Self Initiated Project Evaluation

Research

I felt the research for this project could perhaps have been expanded to help me produce a more informed and uniform art style. It seemed at the start I had this vision of a very harsh shaded world which really gave the impression of folded paper, but what I have ended up with is a more smoother 'child like' style lending itself more towards a Wii game instead of something of more artistic merit.

I believe this issue has come through using Unity3D as my game engine and restricting myself to mobile development, these two factors combine to make it much harder to produce high quality/semi-realistic rendering techniques. I have tried later on in the project to rectify this by rendering all of the assets in Cinema 4D using custom made material shaders which can better replicate the look of thick paper on the models (it also helped improve the antialiasing on the models and consistency of the lighting setup).

I felt it was quite a blurred line in terms of polycounts for objects (due to the variety of objects, workflows for polygon reduction and quantity and quality of assets) and also the ordering in which things were produced. Working solo on a project can really test you to see if you can stick to a 'regimented' way of working and many things can affect this process. I found if I had a particularly intensive week and fell behind on my schedule, I'd be doubling up some weeks and on others I'd advance other parts of the project that interested me more at the time messing up the schedule for the later weeks. 

It might have been worth exploring UDK at the start of the project and doing some comparisons between the two game engines to see which would benefit me more visually as the end result for the project was not a game, it was a visual representation of the artwork for the game.

Character

I also believe like the art style, the cat needed a lot more research and development. I think through the project I realised I was much more interested in the design and visual appearance of the environment that surrounded the character than the character itself. I seemed to neglect the work I had to do on the cat until I HAD to do it, perhaps it was past experiences with animation and rigging characters that put me off. I found it quite hard to source any usable footage of cats walking in profile and frontal perspectives, or for that matter any movements (such as jumping, idleness etc)... which is surprising considering the internet is inundated with cat videos.

I felt the character needed to be defined more before I began designing him, I should have built up a back story which had solid foundations from the games core mechanics and then iterated concepts before jumping in with the modelling and texturing process.

I am glad that I did not settle with the first 3D iteration of the cat model, after review it became painfully obvious the cat looked very un-natural, thin and rigid. All these things came from other people reviewing my work, myself and through trying to animate the thing which was like animating 4 cocktail sticks to move like legs.

Scale of project

I'm quite happy that I managed to achieve so much on my own. To produce an entire world including landscape, character, the sky, the atmosphere, the foliage, structures and all the smaller details within the time frame alongside my other modules has proven to myself that I am capable of more than I thought. 

I think at the start I was quite naive in my approach, it is hard at the start of a project to really know how much there will be to do and especially to try and plan an entire years work is no easy task. The problem is everything is based on estimations as this isn't something I have done before, so if one section of the project was underestimated it creates a back log of work which can lead to becoming enthused with the project.

I believe one of the core things I will definitely take through to new projects from this one is having a good workflow. It took me a month or so to really get a good 'flow' for my work method, understanding how I would get from concept to implementation and maintain a consistent quality was a real worry for me at the start.

After much trial and error, and experimenting with software that either sped up or hindered the process I managed to get something that worked for me. Here is a simplified diagram of the process...


Workflow

Asset List > Drawing/Concept > 3D Model > Unity > Block colour >Optimise 3D Model > Texture > iterations L/M > Render Book

It has also made obvious why so many studios have different pipelines for the asset production. Each game is so unique that it requires a tailor made process to produce the assets to a standard that looks right. Not to forget the advances being made in technology year on year, as studios embrace and tailor new softwares to suit their needs. 


Coding

I had hoped I wouldn't have to get involved much with coding as it isn't really a big part of the criteria for this project, but it was unavoidable. Designing assets for a game required me to implement the objects in to the actual game world and view/use them from a players perspective. I created a simple character controller and through learning Mechanim (Unity's new animation module) I hacked a movable character together to be used to test things. I later implemented a few actions the player could do, such as chopping a tree down and mining ore. The final piece of code required was to allow scene switching, this was important to get a feel for the change over between spaces and to see how (and if) it worked.

Book

I think the end result of my project was a little rushed. I hadn't taken in to account render times (which were 30 minutes per object), composing the book together and ultimately getting it printed. I had been quoted 14 working days for an A3 high quality book, by this time I had two weeks remaining on the project and still had to render and compose half the book. I have for this reason fell short on what I hoped I could put inside the book. Some of these things are not necessarily stated in the project plan, for example I wanted to make 3 iterations of each model inside the book (using an appendix to show the low/medium poly models, linked through alphabetical/numerical order). I also wanted to include more renders from Cinema 4D of the landscape and interior scene.

I have purposefully excluded any GUI from the book as I felt it would break the layout and flow of the book, but more importantly I didn't feel the GUI elements stood out enough to warrant its own pages within the book itself. I had hoped I could include the texture maps for all of the 3D elements too, but again it didn't suit what I already had within the book.

Summary

Ultimately it has been a great experience, to be given the opportunity to develop myself at my own pace with a project set by me is something I haven't experienced before. I learned an awful lot from trial and error, and I believe that is one of the most effective ways to learn and remember processes and software specific tasks. What I have come out with is something (when printed) I can take to future interviews and feel confident in the quality and quantity of my work.

 

Digital Rendition of 8th Square Book

 

 

Scene Changing

Scene changing was one of my last parts to this project, originally I wanted the inside of the buildings to be immediately accessed through the door (and remaining inside the game world, not a seperate space) but this soon became impractical when put with the current control system, camera logic and size of structures (and working out how to make the camera cope with third-person perspective in such a tight spaces was impossible!). 

I decided to go for a Pokémon-esque room enter process, a player would walk up to an open door and instantly be transitioned to the inside of a building. I had the idea that the inside of the room should feel completely isolated from everything else and felt the hard-edged interior should would benefit from a simple fading black backdrop.

This was only an example, I was running out of time really especially with the plans to get a book prepared for the hand-in date. I was able strip out the town hall's exterior shell as a base, ensuring I was using the exact same space as the game showed. I had to produce some simple collision boxes so that players couldn't just walk out the front of the structure and in to the fading backdrop.

The camera switches to an isometric camera which keeps the player locked to the centre of the screen, I think this works really well particularly with the control setup currently. 

Below is an in-game screenshot of the interior of the town hall (prototype), and following that is some simple renders of the interior assets.  

Book Development - Process and Problems

I have just finished the process of producing the artwork for my book. This post will explain the process I have taken to produce this work, and also the issues I have faced in the development process. 

I begun by looking through an array of books to find a style I felt would suit the simplicity and clean design of my low poly artwork, I looked at cookery books (specifically NOMA) and an arrangement of game design concept books to see how they laid out the work (and in what order).

I began to take some models from the Unity project and import them in to Cinema 4D R14 to render to print quality, I also developed a procedural texture that would make my models appear more 'believable' having a none-repeating paper applied over them. 

The texture is made up of both seamless textures I have developed in Photoshop and built-in modifiable Noise shaders stacked up using the Multiply filter to produce different scales (and shades) of grain over the texture. I also applied an Environment shader which helps bring out the model from the stark white backdrop I planned to render the images on. Finally a normal map was applied to catch fine details on the flat surfaces (to give the shape more detail at such high resolution rendering, and to give a more realistic feel to the heavy paper).

I then began rendering the models at no particular scale to a 'paper' backdrop, but after a while i realised that it became very repetitive to see the exact same textured backdrop in each render. Here is an example of the initial renders for the book, and also the layout of the text on the page.

  
I decided it didn't look professional enough, and at this point I had not decided what dimensions my book would be so anything I created and rendered might well have to be re-done later on to fit the books dimensions. I developed a cleaner rendering style which brought out the models, and by setting up a grid for a consistent composition (different for left and right pages) and deciding on an A3 portrait setup for the book.

Here is an example of the new studio I have set up in Cinema 4D and the guides I have setup in Photoshop to keep consistency through the pages. 

 

I will save the final images to be shown in the actual book as I feel they will not be shown off here (with all the white space around!).

AO Unlit Trees

After experimenting with lots of different ways to light the trees, be it basic diffuse maps, vertex-lit or lightmapped - they never seemed to fit in with the surrounding environment. I finally came up with a solution to this problem, I sampled lots of plants from photographs of Arizona to get a sample of greens/reds that are naturally found in a landscape similar in tone to my own.
I then brought each tree model in to Cinema 4D and baked on the lighting to achieve a soft AO finish to each of the models. It seems odd that I am baking the shadows in to the trees, but  for them to sit right in the environment they need to use an unlit shader (which still allows to cast shadows as if it were a normal diffuse tree). Baking the shadow directly on to the texture and using an unlit shader makes for the quickest render pass for the object, which is perfect considering how many are used to fill up the level. 

Here is a selection of screenshots which show the trees casting shadows and having AO and the texture baked on from Cinema 4D. 

 

Skybox Surround

The initial idea for the skybox was to have it composed of multiple layers to give a slight parallax effect when the player traversed the environment, it worked well when the images were applied but because of the sheer scale of the skybox the parallax was hard to notice.

It also seemed to ramp up processing on the devices I tested, this was due to how materials behave within Unity. I was using a transparent cutout shader on the single skybox mesh, but when the game was run the engine had to render a transparent texture on top of another transparent texture exponentially. Not only was it rendering each layer, but it was having to re-render the back drop 16 times and frame. The skydome was also divided up in to seperate material IDs which allowed me to apply 5 4096x1024 textures (another bad idea!).

I believe there is place for a skydome like this, but not for a mobile game with current technology.

Here is an example of how the multi-tiered dome worked :  

After working on the graphics for the surround I managed to get something that worked well and blended nicely with the foreground elements. Here is a visual progression of the backdrop, I found it helped to bring both foreground objects gently in to the backdrop to help blur the seam and viceversa.

After producing a final backdrop I needed to optimise the skybox itself to run on iOS and Android devices. The most effective way to do this and maintain a good quality texture was to merge the layers together and have a single cylindrical capped skybox. The second problem which arose was iOS devices up to iPad2 can only manage 1024x1024 textures and therefore my 4096x1024 wouldn't work. I got around this problem by dividing the cylinder in to 4 seperate material IDs and dividing the large backdrop texture in to four 1024x1024 textures. This worked perfectly, the only drawback is now editing them takes four times aslong (to ensure they match up, and they are in four different documents).