+Tim Duff I appreciate what you're saying. I did toy with the idea of going into detail about more than just the parts explained. But made the decision to just keep to one narrower topic. However in the future I may well do a part 2 going a little more in depth into what you're talking about.

+Tim Duff You have the right idea but it doesn't have to use any calculus. (Transitions:) Just give each massive object a volume of influence (like you said). Then use Mario's position and velocity vector to determine if you should transition between two overlapping volumes of influence. It's just about transitioning, not maintaining realistic gravity. You don't even need whole volumes, just invisible bridges between masses -- just the overlaps, not entire volumes. (Slopes:) Use two normals for each surface: one for rendering and one for gravity. A single center of mass would not be enough to figure out what the gravity normals should be. The right way to generate the gravity normals can be one of two options, but I've written enough already. (This video:) Even if it's not entirely right, the video isn't wrong to keep it simple and just discuss the surface normals.

+Tim Duff I was the head of Super Mario Galaxy 2.5, the SMG hacking project, and you're right -- in level files there are "areas" that determine the gravity. One of those gravity types probably works like described in this video, but there are also some built in ones with specific types of gravity (spherical, disc, etc.) to make things easier. Then, you can also determine a "priority" level for gravity that determines which gravity will take precedence over another, as well as the shape of the gravity, determined by an "area shape number".

glem3 Thanks for the confirmation! ICYMI, I invented a universal gravity system that allows many many thousands (perhaps in time, millions) of collide-able objects to act on each other gravitationally in real time.

The same applies to the tower on the first galaxy. The tower has some paths that get twisted in strange ways and the path to go up the tower. If you fall from edge of either a twisted path or by going up the tower instead of being subject to the center of mass of those objects the game detects you are no longer on the tower so you are drawn to the bottom of the tower.

+SwankyBox Ok now there's a problem! Why do I see you everywhere???

+Dick Gumshoe - You must have good taste in videos! I'm a total KZfaq nerd, so I tend to lurk in the shadows everywhere.

SwankyBox Well I kinda subbed to almost 400 people on KZfaq... Maybe that's why...

he's wrong

+ThunderfilmStudios That's a pretty good idea!

+Jesse Flello They render the screen to a 2D surface. Technically, you don't really walk "through" the portal. You get "teleported" (they ofcourse try to hide that) to the actual location.

+Mariano But then how does being inside a portal work? You CAN be partially in one portal, after all.

I think cameras are used to show what's through the portal as well as a teleportation hole

Yeah thought this video was obsolete once I saw that one haha

+SharabTi Oh that's inverse kinematics. It's the same technique used in Dark Souls to make it so the players feet on the ground always make sense, whether he's climbing a hill or fighting on stairs.

+Preston Hanford Thanks man, it's nice to hear I'm improving!

i know 5 years late, but this video is all guesswork and also completely wrong

That makes two of us. Everything makes sense here, and it may even help me get better at game deving

I have to thank them too! Glad you're liking it!

+Elise Spalding Hey thanks! I appreciate your faith! As for your suggestion, I'd imagine it's fairly similar to the Banjo video I did a while back. Having said that, I will look into it.

Oh yes I totally forgot about that... Offering suggestions is tricky!

+cRoxy86 For the view inside the portals, the only have to render a camera view on a place (2D surface). Try imagining the portal as a flat TV. To have two half of an object at different places, you just have to render a copy on each location. Now the only part I don't know is how to render only half an object.

+NSNick Camera's are pretty darn complicated, and not very easy to figure out as every game seems to have a different method! However if I can find something episode worthy at some point, I will definitely cover it.

+Power Max That was my first thought too, but it makes irregularly shaped planets impossible (or at least more difficult).

well thats because its simply true. Here is the object you're talking about embed.gyazo.com/c58265ca25a29e086d1dc4f2178d1798.png

How is the gravity for the question mark planet calculated?

+crispy_doodles It's saved in the header of the Game Boy ROM.

+Chadunda Well I try to keep each video to one specific mechanic, simply because people tend to like smaller digestible chunks. I may well cover this in a part 2 the future. Glad you're liking it!

+Michael Green Dragon (MKCoolKid) Puzzle Plank Galaxy theme

It's from Galaxy 2, the wood galaxies (I don't remember their names)

LB I already answered them with the exact answer. Why would you give him a vague answer after I already said the exact answer?!

Twilight Yoshi Your comment was not there when I replied, I loaded the comments section before you replied. It doesn't update in real time.

Annoying Sabaris is that supposed to be offensive?

+Mario Kart I've read this somewhere. It's stored on the carts, if a cart didn't store a border then there were a couple default borders that the player could choose from.

+JumpCutThoughts I was cramming for about 15 hours straight to get it done, so I flew off the rader somewhat! Glad you did though, both Galaxy games are fantastic: and damn their soundtracks are epic!

+Akaji Blubb Thanks mate! I've just looked at it was featured on Kotaku Aus. I may well write them though, thanks!

I think covering it in a video would be cool but I'll take the liberty of being a show off and tell you here. when each portal is fired it maps itself to the polygon displayed over the crosshair, say a wall. Using this map the game places the corresponding texture of the portal on top of this mapped texture creating the illusion the portal is tearing through space. For production (and I'd imagine debug reasons) the shape of the portal can only map itself to one polygon at a time which is one reason why all the levels are made of flat surfaces, as the devs want to give the player the freedom to place portals wherever they please. (also curves look bad in the source engine) the second thing they do is assign the coordinate of the placement of each portal in a data buffer and set a script so that when an object in source travels through the one of the portal textures (blue) it is transported to the other one (orange). As the source engine is set up to store all objects in memory so long as they are in the same level this means there are zero load times when going or taking objects through the portal.

+Dan August For the look, it's simply putting a camera on each end of the portals and film the other side in relation to the angle between your eyes and the wall where the portal is. Then you render the image as a 2D image to the wall. In other words, each portal are basically flat screens streaming a video from each others' view. As for going through them, I assume they use a copy of the object going through it so you get two objects at different locations, then turn invisible each part that goes through the wall to give the illusion of half the same object at each end.

+Doroc0 I've befriended the guy who programmed the graphics in Splatoon (Ken Watanabe). He told me that the environment is made up of many textures. The ink is stored as a texture for each team in each world and is mapped under the diffuse texture. It uses a technique called a stencil buffer, where it makes the diffuse texture become transparent at the texture co-ordinates where the ink had landed.

+James Moore OK, thanks.

+SuperZalkin Naughty Dog explained it in a video somewhere on youtube.

+Khast P Tigris I am not sure about the Gameboy Advance, but in order to solve the problem on the Snes, they installed an additional chip in the cartridge to render 3D. That is how Starfox came to be. The same process could have been done on the GBA, but I can't confirm.

+Khast P Tigris Whoa, I didn't know the GBA was capable of that. It looks like both the GBA and SNES did in fact have primitive 3D capabilities. Of course, for the game to run at full speed, actual 3D had to be used as little as possible. I can't figure out exactly how they did the buildings in Crazy Taxi, but if you look at SNES games like Super Mario Kart, Star Fox, and Pilotwings, you can see that the "far distance" backgrounds are just scrolling 2D images. You might be surprised to hear that the ground is 2D as well as far as the game is concerned: the SNES has a special hardware feature allowing it to easily display "tilted" 2D images. (See the intro to Illusion of Gaia for a great example of the effect being used pretty much by itself. The entirety of Pilotwings is a more complex example.) This "tilt" feature is part of what's called "mode 7" on both GBA and SNES. (Another thing to note is that the GBA and SNES have extremely similar hardware - Nintendo basically just miniaturized it, and this is also why there were so many SNES games ported to GBA.) Another trick that I spotted in Crazy Taxi is passing off 2D images as 3D objects. If you watch the taxi closely, you can see that when it turns, the game simply switches between still images of it at different angles. The whole time, it's technically 2D, but there's enough different "pre-rendered" images of it that turning appears reasonably smooth. This form of fake 3D is called "billboarding", because in the 3D world, the 2D image is always upright and facing the viewer. The billboarding effect can be seen very clearly in Super Mario Kart, because there are far fewer "angles" for the karts. Pipes sticking out of the track only have one angle, because they're cylindrical and therefore can look the same from all sides. Billboarding can also be seen in real 3D games - the trees in Super Mario 64 are billboarded, allowing it to display fitting/convincing trees (always a tricky feat) with only two polygons. In fact, just like the pipes in Super Mario Kart, the trees also have only one angle - they're treated as "cylindrical" too.

+Myniced I know enough about it, that it is all software driven. The GBA wasn't able to do the enhancement chips quite in the same way, and that would draw the cost up quite a bit. I'm more interested in knowing how the software based engine was able to do so much, what shortcuts they had to do to produce the effect. And what did they have do to get the huge game map into RAM. While the GBA had a few full 3D titles that used the same engine, Crazy Taxy was one of the best looking, and has nearly everything that the Dreamcast version had. There had to be some level of wizardry going on to get it working.

Real 3d on the GBA was done with software rendering. Basically, the CPU renders the scene and the frames get rendered out to the GPU as bitmap images.

+Khast P Tigris Like QWERTYCommander said, the GBA has no hardware support for 3D primitives. It's conceptually the same as writing parts of OpenGL or DirectX for the GBA but ofcourse without hardware support.

+Gabe Nathan My best guess is that the entire area is a trigger that tells the camera how it should interact, once you leave the trigger the camera goes back to normal, unless you entered a different trigger, which would tell your camera different ways to behave again. Something like that.

I'm more talking about games whose cameras are always changing position in order to get the best angle

+Gabe Nathan First, you can put collisions on the wall to avoid having the camera going through them. That's a start. You can also tell the camera to switch angles if the distances between the character, the walls and the camera are too shorts. But to get the best view of an area, you can put constraints on where the camera can and can't be so it only travels on locations where the developers judged to be the best views. Those cameras could basically be on rails, going along as you progress.

I know for a fact that camera systems exist without rails. Take the camera on the morph ball in Metroid Prime

Gabe Nathan I was simply naming one system. The morph ball have the camera following you from a certain vertical angle and distance and will adapt to the walls' collisions. It also checked whether there's an object blocking the view between itself and the ball and change position from that. Finally, to keep a smooth flow, the camera allows the ball to move a certain range of distance and angle from the camera before it reposition itself. For example, if you turn around in a small circle, the camera won't move, but as soon as you get too far or too close, it will.