Why shouldn't I permit changing the field of view, if the technology allows for it?

Question

This is less of a question aimed at developing a game and more about the technology behind it. A number of gaming journalists and pro gamers (both in the tournament and the home entertainment sector) often hammer on the matter that any first or third person game should have a way to change the field of view, and I agree with them. The main reason they mention this is because a pretty large number of players get motion-sickness related headaches and eyestrain if they play with an overly constrained field of view.

I can understand if an indie developer does not allow this due to budget constraints, because they need to add support to the engine for that. However, a number of developers use high profile third party engines like Frostbite, Unity, Unreal, or Cryengine, but don't offer an option for changing the field of view.

Why shouldn't I permit changing the field of view, if the technology allows for it?

Answer 1

4A Games, the developers of Metro: Last Light were receiving a lot of criticism for not having a FoV slider in their game. Community manager Maurice Tan defended their decision on the Steam forums as follows:

The main reason for maintaining a fixed FOV is because we have 3D elements like the watch and weapon ammo that need to remain visible.

In addition, all the game's first-person cut scenes and cinematics and each and every animation involving Artyom's hands - idle weapon animations, reloads, ladder climbing, melee attacks etc, - were created assuming the same, fixed field of view.

Changing the FOV would break all the cut-scenes and animations - you would be able to see inside Artyom's arms, or they would appear to float in the air in front of you. Or worse.

We had considered offering three FOV pre-sets, but this would still require significant work to re-do every animation, adjust the HUD and UI and other seemingly small but incredibly time consuming tasks.

Even with a wider but still fixed field of view, Artyom's hands would look too far away. We know - we tried.

Game performance is also tied to FOV - the amount of geometric detail we put in each scene has been partly determined by this set FOV, and setting a wider FOV would have a performance impact.

The main reason for not having a FoV slider was that they were going for a no-hud high-immersion approach were all information important for the player was displayed on 3d elements in their field of view. When your GUI is part of your render-scene, changing the FOV changes your GUI.

Something Tan didn't mention in above post but which likely also was a part of their decision is that the FOV is important for level-planning. Metro: Last Light is a horror-oriented game with a lot of jump-scares and other shock-moments. There are lots of situations where the player is distracted by something interesting so they look into one direction, and are then surprised by a sudden attack from the side. This shock-effect can only work when the level designers know the FOV of the player and so know what they can and can't see when looking from a certain point to a certain other point.

The same problem also appears in a different way when directing cutscenes. When you have high-immersion first person cutscenes, you likely want to have full control over the camera to be able to control what the player can and can't see. But when you don't know the FOV, directing the cinematography properly can be impossible. When the FOV of the player is higher than you expected, they will see objects they are not supposed to see yet. When the FOV is lower, the player won't see things they are supposed to see during a cutscene.

Answer 2

Not having a FOV slider in your game due to budget or technical reasons is a non argument. FOV is just one parameter in the creation of the view-projection matrix. After you have a proper matrix all the other calculations will follow by themselves. To word it in another way, making the FOV changeable means that only one line of code needs to be changed (in the camera class usually) and that an option needs to be added to the menu. The change does not propagate into other systems, they only 'see' the 4x4 matrix and don't care what the meaning of the values inside of it is.

public static Matrix CreatePerspectiveFieldOfView (
         float fieldOfView,
         float aspectRatio,
         float nearPlaneDistance,
         float farPlaneDistance
)

Matrix creation with field of view in XNA

However there are good reasons for not having a changeable FOV slider. Setting a higher FOV means that you can see more of the world on screen, this will have an impact on performance. Another reason is that in competitive play the player with a higher FOV setting will have an edge since that player can see more around him. Lastly it can ben an esthetical choice. A narrow FOV will give the player a different feeling than a wide FOV. Especially in shooters with narrow corridors a narrow FOV could add to the experience. A good example of this is the Call of Duty games where when you aim down your sight the FOV becomes smaller.

(Note that personally I do not believe that any of the reasons above should put a very low on the FOV, I just want to illustrate that in no way it is a technical problem.

Answer 3

One thing no-one else has mentioned is testing.

Every aspect of control you give a player is another variable you have to test for. For approximate full coverage, you effectively have to play through at each potential FoV, in case you get graphical issues only on one level under particular circumstances when you're in widest FoV or narrowest, or "when you saved the game at narrowest and open the save at widest".

This additional cost/delay may not be worth the additional value of FoV widening. Just because the engine supports it doesn't mean it will "just work" (although it does make it more likely). And just because the calculations should all work out doesn't mean that the user experience will be positive.

Answer 4

From a user perspective the main reason is that in multiplayer a wider FOV gives an unfair advantage.

Answer 5

As a developer, changing the field of view is a precarious decision to be made. This would also need to be a decision made early in game, in a multi-person studio it's important that a minimum frame rate and FOV are determined early on so that animations and Occlusion culling have something strict to work by. Of course Occlusion Culling can be based of numerous factors, not just Camera view.

I would say that the main reason I have never allowed FOV to be changed in game is that when this is the case gameplay generally is greatly altered, I've also known it to cause motion sickness. It will also mean that the camera is rendering more meaning a massive performance hit as games are generally optimised for their set FOV.

However, this is assuming I'm allowing you total control. There might be scope to allow the user to alter view cone from say 70-110 or something, but I don't know without testing that myself.

Answer 6

The others gave a bunch of good reasons - Aesthetic (cut scenes, desired look&feel of the game), Multiplayer balancing, Performance drop with wieder FOV, but I want to add some ideas to the technical aspect:

Especially the technical issues are probably more convoluted than many of you think, since a lot of processing is done on GraphicsBuffers and these Buffers usually have a fixed resolution - so a lot of Graphics detail will depend on the translation of these Buffers to the actual Geometry. For example many modern Shadow algorithms (apart from volumetric shadows) look more and more Blocky and Aliased with a higher FOV and the same Buffersize. The same is true for any algorithm which does ray-tracing like operations, because the FOV is wider, the single rays are spread in a wider arc, and more detail might get lost in approximations...