How Uncharted 4’s artists created their amazing landscapes

Uncharted 4: A Thief’s End is one of the best-looking console games of all time. Now you can understand why.

Specifically, you can learn why the action-adventure game for the PlayStation 4 has such stunning environments. Places from tropical islands to laundry rooms are full of detail in developer Naughty Dog’s newest game. That’s thanks to a talented team of artists and engineers at the studio.

[aditude-amp id="flyingcarpet" targeting='{"env":"staging","page_type":"article","post_id":1966724,"post_type":"story","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"dev,games,","session":"C"}']

GamesBeat interviewed Uncharted 4’s lead environment artist Tate Mosesian and environment artist Rogelio Olguin about Naughty Dog’s process for creating those beautiful locations.

GamesBeat: Maybe this question is too vague to answer, but why does Uncharted 4 look so much better than any other console game?

Tate Mosesian: This is a subjective question. I’m sure there are many people who would disagree. However, I think it’s fair to say that Uncharted 4 is one of a handful of current gen games that have set the bar for what a realistically rendered game can look like.

I’ve been a texture artist at Naughty Dog for 15 years. The intensity and passion we all put into creating the best looking games we can has always been part of the studios DNA, and with each new game we challenge ourselves to raise the bar. There are many elements that come together to create a game as beautiful as Uncharted 4. However, I’m a texture artist so I’m going to focus on the role textures and materials play in the pursuit of beautiful graphics.

When we started developing Uncharted 4 for the PS4 we chose to build our new render engine around the principles of Physically Based Rendering (PBR). More importantly, Naughty Dog has always been about handcrafted graphics, and we were going to continue in that tradition with Uncharted 4. The trick was to integrate the artistry of handcrafted assets with the rigid requirements of PBR.

A primary component of the PBR pipeline is textures and materials, and it is critical that the material parameters are correct for achieving a believable and consistent response from lighting. This meant that we needed to find new methods for generating vast amounts of content of the kind required to feed a PBR pipeline and at the same time maintain the unique quality of the handcrafted look that we are known for.

Early on in our research we connected with Allegorithmic (creator of the texturing application Substance) and together we began working on developing a texture/material workflow that would meet the PBR requirements as well as complement our established methods and allow us to continue to be creative and artistic.

I am a believer that if you give great artists the freedom to be creative and provide them with the tools they need to do their best work you will get the best results ,and if a particular tool yields consistent great results, artists will gravitate to it naturally over time. That’s the approach we took in-house with Substance Designer and Substance Painter.

[aditude-amp id="medium1" targeting='{"env":"staging","page_type":"article","post_id":1966724,"post_type":"story","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"dev,games,","session":"C"}']

We took a slightly different approach with our outsource pipeline because we needed to ensure consistent and predictable results across several external studios. So we created a more “black box” approach for the outsource vendors. That’s one of the great features of Substance Designer. Its versatility allowed us to tailor the workflow to meet challenges both internally and externally and the two pipelines integrate relatively seamlessly.

Rogelio Olguin: The quickest way to explain would be through our aesthetics. The Uncharted series and the games that are created at Naughty Dog have a strong sense of aesthetics. We strive for something beyond the raw data of what a texture is or any asset in-game.

We, as artists and designers, strive for something more, the intangible sense of the sublime. I think at the core that is why our games look better because we have a reason behind each asset placed in the world like a rock, phone, or trashcan. Our games are story oriented, and because of that we add as much depth as possible when it comes to the macro and micro of our environmental storytelling that may or may not even add anything to the greater story.

A simple example is like in prologue orphanage chapter with young Nathan Drake you will see a magnet alphabet with one missing letter “Y”. If you explore around you will find it. This small story element adds a playfulness of either a kid in the orphanage who did not return the alphabet letter back to place as asked by the nuns or teacher, hence the written marker “Y” with the question mark next to it. This is a tiny detail that adds nothing to the game, but it creates a sense of place and prior events.

[aditude-amp id="medium2" targeting='{"env":"staging","page_type":"article","post_id":1966724,"post_type":"story","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"dev,games,","session":"C"}']

I think this is why Naughty Dog games tend to have that feeling of being “better”, if you were to compare to other game assets in a vacuum you would see that Naughty Dog assets are usually at least of the same level of most other games. We are more interested in telling stories and when it comes to environmental scenes we want to make sure all sections are balanced in a harmony of special form and storytelling.

GamesBeat: How much more complex are the environments in Uncharted 4 when compared to previous games in the series?

Mosesian: Vastly more complex is how I would describe it. For example, with respects to textures and materials, the previous games on the PS3 were limited to a total of 138MB of texture memory [VRAM] visible onscreen at one time compared to Uncharted 4 which has a limit of 1.5GB. Between the more complex PBR shader, light maps, environment maps, particles, sky domes, wrinkle maps, etc. that vram got gobbled up fast which required us to come up with more sophisticated methods of texture streaming and material optimization.

[aditude-amp id="medium3" targeting='{"env":"staging","page_type":"article","post_id":1966724,"post_type":"story","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"dev,games,","session":"C"}']

Olguin: Environments range greatly in Uncharted 4. Some are about the same scope as prior games while we also have areas that are massive. In general, complexity did sky-rocket. We had more than double the assets than The Last of Us, for example. The biggest challenge came from new tech changes.

Prior to Uncharted 4, we did not use as many level of detail [LOD] meshes to optimize geometry count. In Uncharted 4 we used this almost everywhere — we had LODs almost on 60 percent of the assets in-game, which means the environmental team would have to support these assets in all aspects from modeling, collision, shadow proxies, and shaders.

We had many hurdles that the environment team had to go through. The biggest moment where tech came together was the E3 chase sequence. This by far was the biggest environment when it comes to driven distance and the amount of different environmental styles as you go from the market to the docks. When it came to memory, LOD tech, modularity, streaming, car physics, prop destructibility it was an extreme challenge. Nobody at this studio or any studio for that matter has tried to do a level section with that complexity when it came to the amount of art assets that needed to be created to make all these sections of the chase.

E3 chase became our litmus test for everything else. Without the chase sequence and finally having all of the core tech pillars coming together it would have been really hard to have another opportunity in one level to test out our collective thoughts of what a Naughty Dog’s Uncharted 4 should be.

[aditude-amp id="medium4" targeting='{"env":"staging","page_type":"article","post_id":1966724,"post_type":"story","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"dev,games,","session":"C"}']

GamesBeat: What kind of research do you do when creating environments?

Mosesian: First an environment layout is delivered to the team by the designer in a primitive “gray box” or “block mesh” form. The designer will play through the level describing the game play requirements and important story beats. Then the art director with concept artists will provide concept images that embody the general theme and feel including color keys. More detailed, specific concepts or “paint overs” are provided if an important establishing shot needs to be precisely directed or if a specific game play scenario or story beat needs to be supported.

Once the critical information about a level has been conveyed, the artists will set out to gather reference material. This can come from a variety of places. I often do an initial search online for images of places that look like the concept art. This helps me get an overall feel or essence of the environment. I usually keep my reference material limited to 10 or 15 images that represent exactly what I need to match the concept art. Then I break down the image into its individual components i.e., grass, dirt, gravel, etc. and then gather specific reference for that.

It’s important to note that when a texture/material artist gets a piece of concept or photographic reference they can’t just color pic from the image for local values because those surface values are altered by the light represented in the painting or photograph. The PBR shader requires actual local albedo (base color) values as an input. So one of the things we did at the start of the project was to build a reference database of surfaces, everything from leaf surfaces, front and back, to metallic surfaces. We shot the photographs in neutral ambient light with polarizing lenses so we could eliminate as much shadows and highlights as possible and then we color corrected the database to get as precise local color and brightness values as possible. Then we compiled the values of a wide range of the surfaces and created a reference chart for the artists to check for accuracy.

[aditude-amp id="medium5" targeting='{"env":"staging","page_type":"article","post_id":1966724,"post_type":"story","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"dev,games,","session":"C"}']

Olguin: As early as possible, we start putting together reference images from the areas we are working on. We try to get key images of the look we are striving for and somewhere between this processes we also have concept artists working on putting together the high level themes of the level being worked on.

For my levels I do all kinds of research especially if an area is based on a real world area. For the chase sequence we had many images, but most importantly I gathered some blogs that talked about architectural standards of those areas we were to mimic. With that info we were able to find out the difference of material usage and also what are the standard roofing material which ended up corrugated since it lasts longer than more traditional roofing material like clay, ceramic, and wood. So in general everything we could possibly get info on we try. On Madagascar the entire environment team went to see a documentary in a huge IMAX 3D theater.

GamesBeat: It feels like Uncharted 4’s environments invite players to slow down and examine more than previous games. Was that intentional?

Mosesian: It was our intention to encourage players to explore more and give more opportunities for traversing alternate paths. It wasn’t our desire to invite players to examine minute details, although those details are there, they are only there to serve the whole aesthetic and the aesthetic serves the gameplay and story.

I think that with the PS4 and its increased fidelity we’ve been able to represent much richer detailed and realistic environments and because of that people just naturally stop to soak it all in — much like you would in real life when you see a beautiful sunset or amazing vista, and we did very carefully plan those moments to engage the player and immerse the player into the game experience. Hopefully it’s not too transparent and the player doesn’t feel manipulated.

Olguin: Pacing is extremely important in our games and we test heavily for this. We do have areas that we want the player to look around and examine. This is intentional and definitely part of our day-to-day conversations when it comes to the major story elements. Some areas the player just flies through them, so it is all about the pacing. We also have plenty of small details in the environments that are not seen by many players and that is OK. This is what makes people experience and appreciate the amount of work and detail we strive for in our games.

GamesBeat: Playing through it, I felt like I rarely saw repeated assets. How do you manage to create so many unique objects, from plants to table lamps?

Mosesian: As I mentioned earlier the entire scope of Uncharted 4 is at least two or three times bigger than previous Uncharted games, and the number of assets for Uncharted 4 have at least doubled and in some cases even more. So, yes, in that sense we have increased the number of assets making it less likely to see repeats than in the past games. However, we could never populate a game this size with entirely unique assets, even with outsourcing. Logistically there’s just not enough time to generate that many assets. Take another look and you’ll see that there is actually quite a bit of reuse but even some simple tricks can disguise the repetitiveness. Things like lighting, shadows, and material blending can go a long way to reducing repetitiveness.

We also employed what we called instance features on some prototypes to increase variety while maintaining essentially the same memory footprint and limiting the amount of labor involved in generating all unique assets. For example, the umbrellas in the City Market and City Chase are actually just two umbrellas instanced several times. One umbrella is non interactive, which means that its position in the world never changes and the other is interactive and can be bumped around the world by the player. The non-interactive umbrellas used a shader feature that would remap the local color from a swatch of 16 colors based on where the Umbrella was placed in the world. If we used this technique on the dynamic umbrellas you would see the umbrella change colors as it moved through the world. So in this case we used a parameter on the instance itself that allowed us to assign specific RGB values to the individual instance. This method was slightly more labor intensive because each instance needed to be assigned unique RGB settings. Many of the vehicles were dealt with this way as well.

Olguin: This ties in with everything I mentioned above. But the short answer is outsourcing. We outsource a variety of these unique items so we are able to have almost no visually apparent duplicates in our game to make the game feel grounded. Plants specifically were created in-house by a core talented team of two to three artists. Foliage is extremely hard to get right through outsourcing so that was a decision made early on to not outsource these elements and give them the love needed especially since more than half of our game is covered with plants.

GamesBeat: Do you prefer working with indoor or outdoor environments?

Mosesian: I don’t have a preference. As a matter of fact ,I like to work on levels that have both. It’s a lot more challenging with regards to asset and memory management but I like the variety and I’m less likely to get fatigued by working on the same look for an extended amount of time.

Olguin: It depends for me. Some artists tend to choose one or the other, but I have always liked having a mixture of organic outdoor environments and some indoor also. On a project like this it is not rare to have multiple environments that you are dedicated to. I think it keeps me from getting artistically drained if I have some differences of styles through the length of making a game.