Alan Wake 2 is easily one of the most beautiful (from a technical perspective - IMO also from an artistic one but that's a 100% subjective thing) games ever released and it still looks good even when you turn all the settings down, which is a true achievement. It's hard to make a game scale down to older hardware while still looking good.
Though as commentators like Digital Foundry have noted, the game runs really badly if your GPU doesn't support Mesh Shaders (they mention the use of those for culling in the article). Mesh Shaders in this case enable a lot of really smart culling and dynamic level of detail so that things like coffee cups or tires can be perfectly round without having the 'every NPC has 10k-poly teeth in their mouth' problem that's currently sabotaging Cities Skylines 2's performance, and this is one of the big advantages offered by Unreal 5's Nanite.
You can draw a pretty clean line from the incredible Second Reality demo[1] in 1993 to Alan Wake 2 thirty years later.
As I've commented before on HN, honestly I could write a small book about these Finnish demoscene gods.
Sad to see an otherwise good comment end with a misconception. The problems with the performance is much grander than just "teeth rendered but not visible" (https://blog.paavo.me/cities-skylines-2-performance/), although I guess it's a illustrative point. Missing LODs and lack of culling are the grander issues.
That is the exact start of the sentence of which you quoted the end of:
> Mesh Shaders in this case enable a lot of really smart culling and dynamic level of detail
Isn’t CD Projekt Red moving to UE5 for future titles? Shame, Cyberpunk was gorgeous - would have loved a multiplayer game with that engine.
(Don't ask me for a source, I don't remember where I saw that info)
Have they pulled a No Man's Sky, since, or something?
this strikes me as an extremely silly statement. Their engine doesn't seem to have multiplayer.
CDPR has 1236 employees.
DICE (makers of Frostbite) has 714 employees.
Epic has 2200 employees (before the recent layoffs).
all numbers from Wikipedia.
Frostbite had roughly 300 employees when I joined 2 years ago.
I just don't see it. Looks like RE2make to me. And as far as art direction Dishonored 2/DOTO are dramatically better looking
Don't get me wrong, the engine is incredible and the visuals and systems are some of the best we have seen.
Personally I prefer snappy movement i.e. when I press left the screen character moves left immediately. A more realistic looking animation system introduces a delay while you wait for the feet animation to "catch up" to player input.
They could alter the animation such that feet don't slide across the ground and keep responsive movement. The result would be a worse quality animation, because the movement of the legs would not appear to be pushing the rest of the body around. Instead, it would look more like the feet are following the rest of the body retroactively, while holding onto the ground.
A good example of this is Factorio's spidertron. When the spidertron moves, the legs follow with a walking motion that perfectly tracks the ground below. In this case, it's a great-looking tradeoff, probably because there are so many legs, and not much animation done to the body itself.
It appears much more believable to me than the character pathing demo, where the character moonwalks 1/2 the time. The entire pedal structure is extremely stiff compared to how humans move (feet, thigh and torso can all turn nearly 90 degrees, but they hardly turn at all in the demo). The other demos are better, but their bodies still appear stiff, like they are suffering from hernia.
I do too, but also think it greatly depends on the game. For example, Hollow Knight designed to have snappy response to player input from the start and I loved it. In RDR2, I find the floaty behaviour adds another layer of realism.
In GTA V characters have two different animation modes, the realistic one based on Euphoria when using the third-person camera and the "do what the darned keys are saying right now" when using first-person. Always seemed like a sensible compromise to me, though first-person movement is particularly snappy and direct, more so than pretty much any other FPS, which typically still have some inertia.
I think the real issue is the difference between moving the entire character as an object, with the walking animation being supplementary to that vs the walking animation being central to the character whose object moves because of the animation.
But for a game like this, arcadey instant direction change type movement doesn't really seem warranted either.
However it’s a trade off of performance (the constant ray casting and IK solves aren’t free) but also responsiveness.
Many games opt for faster locomotion and responsiveness instead.
At the end of the day, it’s a technically solved issue and has been for years, but as with every single thing in game design, it’s a choice and trade off.
There are new approaches to generating realistic character motion being shown at SIGGRAPH every year, getting better and better, but the best ones are largely too expensive for games to adopt. Plus, for player-controlled characters, at a certain point you run into a fundamental tension between control responsiveness and physical realism of animations.
I recently saw a great article on the topic which even comes with a WebGL demo: https://theorangeduck.com/page/code-vs-data-driven-displacem...
People don't want the realistic, slow response. Normally people aren't aware of how early their body starts a movement, the conscious brain has the illusion of just having decided it but it's actually started much earlier.
Ultimately, it's more efficient from a cost and productivity system 98% of the time to use something off-the-shelf.
Agreed - there are so many off-the-shelf engines with big communities that you'd really have to need full control to do it yourself.
Example:
Our marketing folks would say the characters are more responsive and lifelike than ever before; our internal dev notes described it as "the characters won't bump or get stuck into objects in tight spaces".
realtor: cozy charmer!
For reference:
- Using an Emerging Language in Quantum Break (https://ubm-twvideo01.s3.amazonaws.com/o1/vault/gdceurope201... );
- DConf 2016: Quantum Break: AAA Gaming With Some D Code -- Ethan Watson (https://www.youtube.com/watch?v=7YjLW7anNfc).
Possible starting point if you want to dig for sources: https://forum.dlang.org/post/lymybpygzfalbdgoaizr@forum.dlan...
From the grapevine I heard there was one person that advocated for writing in D, though I'm not sure if they are still there or not any more.
Wonderful content. I hope we see more shops like this building content for ue5
I am always curious why so many games techs use Lua for scripting. Especially when designed from scratch
As such, we’ve been seeing it as a go to scripting environment for everything from Baldur’s Gate 1, to Warcraft III, to modern titles like Roblox.
It's fast (relative to other scripting languages), familiar, and stupidly easy to embed in a C++ program.
Although to support your point, pretty much every friendly NPC is stationary. Once the hostile zone in on you you’re not looking at how they walk - they’re shrouded in mist anyway.
IMO control was a tech demo for all of this and it also supports why the enemy count is much lower and framed as a horror story.
They should release the engine. It's a solid competitor to Unreal (more scalable due to ECS, easier to write with Luau)
It's a visually stunning engine but it's just too demanding IMO.
I just upgraded to a 4080 and it runs flawlessly though, I guess they were targeting 40-series cards for anything raytraced.
Seems Alan Wake still has a bit of foot sliding. I wonder if they're still essentially playing an animation while moving the character "object" in a direction.
Surely someone's already come up with a system where the character object isn't directly moved but instead moved in relation to an anchor, ie anchor foot to ground, taking a step naturally moves the character, now the foot in front anchors to the ground as the foot behind lifts.
Done this way it would be cool to simulate slippery surfaces along with the ml animation models to get interesting "scrabbling for grip" effects like you see in the BD robots whenever they're kicked/on ice.
Basically gameplay beats the animation in this case. On top of this a lot of the time you don't see your characters feet or if you do you are not really focusing on that during gameplay and won't notice them sliding.
Once we figure out how to render games in true photorealism, it's gonna be so interesting to see how often studios make these sorts of stylistic/practical choices with animation.
This isn’t some gotcha that the game devs didn’t notice or don’t know how to fix. It’s one they likely either decided had drawbacks or wasn’t an issue in practice.
I'd be tempted to work on game dev tooling one day if there were more remote work opportunities.
So do it. Then show it to your team and boss then boss's boss.
I like this tool from Nvidia [1] for exploring CNN feature maps. That's more in line with what I'm talking about but it's is an offline tool. I am imagining tools that provide this granularity of information about networks in real-time during training.
Disclosure: I work for nvidia
There's absolutely nothing wrong with the camera, it's just not to your taste. They've made zero risky or offensive choices with the camera.
> Off-centre camera systems, aka "over the shoulder" cameras make me motion sick. I need the player to be in the centre of the screen or everything feels lopsided.
This is a you problem, though, and an extremely rare problem overall (rare enough for me to literally never hear about it in 25+ years of gaming). I don't think it's reasonable for you to expect the world of game development to cater to your very niche, specific issue.
They did, it's just not one you like.
Most games that position the camera behind and centred have to zoom out or go weirdly high so you can still see what the character is looking at/heading towards. The off-centre camera allows you to be at the same height/viewpoint as the character and still see what's in front.
Look at the Red Dead Redemption 2 camera position when walking vs Alan Wake 2, totally different viewpoints. Both are different artistic choices, I'd say both work well for each game.
(also i yelled at more than one game that had an option to switch shoulder but not put the character at the center - if you are bothering to implement this why not also add the center option? :-P)
I gave Luau a try recently but the syntax for external type declarations is undocumented and unstable, which made it awkward to test properly, and the available VS Code extensions default to a Roblox environment until you mess with their settings.
So mixed feelings for now, I guess this is why they built their own tooling for it.
luau-lsp for example ships this globalTypes.d.lua file[1] for Roblox development and lets you configure your own.
[1]: https://github.com/JohnnyMorganz/luau-lsp/blob/4b7872349d9b8...
But the debugger is so good, I dont know any other debugger with such low overhead in any language
I wish Cyberpunk 2077 had this. It's a very immersive game but man did those weird/glitchy NPC movements take me out of it.
The lightning is really gorgeous in the game and as always I think raytracing is not really worth the performance impact if. The game looks almost the exact same without.
I wonder if you could tune the lightning in UE5 to look like that, I always wonder how much of a games looks is actually the engine and how much is just the art direction and the devs skills to get out of the engine what they want.
I think the game itself was rather disappointing. Too much walking sym, combat was downgraded and less fun, made worse by too few save points at times so you fear to die to not get annoyed to do sections over and over again that makes you hesitant to play on hardest difficulty.
Conclusion wall is way to basic and simple you can just quickly trail and error it. And it FORCES you to advance things on the wall that are clear anyway. I had it only one time where I left it for a while and when I look back into it she was like "we already got this" and put on a bunch of cards automatically, but generally speaking you need to advance it all the time. There is one situation, there is a place, where a reviewer passed by countless times and it turns out the key item you need ONLY appears when you advance the wall. Kind of stupid game design.
Its a movie game mostly, a "press button to advance story" game. The first was like that as well but I think it had better and more gameplay, better pacing that was actually fun.
Writing was obnoxiously pretentious and dragged on for too long. After 10 hours the same Alan Wake tropes get repeated over and over again and the game ends is a cliffhanger to shill some DLC. Really makes me loose my last bit of faith in the gamers, I already do not trust game "journalist" anymore for a long time but I at least could rely on userscores in the past, this game is like praised into high heavens by everyone for no reason. The best think it has to offer is graphics but they to not make for a good game.
There are large areas in the game the story does not even lead you though, just some stupid collectibles and these boxes with stuff you do not need. I did not even find the crossbow and did not care to look up where to find it, just had a bunch of arrows. I think the game got many people a boner because of the graphics and their brain shut off. The actual game is actually not that good.
And not to forget now how insanely woke the beginning of the game is, thanks to Sweet Baby Inc. they hired to rewrite the game with some woke trash dialog in the beginning and of course the race swap of Saga who was written a white Fin women. Her self made sweater does not fit her at all. 0.5% of Finnish women are black but hey DIVERSITY so they could just keep the story and hot two birds with once stone. My issue is not that she is black, my issue is that she was FORCED in by a company that should not exist, accompanied with woke cringe dialog. Because Remedy does not really care but is taken over by the woke mind virus. If they come up with a original story with a black lead they write themselves I am all for it, but not like this. But nobody cares and I am going to get downvoted for this, I know how it goes.
But I love the numerous references to Control and Max Payne, and how they've integrated it into the same universe. I can't wait for the Max Payne remakes, and hopefully the next installment in the series. I just wish it could be done without association with Rockstar. They ruined the experience of MP3, which was a solid game, but Rockstar's Social Club launcher is a garbage piece of software.
Some of your criticisms are fair I think, but this is actually deliberate and explained as the story continues. Saga is clearly doing more than "profiling".
* While it may keep their creative vision "pure", the lack of lower graphical settings makes it run pretty terribly on an RTX 3080, which is very frustrating. You can brute-force this problem away with DLSS, but I like native resolution and I'm sick of DLSS being used as a crutch.
* The difficulty is all over the place. "Story" mode is likely too easy, while "Normal" varies between being reasonable and damn near impossible. A weaker enemy can take 2 bullets in Story, and 10 in normal. That's frustrating when ammo is so tight early on. In general, the game lacks a good "progression" of difficulty. It's a roller-coaster instead of a curve.
* To add on to the difficulty, the mechanics do not feel as consistent in AW2 as AW1. Why do some enemy types have a shield you need to burn away, while others don't? (Despite looking similar) For others, it's unclear which ones will vanish with a little light and which ones require an intense light - and what exactly gets them to disappear. Why can't the flashlight have a bit more grace when a target leaves the crosshair for 0.1 seconds? Why do so many enemies have extendo-reach and teleporting? Why isn't dodging timing better telegraphed? Etc.
* The story, at least so far, hasn't hit the same highs for me. It feels a little like we're doing the same thing over again, and it doesn't hit as hard the second time after the reveals and twists from AW1 are already used up.
I much prefer setting DLSS on performance and playing with medium settings and raytracing on a 3090. The game looks way better that way.
After playing the first ‘boss’ in AW2, this is an amusing claim to read. The controls and environment interaction are maddening.
Remedy/Sam Lake dropped the ball.
Instead of making another game cut from the same brilliant Alan Wake cloth they botched it and took out most of what makes the original game great.
"ECS meant that iteration was quick because adding new or modifying existing systems or game objects was easy, and performance gains were clear when saving and loading the Case Board."
and there's an accompanying screenshot with perhaps 20 objects on a case board.
I scanned through the rest of the article with one question in mind: what the hell? ECS is not a solution to putting 20 photos on a case board. You're pushing 100Ks or Ms of objects and need performance and are willing to suffer for performance = ECS. Not this.
So, despite world class developer, I did not find the article credible. The main cause of the lack of credibility is the author of the article did not anticipate the reader glancing at the case board and the ECS claim and saying: bullshit. Alternatively, they did anticipate the claim, and were told to ship the article anyways.
We built a new Voxel-Based Character Control that enables smooth navigation in cramped, complex and dynamic environments; it makes character movement more natural and fluid.
Fluid movement can be a problem, but "Voxel" is not what you did here, it's a marketing term. Maybe start with how a 2d collider for a 2d game should perhaps be a sphere smaller than the character (not a square, they get stuck on corners), and then say sth interesting about how you solved 3d. I am fairly certain that "voxel" isn't what happened, esp with the prior of "ECS".
ECS also has a lot of overhead. You almost certainly don't need ECS, and whatever you're writing would almost certainly be much simpler without ECS. Use ECS because it solves a serious problem that you have.
Edit: And my seriously downvoted point was putting objects on a "Case Board" does NOT justify an ECS; it's a problem with a /lot/ of good and good enough solutions. Were there any problems that needed an ECS, or did the article just want to say ECS (and voxel) because it's hip?
