This idea of having idealised curves that get sampled at discrete points in time reminds me somewhat of Functional Reactive Programming and its predecessor, Functional Reactive Animation. Were any of the architects familiar with that world when they planned all of this out, or was it an independent discovery?
There was at least one FRP fan on the team, but it was not a meaningful source of inspiration.
IIRC the motivating goals were:
* Client-server architecture (not p2p)
* gracefully degrade under constrained bandwidth
* replay support
The in-game replay was a by-product we simply thought was cool and novel. I’ve yet to see anything else like it!
The system also served as the basis for save/load. I think if you had a full replay you could resume from any point. Pretty sure we shipped that feature, although it’s been so long I forget what shipped and what was merely an idea!
Hey! First I loved PA. It was such a great game. Second, I’m curious how you simulated combat. Was it simply every every server tick you updated all units or was it something smarter?
IIRC the server simulation ran at 10Hz. I think the goal was to update every unit every tick. But there may have been some round robin style updates as you get to late game. Pathfinding may have had some limits and deferred requests.
Nothing particularly clever that I can recall. But it’s been quite a few years!
Thanks for your work! I played many a PA LAN party back in the day and SC was my favorite RTS during high school. PA is one of the few RTS where PVE could have fun scenarios.
I'm simply not smart enough to manage 8 frontlines. I'm excited for Industrial Annihilation and hope you're involved with it.
This is my favorite technical writeup. I think I asked you before on here if you still had info about how PA's pathfinding or csg geometry worked, those were two other cool things I remember from the OG tech demo.
I don't know anything about the CSG. I didn't start the pathfinding system,m but I did a lot of work on it. I could maybe someday turn that into a blog post. But I don't have source access anymore.
The tldr of pathfinding is voxels and flow fields.
The interpolator for the current value is generic. It could represent any function, including curves like parabolas. I don't remember whether or not we actually used that capability though.
I was expecting early access steam, not a literal preorder page with no release date set.
I feel like that's the wrong way to do this. You either use preorder with a scheduled full release window, or steam early access. Taking money as preorders for a steam early access with pretty much no release window... That feels more like vaporware to me.
The initial roll out for Industrial Annihilation was really janky. With Planetary Annihilation, they did the Kickstarter thing and it worked pretty well (if I recall correctly, it was the highest grossing Kickstarter of all time at that point). With Industrial Annihilation, they revealed it as an investment platform for wealthy stockholders. Absolutely no options for people who wanted to support it but were either not ridiculously wealthy or just didn't want to own stock in the company.
Interesting. I'm concerned that I'll have to try and play satisfactory while I'm under artillery fire, and their last foray into the genre had too much frustration compared to fun, but assuming that those systems stay relatively light this could be exactly what I'm craving. Annihilation and it's successors have always had a special place in the RTS genre.
Well given the top-down point-click interface, I think you mean factorio, which already has near-constant enemy attacks, bases, etc but they are low-tech zergs.
Factorio is a fantastic game, so I'm optimistic that copying that model might be good-ish.
I was addicted to following the dev updates when this was in original development. Was a huge fan of RTS at the time and found the space theme and the early concepts of slamming astroids into your enemies to be amazing.
Hey, just dropping in to say: the moment I saw the demo of the feature, I said to myself, "that's gonna be due to a good underlying data structure and this article is going to be about that data structure".
Was not disappoint.
This is a lesson I learned way too late in IT: Get the data structure right and interesting features come at reduced cost.
Opinion time! Planetary Annihilation was close to being a good game but was held back by its ambition to be a novel engine. The one that got my nerves on edge was they have a zoom mechanic where the strategic map was a weird shape - there were multiple worlds and each world was spherical rather than a 2d grid. It did terrible things to the gameplay because suddenly orienting the map was a challenge. It was really well implemented and impressive visually; but unfortunately it made it remarkably easy to get lost. It was a great case study for why so many wargames go with the unimaginative "rectangular table" setup.
Things like that coloured my view of PA. They have an interesting engine but the consensus back when it was released seemed to be that they didn't manage an engaging game. I got the impression the engine was consuming too much of their attention.
> Planetary Annihilation was close to being a good game
PA: Titans is 85% positive on Steam with over 10,000 reviews. It's objectively a good game!
> I got the impression the engine was consuming too much of their attention.
I think you have the order of operations wrong.
Having multiple, spherical battlefields was arguably a mistake. It absolutely limited the game's audience and made controlling things much more confusing and difficult.
But you have to remember this game started as a Kickstarter. We didn't have a prototype, just a vision. The game was sold on multiple planets of combat. There was no way we could have possibly changed the game from multiple spheres to a flat 2D plane after raising $2.2 million on KS.
In fact I'd argue there were multiple "bad ideas" that we were forced to ship because the community decided they were PROMISED those features. Sometimes those promises were things like the core vision sold on KS. But sometimes the features were small things casually mentioned as ideas being considered. Managing community expectations is very difficult!
PA 1.0 was a bit a flop for reasons. But PA: Titans was quite good. The game was shipped on a custom engine in just 2 years, or 3 years for Titans. In hindsight it's one of the most impressive team achievements of my career. I don't know how we did it.
I also still play TA from time to time. It is kind of annoying that if your screen resolution is big enough to display the entire map at once that it crashes.
1) Games are meant to trigger the imagination which is fun. In this sense, Planetary Annihilation looks successful.
2) Games are a computer interface that should be fun to interact with. In this sense, Planetary Annihilation looks like an extreme failure. Managing battles on multiple planets in real-time sounds incredibly frustrating and the often overlooked aspect of actually interacting with the game would be horrible.
I always found StarCraft limiting, it didn't excite my imagination. I couldn't understand why Starcraft 2 was (and still is?) the most successful RTS game. You fight with 10s of units and you can't zoom out; where's the spectacle to trigger my imagination? Then I realized StarCraft was an game about user interface, and the user interface is pretty good. Select units, click them around, watch and click the mini-map, keep it simple and responsive. It's under appreciated how fun it is to simply select a few units and click them across the screen.
StarCraft isn't my favorite, but the subtle user interactions in the game are limited but not frustrating. Whereas Planetary Annihilation always looked like a game where user interactions have fewer limits, but a lot of frustration.
One thing that SC2 gets right which is endlessly frustrating when trying to play most other RTS games is how the units react to repeated similar commands: if you click a destination and then adjust that command slightly, the units don't miss a beat in their movement, and just adjust their course slightly (SC2 being the kind of game that it is, SC2 players will often just spam that command with only minor adjustments). In most other RTs games I've played, that second command will often recalculate the movement of the selected units sufficiently differently that it at least slows them down, and spam clicking will just cause them to derp out completely, making unit movement a frustrating task.
(To say nothing of Supcom's annoying habit of units setting off in the opposite direction. Especially irritating for your ACU when trying to beat a hasty retreat, which arguably shouldn't even have the vehicle movement mechanics which causes this)
"where's the spectacle to trigger my imagination" Please see Supreme Commander. Its a regular 3D rts map so its easy to navigate, you get ALL the zoom, and most stuff gets bigger (in size & qty) as the match goes on till either someone wins or no one in the map can mentally manage it all!
I've played it. I prefer the open-source game Beyond All Reason which runs better on my computer.
And yeah, that was an implicit comparison I was making, I would look at games like SupCom and wonder why StarCraft was more popular, and thus my previous comment.
It was definitely lacking in depth and strategy. Battles took so much effort to manage, that for normal skill level players the game was more about turtling up your planet and hitting an exponential growth curve so that you were basically invincible, then using a super weapon to end the stalemate. Either that or ending the game within the first few minutes with a well-optimized rush.
However, I don't think I have seen a dev team flex their engine-developing muscles as much since in a game. It's clear that a team of Uber-talented (pun intended) people with a passion for their craft set out to use a lot of cool technologies, and I think that's worth something by itself.
> each world was spherical (...) suddenly orienting the map was a challenge
The other downside was the fact that it was possible to assault arbitrary points - e.g. drop units from orbit in the middle of enemy base. It required much more awareness about the whole map. In 2D game one could mostly focus on front lines and choke points.
I can imagine the discussions to be had in the dev-kitchen - between we go natural on zoom out - or we start a projection war right there, right then..
https://store.steampowered.com/app/109700/Achron/
It wasn't a very good game, but it used a similar time-scrubbing mechanic to support time travel.
You could send your units back in time to harass the enemy before they built defences.
Time jumps cost energy, and paradoxes were resolved at wavefronts that moved forward in time faster than 1s per s.