time, discussed and acted upon.

Asset list for this project

Additionally we created a timetable on what we would do each week. However this did not go as planned. This mainly related to communication and preference place of work. Some members of the group preferred working at home. This caused problems on giving and receiving peer feedback on issues found with assets. Another factor related to the varying skills we had in 3D modelling meaning some of us were simply slower modellers.

On reflection in order to have prevented this we should have made more effort to create an atmosphere which encouraged helping each other from falling behind.