Hunter Thinks About Games

How Long is a Frame?

2022-06-05T00:00:00+00:00

I’ve recently stumbled upon a topic that differs greatly depending on the background of each game creator. It’s a fundamental question of time; namely, should time be measured in terms of seconds, or in terms of frames?

I am a stalwart defender of frames… but I still think seconds have their use. As for you, let’s dive into the merits of both, and then see how you’re feeling!

The Usual Defense of Seconds

I understand the appeal of seconds, I really do. They are a natural, universal way to measure a relatively short interval of time. A second is 1/60th of a minute, which is 1/60th of an hour. The arguments I hear in favor of seconds generally focus on this point: that seconds are common and natural to understand.

There are other benefits to seconds too… for example, it’s easy to hop online and find physics equations that are in terms of some value per second. But these benefits still tend to reinforce a bias of familiarity with the second.

Just because seconds are intuitive to most people does not mean a second is always the best way to express intervals of time in game design. That familiarity bias can even impact our ability to perceive time in more granular intervals.

Where Seconds Break Down

(For these examples, I am hacking around with some JavaScript code by Richard Davey I found here)

Let’s say we are making a shoot-em-up game, and we have a spaceship that needs to shoot rapid fire projectiles. To start off, let’s make it so where if you hold the fire button, it shoots 1 bullet per second:

When you see this animation, you’ll immediately think “Wow, those bullets need to shoot a lot faster!” So let’s try them faster… instead of a bullet every 1 second, let’s do a bullet every 0.5 seconds:

Still a bit slow! It really only starts feeling naturally like a shoot-em-up game once we get the shot interval down to 0.125 seconds:

When we’re dealing with gameplay actions, especially ones that need to respond to a button press, game feel is determined by changes that transpire over the course of fractions of a second.

Insisting on measuring with seconds means that the programmers and designers are going to be juggling a lot of fractional numbers! Fine tuning these very quick gameplay actions often means deciding over the difference between numbers like 0.2 and 0.25 seconds.

That’s where the familiarity benefit starts to crumble. While one may be able to easily discern the difference between 1 and 2 seconds, can one just as easily discern the difference between 0.2 and 0.25 seconds? For most people, fractional differences between numbers seem negligible. So, these seemingly tiny values hinder one’s ability to perceive the difference between them.

Re-Framing Our Perception

What do I mean by “frame” exactly? In a time sense, a frame is the time a game takes between drawing a screen and drawing the next screen after it. How many seconds this is depends on the frame rate, but in common practice we are dealing with 60 frames per second. This means we are dividing each second up into 60 parts.

To convert 1 frame at 60 FPS to seconds, the math works out like this:

1 second / 60 frames = ~0.01667 seconds per frame

When we recalibrate our perception to declare 1 as a much shorter unit of time, negligible fractions suddenly become substantial intervals.

Our 1-second shot from earlier is a shot every 60 frames. Our 0.5-second shot is a shot every 30 frames. Now the numbers we are seeing better match the “feeling” of those too-long intervals. In the end, when we settled on 0.125 seconds, those shots were firing about every 8 frames.

Feeling The Difference

(For this section, I got the GIFs from this imgur link, and the frame data from EventHubs)

I’d be remiss if I talked about frame timing without bringing up fighting games. In fact, I’d say my own tendency to think in frames is heavily influenced by being a fighting game fan.

In fighting games, the action unfolds quickly, and a lot can happen over the course of a second. Hardcore players pore over frame data, which is information about how long attacks take, the position of hitboxes and hurtboxes during the attack animations, and how much damage those active hitboxes deal to opponents.

As a rule of thumb in fighting games, slower attacks which take longer to execute tend to do more damage than faster attacks that can be thrown out more rapidly.

In the Street Fighter series, attacks are divided into three grades: Light, Medium, and Hard. As a case study, let’s look at three grades of Sakura’s standing far punch attacks in Street Fighter IV.

Sakura’s Light Punch (LP) animation takes 10 frames and does 25 damage:

Her Medium Punch (MP) animation takes 22 frames and does 75 damage:

Her Hard Punch (HP) animation takes 28 frames and does 110 damage:

Although we are ultimately dealing with a difference of 18 frames (0.3 seconds) across all three attacks, you can feel the difference in speed and weight among them. You can throw two LPs in the time it takes you to throw one MP!

Once you start to tune in to the differences of frame timings, you can start to train your sense of “how long a frame is.” I recommend reading Masahiro Sakurai’s Famitsu column on this subject (Or Source Gaming’s English translation).

Settling The Great Debate

Hopefully now, you can see some benefit to thinking about time in frames if you didn’t before. So now we’re all 100% frame-thinkers, right?

Well of course, nothing is ever that cut and dry. In a practical sense, you may want to take advantage of differences in perceiving seconds vs. frames.

At my studio Evening Star, we accomodate both approaches in our Star Engine. We have a delta-time value programmers can use called Time.deltaSeconds, which is generally 0.01667 when running at 60 FPS. We also have a value called Time.deltaFrames, which is 1.0 when running at 60 FPS. We use either-or to drive timers depending on the context.

For most people, numbers between 1 and 100 are easiest to think with and visualize. To play into this natural inclination, we generally use this rule of thumb: favor expressing “fast” intervals less than 1 second in frames (which avoids fractional second values), and favor expressing “slow” intervals longer than 1 second in seconds (which avoids large frame values). One may show up a lot more often than the other depending on the nature of each game.

Either way, it is useful to develop your perception of frames, because they have the power to help you make the most out of timing details!

The Price to Fill Your Screen

2022-01-23T00:00:00+00:00

The video game business has changed a lot in the past 40 years. Game systems have grown more powerful, and the TVs they’re plugged into have gained a much higher pixel density. Generally speaking, development cost has increased as well. This got me wondering, what kind of relationship does screen size have with game development budgets?

The Quest For Higher Def

Video games are played on screens, so game technology is intrinsically linked to display technology. For decades, the most ubiquitous screens people had in their homes has been TVs, so the consumer TV market has always set the benchmark for what game systems, and the software running on them, should be able to do.

When the Famicom/NES came on to the scene in 1983, it supported an output resolution of 256x240 pixels. That is, 256 pixels wide and 240 pixels tall. The height of 240 scan lines was the standard for analog TV, and this resolution retroactively became known as Standard Definition (SD). For console game systems, this 240-line height was the norm from the NES, through Sega Master System, Sega Mega Drive/Genesis, Super Famicom/SNES, Sega Saturn, Sony PlayStation, and N64. That’s over 20 years of standard def dominance!

In the 1990s, in the midst of SD’s reign, the gaming world expanded into polygons, propelled by advances in 3D rendering and CD-ROM technology. Production values and development budgets increased to meet the potential. Then at the end of 90s decade, everything came together TV technology began expanding into High Definition (HD) resolutions.

As the game industry entered into its Fifth Generation (PS2, GCN, Xbox, Dreamcast) in the year 2000, output resolutions doubled to 480p, or a 480-line height. Then the Sixth Generation (PS3, Xbox 360, etc.) brought a 2.25x increase to 1080p. While the Seventh Generation (PS4, XB1, Switch) stayed steady at 1080p, we have just seen another doubling with the introduction of the Eighth Generation (PS5, XBSX|S). What this means is, over the past 40 years or so, output resolution has been increasing closely to an exponential rate.

Exponential growth of console output resolutions. I guess 8K consoles will be coming around 2030-2032… we’ll see!

Meeting The Pace of Change

As display technology has become more advanced, game developers have constantly risen to the occasion with sharper graphics, crisper audio, and faster action. As the technology behind games has changed, the skills and tools needed to create them have changed too. Productions became more complex, and demanded larger teams.

While the state-of-the-art has evolved, the gaming public’s tastes have kept in tune too. Staying competitive in the eyes of the critical public demanded that developers increase production values of their games. In this environment, developers who couldn’t adapt to shifting technical constraints and consumer expectations perished.

Since around 1990, a new wave of game systems has been coming out about every 5 years or so. To keep up with the times, these competitive developers have had to adapt their productions to these new systems. But not all platforms are equal! Developing a GBA game was obviously a less expensive affair than a AAA game on Xbox 360. Anybody who’s familiar with either system can tell you that the differences in capability are vast.

Screen Height as Proxy for Potential?

Interestingly enough, the GBA was a handheld system with a screen height of 160 lines. It was prominent in the early 2000s, alongside its console bretheren which were blazing the trail into 480p. So I had a hypothesis… can we use screen height as some kind of general benchmark for technical potential?

The GBA also started a rule-of-thumb that a given console generation’s handhelds could generally match the power of the consoles from two generations prior. So a GBA’s power was comparable to the SNES (Super Mario World: Super Mario Advance 2), DS to N64 (Super Mario 64 DS), 3DS to GCN (Luigi’s Mansion) and so on.

Through this stretch of time, two console generations was roughly equivalent to 10 years. Trying to make a Nintendo 64 game would have been a lot easier in 2006 than it was in 1996. After all, a developer in 2006 would have 10 years of technical progress at their disposal. So, if screen height is a measure of potential, is there a way we can we compare screen heights with game budgets, and derive some industry-wide indicator of efficiency or technical mastery?

Gleaning Some Intel

Since I was curious, I decided to try and take a look. After all, it must be easy to examine 40 years of game dev budgets, right? Well, unfortunately the game industry is notoriously secretive, and production budget numbers are some of the most closely guarded business secrets in the industry. Fortunately, various industry insiders have been willing to talk in the abstract over the years, giving me enough breadcrumbs to start piecing it together.

There’s also the fact that this is a complex problem with multiple variables. The games market has become more mature and specialized since 1983, and there have been large market fragmentations over the years: the invention of handheld systems and smartphones, the emergence of indie-friendly downloadable distribution platforms, and the introduction of alternative revenue strategies like free-to-play and subscription models. So for this analysis I thought to narrow the focus on full-price games for console and handheld systems. Also, I don’t want to get into the time durations of dev cycles… I am just attempting to focus on the total development cost.

It’s also an exploration that spans four decades. If we want to compare budgets across generations, we need to make sure everything is adjusted for inflation and expressed in today’s dollars.

So, given all those caveats, here’s a peek of what I came up with:

Some additional notes about this data:

These all should be considered ballpark numbers, and taken with a grain of salt!
A lot of these numbers were gathered from this page on the Video Game Sales Wiki
The budget numbers highlighted in yellow were educated guesses based on the relative development costs from a 2005 CEDEC survey in Japan
I could not really find any data about the costs of developing for SEGA systems like Master System, Genesis, Game Gear, Saturn or Dreamcast, so they are omitted.
I could not find anything definitive about Wii U or PS Vita development budgets.
Switch crossed over the worlds of console and handheld. So, there is not really an average development budget case.

Insights Gained

In looking at these numbers, in all their flaws, I did have some general insights.

First of all, the “dollars per scanline” metric seems to spike when new technologies cause a paradigm shift, then drops as developers become accustomed to the tech. Perhaps my theory holds water… but unfortunately I don’t think this data set is strong enough to confirm it. So from here on out is a bit discursive! But I still had more to think about:

The jump to HD caused a massive shift. Games became at least 10x more expensive to make when we jumped from 480p to 1080p.
When games became 10x more expensive to make, games had to sell 10x more copies to break even. Despite this, the increased dev cost was not complimented by 10x market growth. The combined install base for Sony and Microsoft consoles has stayed consistent at around 170 million since 2000, when games were still in 480p.
The “jump to HD” in 2005 parallels the “jump to 3D” in 1995. Games became 10x-20x more expensive to develop for the PS1 than they were for the SNES.
The MSRP of full-price games has lost a lot of value to inflation. There was a $10 industry-wide MSRP bump in 2005, but by that time over $20 of value in today’s dollars had already been lost compared to 1989. That $10 price bump only reclaimed about $6 of value. We’ve lost another $26 since then.
So while dev budgets for full-price games have ballooned 50x-100x since SNES, the inflation-adjusted full MSRP has lost about 40% of its value in the same time.
The handheld world stayed relatively steady, despite HD making things a headache on consoles.
2D pixel art games are probably the cheapest kind of games you can make!

Lessons For Today

Although I may not have adequately answered my original question, I still feel like I came away with a better understanding of how the market evolved, and where that has led us today. It is pretty staggering to think about those technological paradigm shifts that have driven massive changes in game production economics.

Something that this data misses is too is all of the ways the games market has evolved to allow for different scales of production to coexist. While $50,000,000 games do happen, there are also $5,000,000 games, $500,000 games, and $50,000 games. Unfortunately, their MSRP is scaled relative to the full-price games… they might end up being sold for $40, $30, $20, or $10. Now that Switch has crossed the handheld world over with consoles, is a certain production tier potentially today’s equivalent of yesterday’s “handheld” games?

As ever, there is also a business case for novel technical optimization. If a studio manages to come up a technique or process that can yield competitive games at 1/10th of the cost, they can claw back the value that has been lost to the march of screen sizes. That may seem outlandish or even impossible, but I can’t help but wonder, what would that look like?

Numbers We Can See

2021-12-21T00:00:00+00:00

Video games run on computers. Computers are a lot like calculators: They are machines that take numbers, do something with them, and give other numbers back.

After computers came to exist in the 1950s, people began to realize all kinds of ways numbers could be used to do different things. For example, numbers can be used to represent letters. They can also be used to control other devices, like telling a screen to generate images. Fast forward to now, and these ideas have become so commonplace that we barely even think about them.

Video games have always been built on computer technology. The earliest things we can call “video games” date back to software written on some of the first computers in the 1950s. The birth of video games as an industry was propelled by the availability of affordable microprocessors, beginning in the 1970s.

So, video games are built on top of computers, which are built on top of numbers. That means we can think about numbers as the essence of video games. Just like you and I are essentially made of atoms, everything in video games is essentially made of numbers.

What does it mean for the developer, and what does it mean for the player?

How Much Is Visible?

Numbers may be the atoms of video games, but when you look at another person, you don’t see each atom making up their body. Instead, you see a person; a meaningful whole. Even though in our brains we logically understand that everybody is made up of atoms, in our hearts we see each person as a complete and self-contained being.

There is a similar intuitive leap with games. We do not see them as a collection of rapidly changing numbers, but instead see what the numbers represent.

When a developer is writing the code for their game, they are making all kinds of numerical decisions like “How fast should the player walk? How high should they jump?” But players think more abstractly, like “Can I outrun this enemy? Can I make the jump?” Players only tend to think about numbers when they are displayed somewhere on the screen.

This means that there is a lot of power in controlling which numbers a player sees! A developer’s decisions about displaying numbers has a profound impact on the player’s experience.

Let’s look at an example from Super Mario Bros. (1985)

The player can see four numbers on the game screen: “Score,” “Coins,” “World,” and “Time.” At the same time, there are many other numbers you can’t see, such as Mario’s X and Y positions, his velocity, the camera scroll offset, and so on.

Each of the numbers the player can see serves a purpose. “Score” and “Coins” tell the player how well they are playing. “World” tells the player how far they are progressing. “Time” informs them of their limitations.

What if these numbers weren’t there? If there were no “Score” or “Coins” visible, then interacting with coins, blocks, and enemies would have less consequence. If “World” were not visible, the player would not understand where they are in context of the game. If “Time” were not visible, then the player would die for seemingly no reason when they hit the time limit!

Each of these numbers tells the player something the developer thought they should know, so they can understand the rules of the game. They also reveal that the developer thought displaying a number was the best way to convey the information to the player.

Visualization and Intuition

When we consider all the games out there, it’s always the case that some numbers are visible to the player and some are not. But things are rarely as cut and dry like in the Mario example.

Developers don’t just decide which numbers to show players, they also control how numbers are shown. Numbers will inevitably be present in games, but effective visualization lets players understand them without having to think too much about it. This is the principle of visualization, and it lets players understand numbers with their intuition instead of their intellect.

Visualization is important because players tend to feel more than they think. It is only human nature, after all. Ideally, a player can look at a screen and instantly understand all the information they need, then use that to inform their decisions. In many games, players are making several decisions every second!

Let’s take a look at another example: battle system visualization across Final Fantasy games.

In the first Final Fantasy (1987), only the player characters’ HP was initially visible on screen:

You may look at this and think, “Well, how much HP do I have?” In Final Fantasy II (1988), the max HP and MP became visible:

By the fifth entry, Final Fantasy V (1992), the series had moved to an Active Time Battle system. This brought the Time gauge, unfortunately at the loss of other stats:

Then by Final Fantasy 7 (1997), all of the core battle stats resurfaced as both numbers and gauges:

It is clear that over the course of 10 years, the developers had shifting ideas about how to help the player understand the state of the battle. They also were dealing with the ground shifting beneath their feet as they moved through console generations. Regardless, through experimentation and by balancing needs, they discovered more refined solutions.

Visualization is the lens through which players see and feel the rules of the game, so developers ought to ask, what is the best way to help players do that? Finding the answer is of course a unique challenge for every game.

These examples show that visualization is an ongoing challenge. The ideal solution may be elusive, or could even demand years of exploration. It is a developer’s responsibility to consider visualization thoughtfully.

Suspension of Disbelief

Visualization is just one facet of numbers. For every number we see on the screen, there are a lot more pulling the strings behind the scenes. Sometimes we even get an opportunity to peek behind the curtain, whether that is driven by developers or their player communities.

One such example is Individual Values (IVs) in the Pokémon series. Pokémon celebrated its 25th anniversary this year, and has iterated on its core series gameplay across nearly 20 entries. But IVs have been part of the formula since Pocket Monsters: Red and Green (1996).

Put simply, IVs are a Pokémon’s base stats for battle traits like Attack, Defense, Speed, and so on. In the earliest games, a Pokémon’s stats at a given level could be determined by a math formula that combined their level with the corresponding IV. But, even though IVs were so critical to calculating stats, they were completely invisible to the player!

As the series grew in popularity, the competitive player community did as well. Over time, players got together and figured out how the hidden numbers worked. They also began to wonder why such important numbers were even hidden in the first place!

Game Freak’s Shigeru Ohmori, director of Pokémon Omega Ruby and Alpha Sapphire (2014), explains the philosophy behind this in a 2014 interview with Wired.

Each Pokémon does have [Invidividual Values], but I don’t consider those data as parameters. I prefer to think of them as real, living creatures. It’s the same way that if you have a pet and someone else has the same breed of dog, it’s a different dog. That way people can play the game and my Pokemon will be different to your Pokemon even if they’re the same type.

As the series progressed, the concept of IVs were expanded upon, and gradually became more visible and available to advanced players. The stat formulas also grew more complex, becoming more influenced by the time a player spends with their Pokémon.

Nonetheless, the endurance of the IV system throughout the series is a testament to the original intent of making Pokémon feel organic, and the effectiveness of hiding these numbers from the player to suspend their disbelief.

It’s All About The Numbers

Video games are numbers at work! As players, we see some of them directly, while others are hidden from us. Game development entails making all kinds of decisions about which numbers to convey abstractly, and which to convey concretely. A well-honed game experience is one that finds the right balance.

Don’t you wonder, how might you use numbers to express your ideas?

Introduction to the Blog

2021-12-17T02:16:00+00:00

Welcome to Hunter Thinks About Games! My name is Hunter Bridges, and this is where I aim to post various thoughts I have about games.

I’ve been playing and making games for quite a while. I grew up in the 1990s, alongside the emergence of 3D games, and I have always been fascinated by the craft and the art of game making. Hopefully, if I meet my goal of sitting down to write a post from time-to-time, you’ll gradually get to know me through through what I talk about here.

To be honest, I do not have a super clear version for this blog yet! What I can tell you though, is that I have a keen interest in examining game design and development choices through a historical and cultural lens. I like to see how developers cultivate ideas across a series or within a genre. I enjoy inferring individual impact as people change positions through their careers. These are just some examples, but maybe they give you an idea where I’m coming from!

I do, however, think there are things that I am not seeking for this blog to be. I don’t aim to be a critic, or to convince anybody to see things my way. I also don’t intend to write from any kind of all-knowing perspective.

I just want to have the opportunity to talk about certain games, series, teams, or ideas that have made an impression on me, and see if we can all learn something along the way!

Anyway, thanks everyone for taking the time to read my blog. I’ll do my best to write something good!