In my last post I explained what I learnt during the process of blocking out the Tokyo street scene. I invested quite a bit of time and hard work into this and most of it proved extremely valuable for the project. I hope that this blog will serve as a reminder to myself of the steps I took to get to the end result, as well as provide some insight for anyone interested in my process.

Lighting

Considerable time was spent attempting to figure out the right approach for lighting. On previous personal projects I worked with a baked solution in Unreal. Despite good results, I often wasted a lot of time while the lighting was building. This slowed down my progress and meant I couldn't be as flexible with updates and changes. I researched into LPV's which Epic used on Fortnite. I watched a great GDC video Here which sold me on the technique and with the help of the documentation got it setup. For the Blockouts this gave me decent results. I was able to achieve accurate bounce light and was confident some of the resolution issues would be less noticeable with final art. 

Find more here: LPV UE4 Docs

During my time on the CGMA lighting course I discovered the disadvantages of the LPV system. With the systems current implementation, it only works with the Skylight, Directional and Emissive materials. Not having access to point and spot lights meant it was a lot more challenging to light interior scenes. After further research it was at this point I switched to NVIDIA VXGI solution which used voxels to drive the bounce lighting. I used this for the course and my upcoming project, so I will do more of a breakdown about this later. 

NVIDIA VXGI

Materials 

To create a framework for the materials in the scene I started a mini library. These were all created in Substance Designer and to a first pass level. The idea here was to get the main materials in the correct folder structure and get a consistent art style defined early.  With a good structure replacing them later would be easy.

I pushed for a more stylised look for the art style as my main source of inspiration was the overwatch short. 

I created a trim sheet which I used to texture most of the street and pavements in the scene. Plus, a decal atlas which covers the Japanese signage.  

Meshes

For the blockouts I normally start with the largest brush strokes. First, I create a blockout in 3ds Max, modelling the whole layout in one file allows me to figure everything out and establish a good sense of scale. Once the idea grows i separate the individual elements and establish either modular pieces or unique assets. This isn't always perfect straight away but more of an organic process that grows over time.    

Full blockout street scene

Further breakdowns for the individual assets

Style Asset test

Post blockout, I pushed ahead with finding a style in the unique assets. I wanted to find a balance of exaggerated realism with somewhat simplified details, exaggerated silhouette and realistic materials.           

This started with the post box asset. For art example assets I normally try to find something with multiple techniques and materials to try and encapsulate and answer as many questions as possible. Here I had sculpted worn paint, metal and stone. For stone I was inspired by Michael Vicente (Orb) Portfolio style with slightly more realistic micro noise. The metal was mainly about subtle edge damages and stylised warps. For the paint I attempting to reveal the metal surface underneath. for the Silhouette and mesh itself it was about finding the balance of how far to push reality.

UE4 Shaders and Tech

As I have become more familiar with UE4 I have adapted my approach to environment creation somewhat. I now make simple set dressing assets in blueprints which helps to easily replace and add logic to asset prefabs later. (more of this in upcoming blogs) 

Shader wise i am focusing on master shaders to encompass parameters for flexibility later. The most bespoke work I did was parallax interiors. I built small rooms from blockout assets. I then capture an HDR using a scene capture actor, this is driven in the Shader to create a fake room. These are blurred and provide enough depth along with set dressing to give the look for rooms in buildings and shop interiors for the various shops. The beauty of this is as the assets get to final art quality, a simple recapture improves the HDR rooms.  

Small rooms created for the interior windows

Designing the Robot

I did some initial design work at the start of the project to figure out how I planned to handle the robot. I made a series of kits of arms, legs, bodies and heads and kit bashed them in 3d to generate the designs. Had a lot of fun with this but I plan to redesign this guy with more detail later.

  Kit bashed designs for Robot

Final design choices. In the end I chose design 3/6

Next Post

In the next post I will be covering the switch to the new project idea. Showing some more process and blockout work towards my new idea. 

I hope you enjoyed reading about my process and if you have any questions or require further information let me know in the comment section below.

After the first blockout, I still wasn't satisfied with the layouts. I continued with another composition, which was loosely based around the first blockout from the previous blog post. Visually this composition was my preference but it lacked depth. To counter this I messed with having a strong fore and background where the robot was the main focus. The background trailed off to the city behind.

At this stage something that I thought could be interesting and fed into the tutorial side of things was different time of day setups. I was quickly able to visualize this inside of marmoset and play with what a sunset and night time might look like. 

From here I focused on refining the idea and creating a more finalized blockout. As I grew more confident with my idea I ported everything over to Unreal and started setting up all the assets in a proper folder structure and getting the mood and lighting right. 

Here is a process gif to show how this evolved as I blocked out the scene. 

At this stage it was easy to see the trajectory of the project and visualize the end result when finished. I was pretty happy with how everything felt here but It lacked a little focus composition-wise as the street distracted me too much from the robot.

I tried one final composition which had more focus in terms of camera and scale. I also loved the idea of doing a portrait image as you don't often see this for Environment scenes. 

Decision to change the project

At this stage I made the difficult call to change the project theme. Somewhere along the line I lost the initial theme of an abandoned world with the robot and heading into overly stylized. I realized I was working towards something that my heart really wasn't in. Bearing in mind the plan for tutorials and a continued series of work it was best to take what I learnt and change gears. I am much happier about the new project and I have taken a huge amount of the process/learning forward into this. (will be showing this in a future blog) 

Next post

To round this work up I will be doing an update about the process on the environment blockouts, Shaders and tech I used to get the blockout to where it was at the different stages. As I put a lot of the framework in place even at a blockout stage it was a lot further along than any environments I have made in the past. 

As always love to get thoughts/feedback and comments. 

Anyone who follows my Artstation page may have noticed that there has been very little recent activity or updates lately. This is due to a major project I have started work on that is going to be taking up most of my time for the foreseeable future. In an effort to stay active and provide insight into my work flow, I have decided to document my on going progress and ideas on this journey.

The working title for the series of work I am developing is 'Human Emotions'.  I plan to let this title sit for as long as it doesn't feel weird or until I can think of anything better and more appropriate for the project- feel free to suggest something in the comments if you have time.

My aim with the project is to create a series of environments, assets and materials that I can be recognized for and set a standard of quality in my work. It is an evolution of my original V Ray work here; Previous Blog Post  In my previous post I wanted to work on something themed around robots, I planned to use the robot to tell the story and the basis or signature for future environments.

Story:  A robot in an abandoned world interacting with every day human themes or finding itself stumped by human emotions.  

Aims of the project: 

• Create a tutorial series that would cover themes of environment creation that aren't commonly covered in tutorials.
• Continued series of environment, assets and materials for portfolio
• Cinematic mood and atmosphere 
• Excellent rendering
• A key focus on telling a story 
• Great composition  with no question of subject matter

Here are some images of the first blockouts where I was initially trying out ideas. The assets are blockouts modeled in Max with the scenes created in Marmoset. Marmoset is a great tool for quickly iterating and visualizing ideas and I had a lot of fun throwing around simple meshes and lighting.

Here is my initial ref board for the project at this stage. I was also leaning more heavily towards vegetation to serve the tutorial.

After running several iterations of my CGMA Introduction to Substance course I thought it be good to try to write a blog aimed at Students and professionals learning Substance Designer. I wanted to detail the information that I teach every term and use this as a resource for people who are new to Substance. I hope that it is useful to a wide range of people and if it proves to be popular, I may do another more advanced one.

Using too many keys in gradient nodes

When people are starting out they use the gradient nodes sample gradient feature, it involves sampling an image to generate gradient keys. This is both a fun and practical feature but becomes a nightmare to edit later if it’s not correctly optimised. It's worth cleaning up the keys for refinement later. Plus alterations become near impossible with lots of keys. It's generally good practice when you get into a studio environment.

Greyscale and colour blending

It is good to stick to the general rule of using greyscale to create your height/roughness/metallic and colour for your diffuse and normal. It is not always 100% clear but often there are options in one node for both. You have either two versions of the same node or a switch, which alters between colour and greyscale. This is super important because it’s a lot cheaper to use greyscale. As you use certain nodes, you will see that it tries to convert between the two types if you use the wrong input. This is adding unnecessary nodes and makes no difference to the result so it’s worth identifying these issues.

Different versions of the same nodes 

Converting greyscale to colour/red shaded line dictates colour when it expects greyscale. 

Blending together normal maps 

Often beginners will try to blend multiple normal maps, which for the vast majority of cases makes no sense. It is better to focus on getting an excellent height map first. This will let you drive your other maps later down the line and blending multiple materials will be more successful. There are always exceptions to every rule and sometimes you will want to control your normal map separately to your height. Vegetation being a good example of this.

Macro to micro and grunge map

It’s always a good idea to focus on larger forms first and work your way through the varying scales of detail, this ensures for a compelling and well balanced material and helps you to tackle material definition. One mistake I see very often is using slope blurs with quite grungy maps very earlier on which makes for a noisy material.  You tend to find you will lose the macro to micro detail. Getting finite granular detail is a great process to have in your materials but it should happen once the larger forms are established.

Example of Macro to Micro

Relying on grunge maps to generate base shapes in a material

In the first week of the course, I set some simple patterns for students to create. One of which is a tree bark material. This is quite a complex pattern to create and naturally students look in the library to find the easiest and fastest result, which happens to be Grunge Map 005. It’s good to understand that while easy and fast is a great method, Designer's strength lies in iteration and procedural approaches. It’s better to break down the construction of the bark because as an artist that means you will have superior control later down the line. It's far more scalable and if needed you can create another type of bark in the future. Breaking it down means you have a great starting point to work from if you need another version in the future. 

VS

Over warping

Often when I see artists starting out in Designer there is a tendency to over warp shapes. Warping is a great tool but it can blur pixels together if over warped. Extreme warping can look quite unnatural and give nasty results.

Incorrect PBR values

This is probably the number 1 issues I see with people new to Designer and even in quite a few professional pieces. With PBR, it is very important that the base colour values are the correct RGB value. Often when authoring you tend to look at your material in one lighting environment and do not change the setup a lot. This means you balance your material for that condition and once it’s used in other setups it doesn't work as well. This was definitely something I was guilty of myself but working with different lighting setups and time of day systems across my career has helped me to understand how important this is. The other benefit this gives you is the ability to be very creative with lighting; you can make a super dark night-time scene with hardly any natural light or a very bright daylight scene and know your materials will work great!

To ensure your material is flexible enough to work in multiple setups follow the charts linked below.

Don't Nod Chart

Material Cheat Sheet

Unity Sheet 

Megascans Chart

Adding local colour variety in diffuse/variation in roughness

Observing scan data you can see how much variety exists in your albedo map that isn’t lighting information. As I have pushed and developed my materials, I have tried to push these details and learn from scan data. Using generators as well as grunges is a great way to get this information in and have it line up nicely with your normal and height information.

Pulling back from your materials 

A common issue with materials creation is creating awesome content and detail super close up but without much consideration when pulling back from the material. This often means you run into obvious tiling issues from not observing your material from a distance. Ideally, you want a material that works well from far away and close up. It’s a hard balance to get right. There are cases where putting unique detail into a material can work well such as a vertex blend that you don't see that often. Pulling back can help to identity problems and help to balance your material. Equally getting in close will ensure a good read close up. It’s good to keep in mind the use case of your material and cater for that.

Setting accurate material properties 

Often when marking work I see a great deal of metallic bricks. Although reflections can help to identify details in the material it's always best to try and make your material represent the type of material it's made from.

Use blurs to add detail 

This is quite counter intuitive but the example I always use is the Polycount diagram of baking. By blurring your height map a certain amount, you are allocating more pixels to the map. Similar to chamferring an edge in 3D. This is an important technique to remember when creating good normal maps

Image from Normal Map Wiki Polycount

Scale vs variety

It can be hard to know the amount of bricks to put into a brick material, or the amount of sand dunes on a sand texture. I often find it is best to look at pre-existing materials and experiment yourself to find a good balance. Remember you can push higher resolution the larger the scale of your individual elements in your texture.  This is a balancing act though because you want as much variety as possible. I have seen students putting double the amount of bricks in a material and although they have lots of variety the material can become muddy and low res. 

It's also worth mentioning the content of your substances. I have seen many passionate and exited students wanting to create amazing work which can lead to over complicated graphs attempting to tackle too many things. Try to not create a brick with damages and plaster all in one material but instead keep it simple and break these down as separate assets. 

Example of excellent scale and simple consistent material patterns, Week 1 Alina Godfrey

Thanks for reading, I hope this is useful to you. Let me know if you found it helpful and would like me to tackle more topics like this. 

After the Artstation challenge I took time to reflect upon my work. From university progression developed at a steady rate. Each project I improved and learnt something to take forward into the next concept I worked on. It wasn't until finishing the challenge I hit a bit of a brick wall.

Professionally, learning has continued to improve especially with Substance. I think this is in part to do with the company I share, but also the sheer time spent in particular disciplines.  With personal work I felt I underachieved in the Beyond Human Challenge. That’s not to say It was awful or horrendous and there were definitely some new lessons I took away. Bringing understanding of properly calibrated materials from work benefited the environment the most. This gave me flexibility in lighting the scene that I never had in previous projects. Yet ultimately I came out of it unsatisfied and disappointed with the end result.

Upon reflection I drew inspiration from other artists. Looking into their work and attempting to understand what shortcomings were needed to get there. The common theme included a few ideas I noted here. 

• Often they used offline rendered method such as V-ray

• A key focus on telling a story 

• Great composition made sure there was never any questions in subject.

Links to Artstation profiles. If you are interested in the above work 

Zhelong Xu

Cornelius Dämmrich

Andrew Averkin

Toni Bratincevic

After outlining areas I hadn’t quite hit yet, I began thinking of new projects. Pushing through the brick wall of development is about quality as well as the shortcomings mentioned above. In the past I worked with timelines set by myself as well as project deadlines such as the challenge. Although there was a lot to gain from sticking to a predetermined schedule I have opted for a much wider timescale this time round. The advantage of this means I can work on each stage while reviewing thoroughly before I move on. I want everything I am doing to be perfection.

I began this project in the way I have previously by firstly sourcing reference images. Initially working from other artist inspiration to look into what aspects of the images I found interesting. Why was the composition so successful and how did they achieve that result? What story was it that they were trying to tell and why was that interesting? How did the rendering enhance the quality of the image and what details did they focus on?

I started focusing on different tutorials and learning resources to help gain more background knowledge on these topics. A couple of standouts for me are below. 

Creating a Sci-Fi Alleyway

Lenses, Composition and Camera Angles

Compose shots to tell stories

After researching further into Vray and compositional techniques I started collating images of compositions that I felt were successful and worked well. I then started to think about loose ideas of a story. I have always created quite dark and moody themes within my work in the past and to try and change things up I wanted to opt for a theme that stayed away from this. I stumbled upon Goro-Fujita’s work as I loved the storytelling he created. He has a gorgeous sense of colour and atmosphere in his work which could be likened to a pixar or disney style animated movie. I particularly loved the image below which I thought showed a strong composition and theming which would be perfect for the type of project I wanted to create. 

This gave me a great starting point for my story; an old guy working on his robot to repair him. Giving me the chance to introduce lots of detail in the middle and background and show quite a homely warm atmosphere. 

Fleshing out the story further I started to look at other images that inspire me. I have always been a huge fan of Simon Stalenhags work and I knew that he created emotive imagery often including robotics. This image stood out to me as there was a cool sense of discovery using light and shadow. I thought that perhaps I could introduce that feeling as part of my image with a smaller robot in the foreground. I was interested in playing with the feeling of an older robot discovering the newer model being created. 

I then started to play with compositions.I treated this stage as my blockout, picking a setting and building some primitive shapes. I have been guilty in the past of picking an interesting project idea and being so excited to start that I haven’t seen the blockout process through thoroughly enough. After watching the Gnomon tutorial I picked up a couple of tips on ways to explore this process further and setup different angles. I attempted more angles than below but these were some of my favourites. In the past I never setup multiple viewpoints. Seeing them side by side allowed me to progress and take forward elements I liked from each. I never understood this before but you can't always expect the first idea you have to be the best. 

After attempting different compositions I never expected to work , this led me to create my final blockout image. 

*Robot is placeholder shapes I am aware he looks like big hero 6 

One thing I also did here was get lots of different advice and opinions on these compositions. Although it’s not 100% perfect I am pretty happy that this is going to make a successful image when fully realised. It is complete to a level that I can iron out a few kinks once I work further into the models. I can now comfortably move on to fleshing out props further.

I plan to update in blogs as I work on the project. Hopefully this project pushes above and beyond the quality at which my current folio sits. I would be interested in any thoughts you have and any tips you may have to offer. 

Ben