NPC AI Survey, Optimization; & DLC 1 Level Deep-dive
Attention Officers,
Welcome to the 82nd edition of our Ready or Not Development Briefing, January 31st, 2025.
Welcome to the first dev briefing of the new year! Today we’re highlighting the new NPC AI sentiment survey, discussing a newer approach we're taking to optimization, and offering a deep-dive breakdown to how we create RoN levels-- utilizing examples from the Home Invasion DLC.
We are re-gathering and doubling down our efforts to continue development on Ready or Not after returning back from our winter break about a month ago. Right now, we’re nearing the end of a flurry of careful planning and prototyping.
The breakdown of our level design process in this dev briefing will help contextualize just some of the work that goes into each content release of Ready or Not. Meanwhile, this comes in theme with our community mapping contest closing shortly (check our official Discord for more details)!
Speaking of which, much of our community will be happy to hear that some of the next upcoming yet-to-be-announced levels we're working on will be on the smaller-scale end. This will help balance out the grand scale of some of our more recent levels (i.e. large levels like Leviathan, 3 Letter Triad).
Note: These development briefings serve to keep you in the loop about parts of our ongoing support for Ready or Not, however the briefings do not encompass everything that we’re working on at a given moment. Please keep in mind that the content in our development briefings may still be a work in progress and subject to change.
[h2]NPC AI Sentiment Survey [/h2]
As you may have already noticed, we just put out an NPC AI sentiment survey yesterday! Please fill it out here:
https://forms.gle/DsVQT2x76qEpvWvj8
We are grateful to see that we received an astonishing number of responses already, and we'd love for yours to be one of them-- every bit of feedback counts and helps us interpret your experiences!
If you are not sure what to put for a given question, chances are it is not a required question and you can skip it! If for some reason you fill out a question that was required but you did not have an answer to, please feel free to explain in the written response sections.
One of the ongoing goals we have for the game is the continued refinement of SWAT AI, Suspect AI, and Civilian AI. This survey will help us gather more concrete data on the current state of AI following Dark Waters' release, which included bug fixes and improvements across several areas of AI that we want to analyze the impact of and double-check for unintended issues.
One of the big focuses on this survey is breaking down the specific functions of AI that affect how the player perceives their level of difficulty, and how the AI behaviors can lead to frustration or satisfaction. Although gathering feedback from across our social platforms and feedback channels (like the one in our official Discord) are helpful, this allows us to target specific areas of the AI and gather that info in one place.
[h2]Optimization Efforts [/h2]
We are now intertwining the system called "Unreal Insights" deeply into our ongoing performance optimization efforts to make the game run as smoothly as possible. This system shows the performance metrics of specific game events and processes, allowing us to pinpoint specific problem areas with greater efficiency.
Already we have seen noticeable early performance improvements by integrating this into our workflow, and as our technical artists and engineers continue to implement its use we expect the benefits to become even greater.
[Image below: General example of the Unreal Insights task graph analysis view-port, from the UE5 Developer Page]
[h2]Ready or Not Level Creation Deep-dive [/h2]
The wonderful members of our level design team (namely Tisa for this one!) chronicled their efforts creating levels for Ready or Not and DLC 1 in specific.
The result is this piece that shares our philosophies and processes on how we go about creating levels for Ready or Not in general, using new BTS photos of Narcos development throughout the stage explanation and at the end using a full example of Lawmaker from DLC 1 in particular:
Creating a level in Ready or Not follows a multi-stage pass process from idea inception to final polish.
[Image below: The levels covered (Narcos and Lawmaker) are influenced by the feedback y'all gave in our Home Invasion map poll]
Stage 1 (Planning)
It all starts with production or deciding what level to make. At VOID every member gets a chance to pitch in an idea for what they find would be an exciting level to make, keeping in mind inspiration from community feedback.
Everybody involved writes their ideas in the form of a game design document (GDD) that gets added to the stack and from there, based on the target theme for the DLC and the fun + innovative factor, gets picked to be the level we’re gonna work on.
Our narrative designers and concept artists start to sketch out various ideas for the level and the lead level designer will create a Design brief that each level designer will receive to start the prototype pass.
[Image below: A sketch of the layout of Narcos, along with doorways and other key elements]
Stage 2 ( Prototype)
Level design gets into the first so-called ‘prototype pass.’ We quickly make several versions of the level to explore what’s fun about it and how we want to approach creating the alpha version. Since Ready or not is a realistic game with heavy themes, we like to ground our levels on real life as well.
That means we do extensive research, gather real life references, if we get the chance, go to real life locations, make sure the architecture fits the target level for this area etc. At this stage, it is not unheard off to scrap the level if we find it to be lacking in any way. Prototype passes are there for this particular reason, like sketches for artists, level designers do quick block-outs they can toss if it doesn’t meet the quality bar we hope to achieve for our levels.
[Image below: Layout of Narcos based on research of similar neighborhood development design, made to look like a GPS view]
Once each level designer has created what they wanted, they pass it on to be reviewed by our lead level designer. The lead level designer will then do a review on each prototype and together with the level designer for that level, decide which of these do they want to use for the Alpha pass.
Note that very often each prototype has something of value, so the Alpha may become the joined version of all these prototypes. The best parts of all if you will.
Stage 3 (Alpha)
After the best prototype has been chosen, we start to go more in detail with the blockouts, adding navigation, doors, AI, cover and making the level playable. Once we have the level in a pretty good shape we invite our combat designer who gets in and starts tweaking the AI to get the desired combat behavior. After they’re done our internal QA also starts doing runs on the level to flush out any important issues at this stage.
[Image below: Blockout version of Narcos]
When we have a strong Alpha candidate, each level designer writes a level handover document consisting of all the important information, screenshots, references and reasoning for each part of the level. That level and the handover document then gets passed on to the Environment artist assigned for that level. This is also the time the rest of the team starts working on the level to get all the features out, like audio team recording all the voicelines needed, character art team creating the suspects and civilians etc.
Throughout this time the level designer is in close contact to all teams working on the level in case anybody has any questions, and to make sure the design intent doesn’t get lost in the process of going through other departments.
Stage 4 (Beta)
After the Environment pass has been completed and the level gets its shine, level designers return to the level to do multiple passes.
- Collision pass – The level designer will create blocking volumes around the entire level preventing players/AI from going in non playable places, optimizing movement.
- Visibility pass – The level designer will add in visibility blocker to any cracks in geometry to prevent AI from shooting players in unfair ways.
- Navigation pass ― Level designers will at this point polish up navigation and make sure there is no navigation in a non playable space. Our system relies heavily on navigation, spawns stack up, cover, clear, rooms and other points on the navigation itself so for optimization and to avoid any AI bugs this step is quite important.
- Door pass – The level designer will make sure all doors and it’s features are working correctly for the player and the swat AI.
- Cover pass – The level designer will make sure all cover points are generated and are in the right positions for AI to use them.
- Interactable pass – The level designer will make sure all custom logic still works and is implemented correctly.
[Image below: Red blockout triangles in Narcos indicate the sightlines from one doorway to another. The goal is to allow the player to easily see each new doorway without exposing them to too much danger of being shot in the process]
Stage 5 (Content Complete)
In this stage level designer will communicate with all other teams to make sure all the features for this level have successfully been implemented and are working as intended. Throughout this process the level is heavily tested by QA, making sure it’s ready for public release.
[h3]DLC 1 Lawmaker Design Dev Commentary[/h3]
Lawmaker Layout
For Lawmaker we wanted to create a mansion, big rooms and wide corridors.
Most mansions have really unique layouts and weird corners you won’t find in ordinary houses, and that is something we found extremely fun and unique in a sense that we don’t have many yet in Ready or Not.
Even though mansions do have multiple weird angles and unique layouts, they are very structurally accurate and if you look at it from the top down there is almost always a good architectural plan for the building.
The Lawmaker simplified plan was 2 rectangles intersecting with the third rectangle at 45 degree angle. This gave the building those nice connecting 45 degree corners and unique look we wanted for the rooms and corridors, making it look and feel quite satisfying when clearing out rooms.
[Image below: Lawmaker basic layout concept draft]
To make the mansion feel more manageable to clear we made the rooms a lot bigger than our usual RoN rooms and capitalized on all the affordances from that.
- By making everything a bit wider we gave players and AI more maneuvering possibilities, preventing any bumping into players and getting stuck on small passages.
- Even though the mansion does look huge, by making the rooms bigger the number of rooms is pretty much the same as in all our maps, removing the frustration of having too many doors to breach and staying in line with our Ready or Not formula.
- And lastly, bigger rooms and hallways was something we wanted to target for this level as that is something that is unique and true to real life and we know our players enjoy realism and a good challenge.
Lawmaker Gameplay:We also have to consider what kind of gameplay we want to have in these spaces before we even start creating anything. For Lawmaker we knew we had a home invasion\hostage situation and we wanted to develop systems for our AI to act in correspondence to that.
The invaders are a group of eco-terrorists, not necessarily the hardest group to fight. Other than the two boss characters the group doesn’t have much weapon training, and are quite surprised they actually managed to pull this off and enter this house. So behavior like hesitation, cowardice and running away is expected.
So how did we plan to make this group feel like a threat and make them harder to capture?
Aside from increased hostage taking where you really have to watch not to shoot the civilian, we implemented our zones system across multiple floors.
What that means is the suspects will fire at players if they spot them, and if given a chance they will either try to flank the players or retreat to another zone in hopes of surprising the player again.
In the case of Lawmaker, if they manage to get away from the player they will move up a floor, so if the players fail to catch them individually on the bottom floors, they will have groups of suspects to deal with on the top floor. Which also brings us to the next thing we wanted to add.
In real life suspects will likely move all civilians to the same room where they hold them hostage.
We wanted to add that to the mix as well, making it really important to know where SWAT officers land their shots if the room is full of suspects and civilians.
Knowing all this, we come back to the layout.
[Image below: Zone layout for Lawmaker]
We knew we wanted flanking behavior from our AI, and we wanted to restrain most of the fighting to the house itself and very little to the outside areas. So beside bigger looping paths, we implemented lots of mini loops in the house which the AI can use to do just that. This gives a lot of value to players who wedge the door behind them or watch their back.
Since Lawmaker is a 3 story mansion we also had to consider flanking and AI zone retreating between multiple floors. For that reason we made sure each wing and the center has a staircase connecting the floors.
All these decisions also increase replayability as each playthrough players can choose a different path to take based on the active situation. To increase replayability even more we gave players 3 unique player starts they can use to approach the house. Each with its own challenges.
- Main entrance is the safest to approach the building but places players in the center of the map, needing to decide which side to clear first.
- Side entrance gives players the option to enter through the center or the back of the west wing, but players are exposed to windows on the west wing while approaching.
- Back entrance is giving the players most possibilities of entering, center, west or east wing, but it is the most open as players are exposed to windows from all of these.
Following the narrative for Lawmaker we wanted to create objectives and interactables to tell the story and help players connect it.
Once the eco-terrorists entered the mansion owned by Sven Anderson Lincoln, owner of the oil company they were protesting about, they started destroying the property, throwing oil and paint on priceless paintings, shooting sculptures, beating and rounding up staff that were in the house, taking them hostage etc. Sven’s wife and children were in the upper floors and managed to get to a hidden panic room, while Sven was caught somewhere in the house.
So we knew we wanted “Locate the family in the panic room” to be one of our objectives. We created the logic for that event making sure we tackle a few issues:
- The children must not be harmed in any way in gameplay
- Multiple locations for the panic rooms must be set so that each playthrough it’s different
- We were a bit limited to what SWAT voicelines were available to say, so had to take that into considerations
- Active panic room needs to catch the players attention when they find it
First we placed the wife and two children inside the panic room, and they would never come out as it is likely SWAT would prefer them to be safe until the team has cleared the entire property.
We then created a LIVE camera feed in the security room that also shows other cameras around the property. Some have been spray painted over and destroyed, making the one from the active panic room the only LIVE feed.
To ensure players do get the right picture we added an interact button they can press which switches all the security monitors to the panic room feed and its location in the house. To draw players' attention to the security desk and that interactive button we placed flashing red lights on the active monitors. That made sure we tackled the first issue.
For the second issue we placed two locations for the panic rooms that get randomly selected when the players start the match. That also gets connected to our security room interactable so players can always see which panic room is active even before they reach it.
We solved the third issue by simply having the mother do all the talking once players interact with the panic room. Lastly, we placed red blinking lights outside of the panic room to indicate to players that this panic room is active. We also made sure these lights can be seen from hallways so that players don’t miss them, removing frustration in searching for them.
Another objective we wanted is to locate Sven, the owner of the house that is somewhere inside the house. For that one, similar to our panic room logic, we made sure Sven is in 5 different locations at the start of the map. It’s also worth noting that as an AI he can also flee on his own, making the search for him unique every time we play.
Conclusion
Please make sure you fill out what you can in the NPC AI sentiment survey, it is very helpful in contextualizing future considerations to our AI systems! Last but not least, here's a final closing remark on our level creation process:
With every map we create we want to make sure we bring something new and interesting to the table, new kind of layouts, new kind of mood, new kind of gameplay, new challenges etc. Creating these is really something every level designer at VOID enjoys and puts their heart into, and we LOVE watching how our players tackle each new challenge we present them.
It is worth noting even though our maps are playtested by our QA team we take feedback from watching our players tackle each area in the level and if we feel certain parts need adjusting we are not strangers to addressing these issues in the updates. Our final goal is to create a unique and enjoyable experience for our players.
This concludes our 82nd development briefing, thank you for joining us. Be sure to fall in next time for more development news!
Follow Ready or Not on Steam here.
Our other links: Discord, X, YouTube, Instagram, Facebook, Spotify.
[h3]Stack up and clear out.
VOID Interactive [/h3]
