This website views at its very best in Firefox web browser, and is not available in a mobile version.

      Copyright © All Rights Reserved | Built by Serif Templates

Parchment Background Image for RCT3 FAQ: Master Maps, POI, and Guest AI on

Park Guests

If guests are given park maps they can better find their way around our parks. As maps and umbrellas are available both at the information booths and at the umbrella stalls, it appears that placing either stall in your park will be an asset to your guests. As far as guests navigating their master map (the map of the park drawn for guests by RCT3 coding in the game engine) one would assume that for guests one simply needs to place paths and that’s that, that they’ll find their way around your park and enjoy the eats, drink the drinks, and ride the rides.

Janitors simply need to clear up paths and water gardens. Mechanics only need inspect and repair. How the park guest interacts with his master map is much more complicated as it revolves around a huge number of factors.

During the development of RCT3 guest behaviour needed to be created in a way that would enable potentially thousands of guests to interact in a park containing other guests & their groups, paths and their add-ons, stalls, staff, ride & coaster statistics, ride position information, scenery, weather, etc. all of which had to be graphically rendered into some semblance of reality which included movement, perspective, distance, lighting & shadows, day & night, antialiasing & transparency, flags & texture styles, colors, draw distances, and so on - in other words, the 3-D people living in RCT3 Land had to look and behave like we would when we ourselves are at an amusement park. At the same time Atari had to come up with a solution for the appearance of the game and the behaviour of the life in it in a way that did not require that everyone own a server in order to run RCT3. It seems Atari’s answer to the maximum amount of guest behaviour with the minimum drain on system resources was to create Group Artificial Intelligence.

According to Atari the Group AI focuses on the group guest that is doing the worst with their park visit. What this means is that each group is attempting to meet the needs of the guest in their group who is having the worst time of it at your park. Let’s say we have a group of four guests: Guest A, Guest B, Guest C, and Guest D. Of that group Guest D is having the worst time of it at your park while Guest C is not doing much better, Guests A, B, and C become focused on satisfying the needs of Guest D. Once the needs of Guest D have been met Guest D becomes happier with their park visit after which their group then focuses on the needs of Guest C.  After Guest C becomes happier the group's focus changes again, and so on.

To real people like us who live in the real world this seems rather clumsy and simplified but Atari’s solution makes sense:

one group of four guests each focused on Guest D = four sets of focus at one focus per guest.

In real life we’d have

one group of four guests each focused on themselves and on their three companions = sixteen sets of focus at four focus’ per guest.

Sixteen sets of focus may not seem much for your CPU to process for just four guests but imagine multiplying that by 1,000 guests, i.e., 4,000 sets of guest focus, among their respective groups.  In addition to all that each guest needs to react and interact with the points made in the third paragraph above, plus each individual guest needs to consider these factors:


How many rides, coasters, stalls, and other attractions there are in your park.


The age, condition, statistics, and cost of the items in 1.


The amount of scenery and the consistency of themeing.


The rating of your park (from which is subtracted for untidiness, crowdedness, unreliability and lack of safety).


The happiness/unhappiness of other park guests, the rate at which new guests arrive at your park, and the reason the guests in your park are leaving.

The value of your park and its attraction value to new guests will increase with each additional ride, coaster, stall, and other attraction you add to your park. Conversely, park guests’ needs and attitudes will degrade over time. In RCT2 there was a trainer with which we could make all rides and stalls new. There is nothing similar for RCT3 so every two or three game years, about the same length of time it took rides to get old in RCT2, it’s a good idea to close your RCT3 park & everything in it and take your attendance level back down to zero guests. After you’ve done that you can re-open everything and watch park attendance surge to near pre-closure levels with fresh guests who now see your park with new eyes.

For our park guest, the individual for whom we are building our park and for whom we need to do so much to ensure his happiness, it makes sense that he needs to do so much more than simply get around on his master map. Remarkably, in spite of the virtual nature of the guests Atari has done an excellent job in getting to us some semblance of reality with park guests in RCT3.  Regrettably it’s because of RCT3’s Group AI that we don’t see those great long queues filled with guests in RCT3 like we did in RCT2. The AI is also the reason RCT3 guests spend most of their time joining their group or just standing around in our parks waiting for their group.

Due to the dozens of factors that make up the thoughts, happiness, and activity of each guest there are some situations where guests occasionally have navigation problems, for example, when we place stalls caddy-corner with both stalls having access on the same path tile. These guests will arrive at one of the two stalls and get caught up in a loop glitch of endlessly going from one stall to the other. Oddly, after X number of guests get caught up in such a loop (usually around five guests), each arriving guest that gets caught up in the same glitch will enable one of the existing guests caught in that loop to exit the glitch.

Other challenges with guest behaviour sometimes become evident when we close a ride but the guests remain seated in the ride cars. Some of us have also observed that occasionally the in-game elevator and double-deck observation tower which aren’t broken down at the time will open and close their doors and operate normally with no guests entering or leaving the car. None of our staff have ever noticed the in-game giant Ferris wheel to properly board guests.  At times a guest will leave the path, walk into an enclosure a few tiles away and stop just inside the enclosure, lost. Occasionally in a huge busy park it’s awhile before we discover the dozens of guests all gathered on one path tile that are lost in spite of the fact we’ve built our paths properly. Guests will occasionally walk along a path and seemingly without reason will change direction several times resulting in their walking back and forth along the same path. Guests who are hungry, thirsty, and need the bathroom will ride the same ride over and over without attending to their bodily needs in spite of the fact there is food, drink, and a bathroom available between their ride's exit and its queue line.

However, along with this we have park guests in RCT3 that can tell us how much they like our pool complex, who can admire the view & comment on our scenery, who can take photos of our park, who can find amusement in our zoo animals, who can watch us build our roller coasters, who can hold hands, and who have realistic expressions on their faces indicating at a glance how they feel at that particular moment. And while park guests in RCT2 all looked like clones, RCT3 park guests have individuality in their appearance as they realistically congregate & sport around our parks.

Path Construction and POI

Up until recently this author has placed very little value on how the park path tiles have been placed in relation to guests not getting lost apart from a belief that the paths need to be in a continuous circuit. That was only part way correct.

Here is a link to an excellent tutorial by Fossil who has thoroughly analysed points of intersection. He has taken the building of park paths and honed it to a fine art. For a great number of years while this article was at RCTMart it seemed to apply to RCT3 paths but now Fossil is on AngelFire it appears to be related to RCT2.

Notwithstanding this, the content of Fossil's article is far more than we could tell you about the building of park paths.

  Fossil's Pathing Systems

Please turn to Page 2