There’s a secret world hidden behind the windows of Chicago in Watch Dogs. In future Chicago, one computer program controls everything in the city, from the power grid to rocking horse rides. There are other strange things about the city as well, like how every window reflects some mysterious bizarro city made almost entirely of straight roads.
You’re standing on a hotel balcony with a big neon sign outside. You look in the window, and BAM! you’re looking into an intersection.
You’re standing in front of an office building by a statue, and you look in the window — BOOM — intersection.
Driving your car parallel to some shop windows, INTERSECTION
There’s even an alternate world inside some buildings (via reddit):
What’s this alternate reality chicago like? There aren’t any people there, so maybe it’s more peaceful. Or maybe that’s what it looks like after ctOS kills all the humans.
There is one hint — one of the studios that worked on the game? Ubisoft Reflections.
Comments
37 responses to “The Windows In Watch Dogs Look Into An Alternate Reality”
Staggering attention to detail.
First thing I noticed was this, they’re just textures
yeah i was playing some free roam online with a mate last night and found this, its very strange when you’re walking paste a building on the water front and ther’s a city reflecting back
Reminds me of the reflection in Borderlands 2. You see reflections of completely different areas
This, First thing I thought of was those puddles in sanctuary.
Isn’t this the norm for all videogames in the last decade?
Yep. Realtime reflections are still very costly to process, so most use a technique called reflection capture onto a cube map, that takes a 360 degree capture of reflections from a single location and bakes them for use in whatever location is set to use that capture object. You can have multiple capture objects as well, but it’s still baked information and not realtime.
Realtime reflective surfaces are too computationally expensive to have in outdoor areas with high environment detail.
Yeah but considering the budget put into this game, this is a disgusting oversight in itself.
I think it’s because it needs to run on Low end PC/PS3/360 not exclusively on High end PC/PS4/XBone.
Expensive in terms of processing power needed, not money.
GTA IV had better ones. So yeah, there is cause for disappointment.
http://youtu.be/JGRsrFdwb2c
It’s called an environment map, and this is a great example of it being used poorly.
Yeah, look at the puddles in BorderLands2 and other games, they usually reflect weird shit.
That main city in Borderlands 2 has a reelection of some red building that I swear do not exist – I tried looking for the source of the reflection until realising it was all a cheap trick!
This bugs me about games – Mario 64 was able to put a mirror in a room and give you a real time reflection of Mario (and Lakitu).
Why is this not possible on modern hardware.
Sometimes it’s incredibly bizarre like Assassins Creed Brotherhood shows your reflection in the water, but not puddles; Prototype 2 shows reflections in puddles but not the myriad of glass windows in NYC.
Given we have a tenfold increase in RAM and a boost in processing power you’d think these new consoles could tackle reflections with one hand tied behind their back (or a Kinect strapped to their arse) but I guess not 🙁
Realtime reflections in outdoor scenes effectively destroy rendering optimisations. Occlusion culling becomes all but worthless and you roughly triple the amount of geometry data that needs to be processed on each frame of rendering. Low complexity scenes like bathrooms and indoor areas can usually handle the extra processing required, but to do it outside they’d need to significantly reduce the complexity of the scene.
Which would you prefer? A scene that is three times as detailed but with baked reflections, or a scene that is one third as detailed but with realtime reflections?
I don’t mind baked reflections so much but it is immersion breaking to walk past a window and not see a reflection of your character, surely there’s a middle ground without breaking the polygon budget – like character reflections in front of the baked-in reflection
BTW I kind of got what you were talking about because I like to read DF in a vain attempt to grasp the science behind games but I still to this have NFI what ‘occlusion’ is!!!
Occlusion culling is a way of telling the engine not to render certain things because they’re occluded – blocked from sight by something else, or otherwise outside of the view frustum, so things behind the player don’t get rendered. Reflections change the rules, because suddenly a whole lot more things need to be calculated and rendered than just what is visible in the player’s direct view.
Ahaa – I believe how Crash Bandicoot had such amazing graphics for a PS1 game, you could never swing the camera around to see what was off-screen, so they could put a lot more into the image in front of you!
Also thank you because you’ve just explained to me why Lego Star Wars on the PS2 has real time reflections on many of the surfaces when so many Gen7 games don’t!
That’s two character models in a static room though. It probably also used a trick of mirroring control inputs on the other Mario model to do the “reflection”. Modern games may have more processing power available, but they have to do a lot more work as well meaning that adding the effort of calculating a high poly character (Player characters generally tend to have the most detail because you’re looking at them all the time) is just going to ruin the framerate. Sure, they could do a low poly model render for reflections but that’s render time that could be used on lighting, physics calculations, or other processing.
It’s essentially a balancing act, you only have so much time to render a frame and if certain things take too long, you either drop frames, or give less render time to various objects in the scene. Reflections are generally the first thing to go since you are increasing the amount of things to render.
Somebody didn’t test the cubemaps properly.
If Duke Nukem Forever could do it, why couldn’t Watch Dogs?
Because Duke Nukem looked like a dogs breakfast with very few moving parts.
That’s not really an answer now is it…
It’s a good answer. They had to animate like 10 polygons. Not too hard to reflect
I know its not really a technical answer, but it is an accurate one. There normally like what four enemies to kill on screen and then the model of the duke himself, when put together with the generally small scope of the areas (which also tend to be pretty flat) results in very different requirements on their respected engines.
Watch dog isn’t exactly the most detailed game ever, but it sure as hell has a lot more going on than Duke Nukem.
Duke Nukem Forever, not Duke 3D!
Thank you for that clarification… which was completely unnecessary as I am aware of what was being meant by that statement. I dunno if you remember the game all too well but I seem to remember the enemy spawn waves being capped at a hand full. I can remember no point in the entire game where I was over whelmed by numbers and the enemies are hardly highly technical models.
Forever isn’t the most technically advanced game, graphically speaking it was very average with there being very few models making up the landscapes assets (compared to most modern games that is). I don’t see why you would even use that example in the first place as it was panned for sub par graphics and that is without the open world scope of a game like Watch Dogs, GTA or Saints Row.
the only realtime reflections present in DNF weree the mirrors. THe background reflections were actually still pre baked environment maps but taken from that specific area. only player models were actually rendered in realtime…
Consider the following. If the buildings had realtime reflections, the would then be reflecting not just the background buildings, but also the buildings that THEY are reflecting… so its an endless cycle….
what they should have done is had a large variety of prebaked environment textures and used them a bit smarter so that it wasnt too obvious… being an open world game as well, unlike DNF, it would have been extremely difficult but not impossible to pull off a similar effect, again because of the large amount of character models, and 3d objects.
That’s alright though – IMO that’s a good compromise since you’ll ignore a blurry background if you can see your character reflected in glass or water.
The absurdity of this situation in generation 7 to me was summed up by Arkham Asylum (a game filled with non-reflective shiny puddles) making a Riddler puzzle out of stumbling across the lone mirror in the game in a dingy toilet block, the mirror itself giving you a very low res smudged reflection of Batman!
I must be drunk or something. Since when does a current gen game still botch a technique from the Dreamcast era, or even earlier?
realtime reflections are impossible. essentially it requires you to render the graphics multiple times, Seeing as you are having to re-render the image on every reflection, and reflection of a reflection.
considering every game built thee days pretty much stresses out the capabilities of gaming hardware to its limits, there simply isnt enough processing power to do it…. unless you are building a VERY simple game.
if you want to test this, use UDK or Hammer and build a square room equivalent of 50mx50m and fill it with a maze of mirrors…. run it then watch almost every computer grind to a halt.
They why not put a cut off? Even I know that if you have mirrors aimed at each other the reflections go on until infinity.
Also, this seems pretty rich that the next gen consoles were beating to their own drum about having more grunt only to pull nonsense like this.
As someone with a strange obsession with checking video game reflections, be it water, glass etc, this is a frustrating reality in many, many games.
I also study video game roofs and other curious graphical/tech flourishes. I have am eye for the obscure.
You think that’s weird? I discovered in Halo 1 that while the Magnum spread does reach to the end of the reticle, it’s only reaches as far as the reticle for when the weapon is zoomed in. Meaning, when zoomed out the spread doesn’t reach as far as the reticle.
They’re saving the real reflections for the WiiU version.
If you look closely, the scope reflections on CS:GO are pre-baked as well, and change with what section of the map you’re in. That’s why sometimes it flickers slightly when you move from area to another (it’s pretty suttle most of the time though).
A game where the reflections really are a different world and you have to go inside them or something would be awesome.
Should have just included frosted windows. Id rather see nothing than see something that disrupts immersion