Gameplay Fundamentals Revisited, Part 2: Building a Pacing Structure

By Mike Lopez

[Former EA and THQ design director Mike Lopez continues his analytical series by looking at pacing in games versus films and TV, explaining how careful planning can produce a perfect intensity curve for games.

Continuing from his previous installment, Lopez sets out nine fundamental points that will allow you to developing an Intensity & Pacing Plan for your game -- a document that will help you craft a game with consistent, entertaining pacing and reduce wasted work.]

Introduction

I am continually surprised that many game developers in this day and age still utilize a very old-school, haphazard on-the-fly method of level design. When questioned, they argue they can produce a quality campaign with this technique.

The reality, however, is that the cost of continually reworking levels and missions until a user-appealing structure is realized is extremely prohibitive. Ultimately the end results will always be a rougher progression that falls far short of a riveting experience.

Because both film and TV have the luxury of significant extra film footage, and the flexibility of editing, these linear mediums have the advantage of relatively quick changes to pacing which is quite effective, but these are luxuries that are not available to the interactive nature of games. The only way to get there efficiently is to plan ahead.

I myself have worked on and with many highly experienced and talented teams (Road Rash 3D, Bond, Scarface, Baja, etc.) in the past who, due to a rushed or arbitrarily-ended pre-production, first generation engine/tools woes and/or over-confidence thought their situation was different -- only to end up throwing away almost all of the missions, levels and/or courses deep into production. The level/mission production team then had to start over on a massive body of work with only a fraction of production time remaining.

I only wish those projects could have benefited from these freshly-solidified processes. These talented teams have always regretted the time and cost of throwing away work, and the level/mission production teams have all had to scramble to get the new content built with often too little time left for adequate iteration and tuning, making the quality suffer further.

Even world-class teams with vast resources like the BioShock team (2K Boston/Australia) suffered through the cost and pain of massive throwaway (in their case, I believe all of the missions were entirely redone in the final nine months, and I suspect much of the level layout and content was redone as well).

The cold, hard lesson we must all see is that teams that over-confidently think they can avoid those mistakes using old-school, on-the-fly level design methods are destined to repeat the same highly costly content throwaway mistakes that much of the industry has been making for at least the past 10 years.

For the top high-profile titles of today, that throwaway easily translates into millions of development dollars wasted (enough to probably implement every design feature from your wish list) and often results in large delays and costly ship windows missed. It is time we all stop repeating the same unstructured level process mistakes and learn to utilize pacing and intensity processes like the older, more experienced, and more efficient entertainment industries (film and TV).

At a high level, the entire pacing structure encompasses three major bodies of work:

The Intensity & Pacing Plan is a structured level plan with set intensity magnitude and trend targets for the events in each level and over the entire campaign (Steps 1-7 below).

The Initial Level Implementation is the first pass of production for each level, where the level team will be using the Intensity & Pacing Plan as a blueprint (Step 8).

On-going Level Reviews and subsequent Level Iterations will be conducted to better match the intensity, timing, and gameplay progression targets (Step 9)

As these three bodies will encompass the bulk of design production, they must be led by one of the senior design leaders (ideally the creative director, lead designer, or lead level designer) and also monitored and supported by all the key design leaders.

9 Steps To Achieving A Mind-Blowing Pacing & Intensity Structure:

1. Brainstorming. Assuming the team is not locked into a precise level sequence (e.g. a movie recreation), get the entire development team together ideally for a full day Pre-Production Kickoff off-site to brainstorm (note that with a second day off-site, the team could further work on design/tech/art goals, IP goals, production processes, and pipelines).

More people participating in the brainstorm means more ideas, but it might make sense to break a large group into smaller sub-groups to ensure more people have a voice in the process. These types of large scale sessions also generate great creative and motivational buy-in within the team.

First, start the brainstorming with a session on specific settings for the levels. Try to come up with two to three times as many locations as you plan to implement. List, discuss, and record all the ideas openly without qualification or judgment. Get everyone to list the level locations in order of intensity and appeal.

Examples: Mayan Pyramid Complex, Cambodian Temple, Mesa Verde Cliff Dwelling

Next, brainstorm a list of high-action events. These events may be focused around the locations (say a Halo-style level hub with animated or unique geometry) or instead around the universal themes of the IP (like a vehicle chase in a James Bond game).

Shoot for five or six events per level, so that you can later whittle those down to the best of the best. If you cannot even come up with three events, chances are the location may not be so interesting to build a level around. Separate out or flag the location-agnostic action events that will work anywhere in order to signify that their position could potentially be shifted to another level as necessary. Get everyone to list the action event ideas in order of intensity and appeal.

Event Examples: Earthquake topples temple structure, cliff side collapses, dam breaks and wave rushes through canyon, soldiers repel on lines busting through glass wall to quickly outnumber the player, fuel depot explodes and causes cascading explosions along pipeline, person or vehicle chase through tunnels, cascading car explosions ending in fuel tanker explosion, outrun the lava flow, pillar topples over, stone arch collapses, suspension bridge starts to unravel, barrel explodes a hole in wall to reveal a squad of enemies, etc.

To round out the session, come up with a list of, say, five to seven generic events which utilize main mechanics, which could easily be used with minor variations (i.e. combat with x henchmen, vehicle chase in a different setting, or repurposed puzzles.) This will be a short list of items that will be used to flesh out the rest of the levels in between the key action events, and by definition these will not be as exciting or unique as the key action events identified in the previous brainstorming session.

2. Prioritization. After the group brainstorming session, consult the team leads for task/time guesstimates to each location and event. It may take the leads a week or so to distill the major tasks, meet with key team personnel, and determine better estimates.

Evaluate your project schedule and resources with the team leads, and estimate the number of levels and scripted action events (using brainstorm examples) that can be safely constructed. Take the number of levels possible, and then cut both lists down by 10-20% to ensure you are conservative in your estimates and to reduce schedule risk.

Consider cutting these back even more when also building a new engine and new tools with your game. The conservative estimates also help emphasize a commitment to focusing on a smaller set of content that can be built to a higher degree of polish and quality.

Next, rank the level locations in terms of coolness and bang for the buck. A setting that is 10% cooler but takes five times as long to build is probably not a great trade-off, so this step ranks the relative value. Make sure the number of levels is narrowed down to a conservatively manageable set for the available resources and schedule, but keep the list of all locations intact for the moment.

The team leads need to make a mutual commitment that the order of priority may shift during production, but as the project progresses levels can and will be cut from the bottom of the prioritized list up as the project schedule and personnel needs determines. This risk is the reason why coming up with a high-confidence level number estimate makes sense.

3. Story Framework. Figure out the high-level story and campaign sequences (separating out the cutscenes from gameplay) that will provide the best experience. Skip the details and dialogue, and just do a high-level outline of the plot points and how they pertain to each location, along with a high level of mission objectives.

Note that action plot events will be of relatively high intensity while dialogue or travel events will be of relatively low intensity and that contrast is in fact desirable and useful.
- Boy Meets Girl (dialogue)
- Boy Chases After Girl (action)

The advantage of starting with high level plot events is that it is easier to move some of them around to best fit within the Intensity & Pacing Plan and/or a specific level location.

Next, take the key plot points from the story outline and rate the perceived level of intensity on the same relative scale (dialogue heavy plot events will fall in the low intensity category and will contrast quite nicely with a scripted action event). List this plot-specific event table in chronological order, but keep it separate from the generic event table.

4. Level & Key Event Sequencing. Using the top bang-for-the-buck action event ideas, have the core creative team do a different rating in terms of perceived intensity (just vote on which concepts seem more intense and rate them on a relative scale of say 1-10), then order the events into a table of increasing intensity.

[Note that your relative scale may need to be wider if you are planning a large number of levels, but this makes it harder to manage. It is always a safer development practice to focus on a smaller number of environments that can be polished to a higher level of quality.]

Next, add a column to the generic action table and mark which level locations each event could potentially work with. In a new added column, mark which level location(s) would work best for each action event.

In the generic action event table, add a location column and assign at least three action intense events per location to roughly fit the three ascending peaks in Figure 7. At the same time, whittle down the set of locations to the number previously decided, basing the cuts on the levels which do not work as well with the action event ideas and, to a lesser extent, on which locations provide the lowest bang-for-the-buck appeal.

Once the locations are set, you can hide the columns for Potential and Ideal Locations, or delete them entirely (assuming you have a reference backup). Move the Location column to the far left and group the rows with matching location entries in order of ascending event intensity. Do a manual sort of the location groups to roughly demonstrate increasing progression of the events within the level.

Fig. 7: Structured Level Content Dictates Pacing (the mission starts between the first peak and trough)

In the generic action event table, rearrange the levels so the intensity events increase from one to another to form the initial Intensity & Pacing Plan. Order the levels so that the intensity of action events builds from level to level as in the Table below (values in blue), with each high peak being the climactic action event of each level. Some additional event swapping can and should occur here, in order for the Intensity & Pacing Plan to best fit the ideal targets shown in the graph.

The following table is a fictional example of a first pass Intensity & Pacing Plan with both ideal and projected intensity ratings for the action events. The projected ratings are the perceived gut-feeling intensity values voted by the group to the event concepts. Note that the example values used are fairly rough. The anomalous trend outliers are marked in red and purple. The latter values may be acceptable under a more relaxed trend as per the next item.

Simpler Trend Options: A simpler and perhaps only marginally less satisfying trend option is to focus only on the intensity targets building from level to level (blue values) and building within the level, but not to be so concerned with the magnitude after the first event of each level as long as the values are increasing.

Under this model, the purple values would not need to be considered anomalous, although large value deltas >20% could still be considered questionable. An even more relaxed trend option would allow initial level intensity values for the first action event to be equal to or greater than the one before in which case there is a smaller amount of anomalies to consider (but again, you trade off even more overall experience satisfaction to gain the added structure simplicity and flexibility).

The best approach is to start with the strictest trend criteria and ideals, using a steadily increasing trend as in the table above and as graphed in Figure 8. Then, if you are experiencing diminishing returns on iterations (multiple iterations with only marginal improvement towards the target values noted during Step 9) during mid production, decide if you wish to relax the criteria a bit to make the targets easier to hit.

Figure 8: Intensity Curve Over 4 Levels (fixed pace increases)

5. Plot-Specific Events & Intensity Lulls. Take the key plot points from the story outline and rate the perceived level of intensity on the same relative scale as the event ratings.

Dialogue-heavy plot events will fall in the low intensity category, and will contrast quite nicely with a scripted action event when placed just before or just after. List the plot-specific event table in chronological order, but initially keep it separate from the generic event table.

Identify potential low intensity peaks -- both generic event and plot specific -- and slot them just before and just after the high intensity peaks in an alternating trough, peak, trough... trough, peak, trough rhythm.

Continue to chronologically layer in the main plot points into the event table, ideally replacing and weeding out the least appealing generic action events -- or even those that are projected to provide anomalous intensity and fail to support your target trends. Ensure that all plot point highs and lows have a place in this newly combined Intensity & Pacing Plan table.

Each key action event (generic and plot-specific) will need to be sandwiched with an intensity lull before and after in order to provide the essential contrast described earlier. For simplicity's sake, we will not worry so much about the intensity values of the lulls. We will only set our criteria to ensure that there is a pair of bookend lulls before and after every key action event within each level.

Layout low intensity events as low-risk instances where the player feels safe and relaxed (e.g. level navigation, an opening of view from a tunnel or darkness, a reveal of scenic wonder, a low-risk puzzle, item searches, or a feeling of accomplishment), and these can be used to fill in the lull bookends around each key action event.

The order of these relaxed layout events should be fairly interchangeable, and because they are generic, they can and should be reused in varying ways. If there are not enough low intensity lulls to sandwich every action event, then mark a generic placeholder lull in that spot.

This will communicate to the level designers and/or mission scripters that there needs to be a break in action, and a sequence of low-risk/low tension calm or humor for some period around each key action-intense event.

Typically once the plot points are set in the context of a location, the story can be evolved further for the best fit. Take another pass at the plot outline to resolve any anomalies and to make improvements and stronger connections with the environment location where possible. This would be a good point to lay out any mission objectives for that level since they will inform the plot outline and vice versa.

6. Time Metrics. In a new column set some time metrics for each level so you know roughly how far apart each action event should be in time.

Always set the target time length for levels to be either all the same or all increasing (during production you can allow maybe 5% variance but set the targets more strictly), as discussed in the earlier progression article. A valuable trend opportunity that is commonly missed is to leave the length of levels/missions haphazard and have them end up randomly erratic.

Do not worry about spacing the low intensity troughs too strictly by time/distance or about sequencing their magnitude as precisely as the peaks; just make sure that you have low intensity events (troughs) or placeholder lulls to bookend every major action event (peak) to provide contrast to the action. Wherever possible, replace each placeholder lull with a specific segment of low-risk travel, revealing scenic wonder, or similar tension-reducing event that makes sense in the given level location.

The intervals between peaks within a level should all be the same or increasingly closer in duration. If you are going for a more advanced and complex increasing pace of peaks, (Figure 5) the first event interval in each new level should be reset to that of the prior level, or perhaps just slightly faster, to maintain enough range to make the changes noticeable.

Fig. 5: Relative Intensity Graph of a TV Drama Series (Amplitude Increasing for Greatest Contrast)

If you are going for an increasing level length, then you will likely have to add events after a few levels to keep the intensity peak interval the same or only slightly increased. You might also start from fewer main intensity events (say two or three for level 1 and up to five or six for the final level).

To help visualize the pacing trends, it will be useful to add time gridlines to the graph (say in 30-second increments) to clarify the peak intervals and the ideal pace rhythms.

7. Top-Down Trend Evaluation. Step back from the levels and view the distribution of intensity from one level to the next over the entire campaign, both in the Intensity & Pacing Plan table and visually through the graph of the data.

Verify that you have increasing intensity trends from level to level as in Figure 8. Also identify anomalous points that disrupt or weaken those trends (say one event decreases, stays the same, or has a very slight increase in intensity compared to the one before it -- see the example table below).

Discuss ways to adjust/evolve the high level design of the event in question so that it fits closer to the ideal intensity level (adding or reducing action elements, geo, hazards, props or enemies to produce the desired intensity rankings).

It is important to iterate the Intensity & Pacing Plan at this point so that you can further improve the set intensity targets and trends that you will need to work towards. Because the implementation ratings will differ from the projected ratings, you need not make your projected ratings perfectly hit all the projected intensity targets on paper. Just get as close as possible, and ensure that the appropriate increasing trends are upheld as in the table below.

8. Initial Level Construction. Now, the level designers or world builders should proceed with the initial level construction, using the target specifications and framework from the Intensity & Pacing Plan finalized in the previous step.

The level builders and/or mission scripters now know where the key level events should be placed and roughly how far in time they should be spaced apart. Compensating for projected average player deaths and event completion times, they can turn the time metrics into distance (assuming the player avatar speed is locked down or close to final).

These key designers also know where in time to build in relaxing lulls, dialogue-points, or scenic vistas to reduce the tension after the resolution of a major intensity event.

Using the projected distances between the action events, they can also determine where to place these lulls -- although less precision is needed with placement of lulls than with the action events (just make sure they are close to that event).

As they build out the level, they will fill the sections in between each lull/action event/lull group and the next with the reusable portions of gameplay in such a way that they help build the intensity up to the next big action event (e.g. a sequence of battles with progressively more henchmen or more challenge).

They must also be very careful during implementation to make sure that the intensity of any reusable sub-events does not surpass that of the next key action event so as to preserve the intensity target trends from the Intensity & Pacing Plan.

When working in an Agile Development environment, begin the first pass with a small set of levels that are adjacent (say levels 1-4 if there are enough personnel for four level teams). For the second stage, initiate the initial implementation for a second set of levels (say 5-8) and do a full round of iteration on the first set. In the third stage, initiate the initial implementation for a third set of levels (9-12), and complete another round of iteration on sets one and two (levels 1-8). Continue this method until all levels are working with initial implementation of all the key events.

Note that if the project is more than two thirds of the way into production without completing the initial implementation of all levels/missions, then a full round of level/mission cuts should occur immediately, and the Intensity & Pacing Plan should be reworked. If cuts are necessary, then this signals that the ongoing estimates from the leads have been inaccurate, and thus an even more conservative level of cuts should be made at this point (ideally cut 30-50% more of the unimplemented levels than the schedule predicts there is time to implement).

Level designers should also utilize the lessons of theme park design and urban planning to affect pacing via the level layout. Those of you who attended Brian Upton's excellent GDC 2007 lecture on Narrative Landscapes will have a great perspective on this. Visual pace can be created to coincide or contrast with the pacing of the action to add to the highs and further enhance the lows in intensity. Geographic elements such as sight line reveals can and should be used to help release tension after a high intensity event, just as claustrophobic sections and mental edge barriers can help heighten tension leading into a major action event.

9. Review & Iterate. Throughout level production, the creative leaders and team leads must review the in-game levels and compare them to the Intensity & Pacing Plan targets and trends. They must then schedule appropriate time for the levels to be iterated on in order to adjust the events to come closer to the targets. These reviews are also the best time support and ensure the accuracy of the Gameplay Progression Plan.

Review the levels with those building them, and re-rate the intensity of the in-game main action events (keeping the previous ratings out of view so as not to bias the results).

Compare the event intensity ratings to the target specifications from the Plan. Where an intensity rating is different from the target value, adjust the on-paper design specifications, along with the level designers responsible for mission construction. Consider: adding, building up, removing, or minimizing elements such as enemies, hazards, explosions, or moving geometry. The idea is to evolve the event to bring it closer the intensity and timing targets from the Intensity & Pacing Plan. The pacing trends also must be maintained.

The event ratings for an in-game level will also need to be compared to the ratings for the other levels that fit around it in the campaign. You will need to ensure that the first key action event of the level is larger in intensity from the one in the level before it -- and that the last action event of the level is larger in intensity from the level before (Figure 8 above).

Using the design specification changes, create and prioritize a change list of tasks and communicate those to the producer, project manager, and appropriate lead for each level.

Once deep into production (say 80-90%), then any functioning Dynamic Difficulty system(s) can be leveraged to ratchet up the intensity further for the highest action points and to lower it during the intensity lulls for finer adjustment controls to the intensity of each event.

Allow the level designers enough scheduled time to complete another full iteration to all the levels based on the change-list of tasks.

Repeat this review, adjustment, and iteration cycle throughout production (at least every 4-6 weeks) and even into post-production for final tuning. It is imperative that these reviews continue throughout production so the team can track their progress towards their Intensity & Pacing Plan target trends and to avoid a costly and time-intensive overhaul near the end of production that might otherwise likely result in costly throw-away work.

Every three or four months, the levels should also be reviewed by the whole design team and ideally even shown to the whole development team in order to allow everyone to give feedback, and to ensure the range of viewers and opinions is as wide as possible. This also helps gain buy-in from the entire team and lets them feel as if everyone has contributed to the creative process from brainstorming to Final.

Conclusion

Ultimately, the value of building an up-front structure to the game campaign will always produce a more satisfactory result in significantly less time, with less work thrown away. Once a game team has completed a game using such a structure and realized the quality outcome, there will be zero desire to return to the on-the-fly archaic methods of the past.

With such a structure in place, the quality of the experience will increase significantly. Yet there will still be an ample creative space in which the level designers and/or world builders can operate.

This refutes a common fear among the uninitiated who are resistant to such a process change. The structure imposes high level events and timing only and leaves the bulk of the details up to the level designers and/or mission scripters in both the design and implementation stages.

It is also imperative that all the level designers, world builders, and/or mission scripters who are responsible for building gameplay be a part of the creative team that lays out the Intensity & Pacing Plan. They can then become completely familiar and invested in the process and feel a vested interest in its success.

Ideally, these individuals should not be assigned responsibility and ownership over specific missions/levels until after that Plan has been completed. This will give everyone ample opportunity to contribute to the Big Picture of how all the levels/missions best fit together to satisfy the player. It will also limit any early resistance to on-paper changes (the fastest and lowest cost changes that can be made with respect to level/mission layout).

Once this structural overview of all the levels is laid out (in other words, the Intensity & Pacing Plan is in place) the team can logically move on to actual level design. They can start to work out the remaining medium-and low-level details of the level. This portion is often best achieved with small implementation teams that are entirely focused on the details of the execution.

With the key story, mission, and environmental events in place ,a level designer/world builder/scripter can start to flesh out the rest of the level to fill in the areas between the key gameplay action and story plot events.

They can utilize reusable generic action portions (e.g. varying groups of enemies or hazards) and/or by varying the level layout using the theme park design and urban planning rules alluded to above (in other words, those laid out in Brian Upton's excellent GDC 2007 lecture on Narrative Landscapes).

After the structured Intensity & Pacing Plan is in place and before actual level production begins, there is still ample room for further pre-production planning for the level designers and/or world builders.

Thoughout of the scope of this article, the best proven level pre-production process I have come across is that described by Michael Stuart Licht in his article, An Architect's Perspective on Level Design Pre-Production, so I would suggest adopting many of the elements he proposes (especially those on spatial layout) and integrating them with the staggered events that we have already planned out.

As an industry we can only hope to deliver the knock-out intensity and incredible satisfaction of one of the best seasons of Lost, Prison Break or 24 -- once design teams and leaders have fully embraced well-structured level design and production processes in the same way Hollywood has for years.

The pre-planning of a carefully structured Intensity & Pacing Plan for all the environments/levels (both as individuals and as a whole) and a follow-through of that process with the creation of a gameplay-focused progression plan will always produce a much more engaging result.

Because processes ensure that the pacing, intensity, and challenge progressively increase throughout the campaign, the results are sure to be video game gold. There should be higher quality, higher schedule predictability, less risk, and faster level production. In the end, these benefits will also translate into significantly greater efficiency, through less brutal throwaway of expensive and man-power intensive level production content.

So why does building an up-front Intensity & Pacing Plan make level production faster, cheaper and more predictable? Because it is always exponentially cheaper to iterate level/mission planning on paper than in the tools and editors. In addition, by going through the process of building this plan, you have created a target that you can aim for, like a beacon in the distance that vastly narrows and hones the creative focus of the team.

With a target to aim for, you are more likely to end up with an extremely satisfying experience, which otherwise only rarely happens after many rounds of time-intensive and expensive rework in the on-the-fly level design methods of the past.

If the creative team and leads build and embrace these structured planning, review and iteration design processes, the magic is sure to come. The end results will speak for themselves in terms of reviews, fan appeal, and sales.

It is time for teams to stop resisting level process evolution and adopt the more efficient and higher-quality pacing methods of Hollywood. It is up to our creative leaders and team leads whether they want to improve their likelihood for quality and success.

But also, it is up to all the members of the team to encourage and support such moves -- to facilitate these successful structured creative processes and to allow them to flourish in an industry where they have been strangely absent for too long.

Return to the full version of this article with comments
Copyright � 2009 Think Services