22 replies to this topic

[James Miller]

I know there's been threads about this strewn all over the net, with the question "How long does it take to learn C++", with the answer pretty much being "You never stop learning", however, I have a few questions I'll put out here as specifically as I can, as I'm curious.

1: How long would it take the average person learn C++ to the point where they can confidently and reasonably use all C++ commands/functions etc (even if they're not quite at the John Carmack skill level). I'm just talking, to the point where they could create anything they're asked, even if it's primitive.

 

2: Continuing on from Q1, how long would it take our average Joe to go from making their first Hello World, to a primitive, but functional 3d engine, something on par with say, the original QUAKE or Half-Life 1 engine in terms of functionality, graphics etc (nothing next gen, just simple but functional).

[Álvaro]

I am not sure the average person can learn C++ to the point where they can create anything they're asked. Your average Joe will never deliver a 3D engine nearly as good as Quake or Half-Life.

Edited by Álvaro, 25 January 2013 - 03:29 PM.

[ic0de]

I know there's been threads about this strewn all over the net, with the question "How long does it take to learn C++", with the answer pretty much being "You never stop learning", however, I have a few questions I'll put out here as specifically as I can, as I'm curious.

1: How long would it take the average person learn C++ to the point where they can confidently and reasonably use all C++ commands/functions etc (even if they're not quite at the John Carmack skill level). I'm just talking, to the point where they could create anything they're asked, even if it's primitive.
 
2: Continuing on from Q1, how long would it take our average Joe to go from making their first Hello World, to a primitive, but functional 3d engine, something on par with say, the original QUAKE or Half-Life 1 engine in terms of functionality, graphics etc (nothing next gen, just simple but functional).

No one will be able to approximate the amount of time required for any given project. It entirely depends on the person. My suggestion is just try it and see. Just don't let people tell you that "C++ is too hard". You don't need to learn everything in C++ to make a 3d engine seeing as it's been done in C many times before.

Edited by ic0de, 25 January 2013 - 03:36 PM.

[James Miller]

I know there's been threads about this strewn all over the net, with the question "How long does it take to learn C++", with the answer pretty much being "You never stop learning", however, I have a few questions I'll put out here as specifically as I can, as I'm curious.

1: How long would it take the average person learn C++ to the point where they can confidently and reasonably use all C++ commands/functions etc (even if they're not quite at the John Carmack skill level). I'm just talking, to the point where they could create anything they're asked, even if it's primitive.
 
2: Continuing on from Q1, how long would it take our average Joe to go from making their first Hello World, to a primitive, but functional 3d engine, something on par with say, the original QUAKE or Half-Life 1 engine in terms of functionality, graphics etc (nothing next gen, just simple but functional).

No one will be able to approximate the amount of time required for any given project. It entirely depends on the person. My suggestion is just try it and see. Just don't let people tell you that "C++ is too hard". You don't need to learn everything in C++ to make a 3d engine seeing as it's been done in C many times before.

 

Fair enough, thanks for clarifying that.

 

With that being said, I have another annoying question in relation to the above, is creating a simple 3d game engine something that could be achieved with a good solid year of studying (or less?) or is it one of those things that takes many many many years to learn (like 5-10 years?)

Edited by James Miller, 25 January 2013 - 03:35 PM.

[ic0de]

I know there's been threads about this strewn all over the net, with the question "How long does it take to learn C++", with the answer pretty much being "You never stop learning", however, I have a few questions I'll put out here as specifically as I can, as I'm curious.

1: How long would it take the average person learn C++ to the point where they can confidently and reasonably use all C++ commands/functions etc (even if they're not quite at the John Carmack skill level). I'm just talking, to the point where they could create anything they're asked, even if it's primitive.
 
2: Continuing on from Q1, how long would it take our average Joe to go from making their first Hello World, to a primitive, but functional 3d engine, something on par with say, the original QUAKE or Half-Life 1 engine in terms of functionality, graphics etc (nothing next gen, just simple but functional).

No one will be able to approximate the amount of time required for any given project. It entirely depends on the person. My suggestion is just try it and see. Just don't let people tell you that "C++ is too hard". You don't need to learn everything in C++ to make a 3d engine seeing as it's been done in C many times before.

 
Fair enough, thanks for clarifying that.
 
With that being said, I have another annoying question in relation to the above, is creating a simple 3d game engine something that could be achieved with a good solid year of studying (or less?) or is it one of those things that takes many many many years to learn (like 5-10 years?)

To tell you how long it took me specifically to become good at C++ it was about a year and a half but you will end up with numerous completed projects before then. To make a 3d engine it takes as long as it takes, I'm almost done mine and I've been working on it for a little over a year. To answer your question I would say it is possible but obviously the more experience you have the more solid your engine is likely to be.

Edited by ic0de, 25 January 2013 - 03:39 PM.

[swiftcoder]

How long would it take the average person learn C++ to the point where they can confidently and reasonably use all C++ commands/functions etc

3-5 years, or more, depending on how much effort you expend on learning every nook and cranny of the language. That said, read on...
 

I'm just talking, to the point where they could create anything they're asked, even if it's primitive.

That's a completely different question. You don't need to know the whole of C++ to write programs with it - I'll go out on a limb and say that most people who use C++ don't know everything about it.
 
6 months to a year ought to do it, if you are fairly diligent about continually writing programs to expand your abilities.
 

how long would it take our average Joe... a primitive, but functional 3d engine

A lot longer. You need to learn a 3D API and a bunch of math, in addition to your programming language.

That said, write games, not engines is required reading. A "3D engine" should probably never be your ultimate goal, and it certainly shouldn't be your first goal.

[Dan Mayor]

I would like to chime in and address the second part of the question from my view point.  Programming itself is more of a proper implementation and understanding of the way that a computer thinks.  C++ or any language for that matter is little more than an instruction set that allows you to implement commands that the computer (or console) can act upon.  What I see quite a lot of are people that read a book and they know some commands in the language but they don't always make the underlying connection of what the command is doing.  For example (all be it possibly poor as I'm just making a quick reference example here..)  In C++ many people learn to declare and reserve dynamic memory through new or memset methods, but very few actually understand what this is doing.  The common and quick answer everyone will spurt out from memory is "It makes a dynamic variable in memory duh!"....

 

What is a dynamic variable in memory?   Something that can be constructed and disposed as needed...  Ok but what does that mean? ...  It means that the computer is reserving a range of memory addresses in RAM to store a variable that could be of the size requested.  And when you release or dispose of this dynamic memory the systems reservation of these memory sectors are made available for anything and everything else to reserve and use later.  How about using Direct X for another quick example.  There are many many many programmers out there that know how to use Direct X and if you ask them what it does they will answer (It renders graphics to the screen).  Which actually is a bit of a false statement to be made.  Direct X is an API layer that exposes command sets that allow you to activate capabilities of the video card's GPU hardware, shaders and on card ram (as well as many other features).

 

I'm sorry it seems vague but the point I am trying to make here is that learning the language is actually very easy.  "cout prints text to the console", "if compares values"  "for loops until it's end condition is met".  The thing that makes a difference between someone being able to play with the language (and maybe even effectively produce some software) and the guy that can reasonably do whatever is asked of him is more the underlying understanding of the language actually causing physical things to happen on the computer and even some understanding of how the hardware itself works.  This is the level at which you become more of a problem solver where in you know a solution you are aiming for, you can quickly think of the logic behind it (and in gaming / engine writing what the most effective use of the hardware may be).

 

So to the answers, how fast can the average joe learn C++?  I don't know, how fast can you memorize a few hundred commands?  How long will it take to elevate above that to the point where someone tells you "Make a thing to do this" and you can figure it out?  That depends more on how long it takes you to make the connections between the commands your using and what they are actually doing not what the results they produce are.  Especially when your getting into something like a game engine that requires advanced knowledge of hardware capabilities, hardware api's and advanced performance oriented design of the code.  It's not just knowing how to use DirectX to render some geometry you need to make it "smart" enough to still work well when your users use it wrong.

 

This is why you can't really get a definitive answer on a time frame to expect.  Everyone learns differently, many never make the underlying connections but they still get results, these people normally do better using someone else's engine and they can eventually get pretty good at it.  However should you ask them to build an engine you'll start finding they will tell you "Um I don't re invent the wheel" or "Why? <insert favorite engine here> already does it good enough".  This isn't laziness, more so this is actually the more experienced answer where in they may not know it but they are really saying "I don't understand how these things work I just know how to use them".

 

Just as a little more information as it would pertain to making an engine, here's the first few questions you should be asking yourself before writing a single line of code on an engine.  "What is the most common video card type that PC gamers are using?"  "What shader models are most common?"  "What are the differences between the various sharder models?"  "How GPU expensive are these various shader technologies on the most common video cards?"  "What Direct X API's impose the greatest bottlenecks when attempting to trigger GPU calculations?"  "What is the fastest and most efficient way to transmit the shader apps in to the pipeline?"  "How can I manage my code to maintain lower bandwidth even when my users are trying to render a bajillion polies?"

 

I'm sure that many people will chime in after me on this and say that none of this matters.  Compare their engine to Unreal.  Why is Unreal so much more powerful?  Is it because they use something that you don't?  No it's because they address issues like these (and I do mean LIKE these maybe not particularly these).  So all in all learning to "Program" and learning to use a language are different things all together and their importance varies based on what you are attempting to accomplish.  When your not talking about games performance actually goes right out the window in favor of getting the expected result.  Thus making it not so important that you understand how the language works just how to get what you want.  However when you get into engines an the like the question is different.  How can I leverage the hardware to get the result I am looking for.  Amount of time for people to make these connections and actually start doing the proper research vary more than snow flakes, you are the only person that can have any idea how long it would take you to memorize the functions, learn how they make hardware level things happen and then making the connection between those and learning to solve problems.  Basically I guess I'm saying learning the language is learning to get results using what is available.  Learning to program is learning to solve problems.  It's like detailing a car versus building a car, you can give anyone a sponge and a bucket and they can make your car look shiney, that doesn't mean they can actually make it go faster.  So how long would it take the guy at the car wash to learn to make your car go faster?  How long will it take billy anybody to learn to wash a car OR make it go fast?

[ic0de]

That said, write games, not engines is required reading. A "3D engine" should probably never be your ultimate goal, and it certainly shouldn't be your first goal.

This article seems to have become a bit of a thought terminating cleche around here. I don't think people shouldn't make engines but they should obviously be developed concurrently with a game. I don't think this article should be used to smack people in the face who want to write an engine and I don't think it should be put on any kind of game development pedestal.

Edited by ic0de, 25 January 2013 - 04:33 PM.

[Aeramor]

Learning all of the commands/syntax in C++ is barely 10% of the battle. Really it comes down to logic and math, while there are certainly different paradigms used in different languages the process is mostly the same. Still C++ requires you to think about more of these simultaneously as it doesn't handle any of it for you. 

 

Think of it like learning a musical instrument. You can learn all the keys/strings etc but you still won't be Aerosmith overnight.

[swiftcoder]

I don't think this article should be used to smack people in the face who want to write an engine and I don't think it should be put on any kind of game development pedestal.

That's not what I use it for (nor what the article actually says, if you ignore your knee-jerk reaction to the title and really read it).

 

I'm merely pointing out a trap a lot of newcomers fall into, and sure, if the OP is purely looking for a learning exercise, then no one will stop him, but it never hurts to inform. If the OP had asked about "a primitive, but functional 3d game, something on par with say, the original QUAKE or Half-Life 1" I wouldn't have mentioned it at all...

[Dan Mayor]

That's not what I use it for (nor what the article actually says, if you ignore your knee-jerk reaction to the title and really read it).

 

I don't think this article should be used to smack people in the face who want to write an engine and I don't think it should be put on any kind of game development pedestal.

 

I'm merely pointing out a trap a lot of newcomers fall into, and sure, if the OP is purely looking for a learning exercise, then no one will stop him, but it never hurts to inform. If the OP had asked about "a primitive, but functional 3d game, something on par with say, the original QUAKE or Half-Life 1" I wouldn't have mentioned it at all...

 

I have to agree with you on this one swift coder.  People need to be made aware of complications and heart aches of what they are getting in to and they need to weigh those equally or even more so critically than the idea that things are easy.  I seem to have been on a rampage of "Game Development is hard!" today (probably just gripes coming up from a difficult week) but it's still something VERY valid to say.

 

Engines power games yes, games can get very complex and time consuming yes.  This is my big argument when it comes to building games and engines together.  Simply put engines are extremely complicated maybe even the most complicated things that are ever coded.  Games can also be complicated and in some cases just as complicated as building the engine they are powered by.  This is why to me it's NEVER a good idea to do both.  If you want to make an engine just to learn things and your making a game using it just to see how it all ties together, great!  If you want to make an engine that other people may actually use is another story all together.  You will need to invest LOT's of time on making it high performance, easy and providing a wide range of tools.

 

Game content, development, coding and logic are also complicated and at times can take quite a large time investment.  Same as with the engine if your just doing the game to learn how things work and your coupling it with the development of a simple engine GREAT!  However if you want to get other people to actually play your game (not just test it for you but actually play it) you need to put in some effort and make it worth playing (keep in mind there is a lot of competition out there and tons of really good games).

 

So here's where I jump in on the argument between making engines and making games.  Both are required and both have nearly equal importance to each other.  That is to say without better engines we don't get better games.  Without better games we don't advance the engines to power them.  However when we get into the world where everyone builds their own engine and their own game at the same time we completely negate the first statement of this paragraph.  You build the engine to power your game and you build your game to run well on your engine and neither receive the proper attention that they deserve!  So my opinion on this whole thing is the more people try to do everything the more nothing gets advanced.  To me the argument isn't saying you can't build an engine it's that it requires just as much if not more dedication than the game itself.  As independent and or a small game studios we simply can't afford to take a year and a half to make a half way decent engine followed by another year and a half to complete the game itself and by the time we hit the shelves we're 2 - 3 years behind all the competition in performance and quality!  Even the larger studios that do build their own in house engines to go with their games have two different departments.  One focused entirely on the engine itself and another focused entirely on the game itself.

 

So if it's just for hobby then writing an engine is perfectly fine.  If there's any concern for profit you need to compete with the big boys who are dumping money into the development of their engines by the truck load.  Your one or two people working in your spare time you'r not likely to ever get on par with them.  However making a game with two or three people can be profitable...  Make a few games and get some money coming in first then take your experience and knowledge from making games and apply it to making the engine that powers them.  When you do make that leap into making the engines make sure that those involved are dedicated to just that and not the both at the same time (otherwise you will end up cutting corners and sacrificing performance and tools that will lead to loss of money and less chance or sales).

[Anri]

Its not a question of "how long" but "how to go about it".  You'll find your own pace, and if you are smart about which areas you choose to tackle first, it might only take two or three years.

 

The "Quake" engine is certainly not a terrible long-term goal, but I'd recommend looking at the goal of making a Wolfenstein3D style ray-caster engine before even thinking about a full 3D engine. All you need is the ability to draw a pixel on the screen and obtain keyboard(possibly mouse) input.  The maths and programming skill required is slim and is a much more manageable project.

 

I'd learn(in order) C++, Software Development/Engineering, Maths(Trig & Algebra) and WindowsAPI(if you are using Windows).

[James Miller]

Its not a question of "how long" but "how to go about it".  You'll find your own pace, and if you are smart about which areas you choose to tackle first, it might only take two or three years.

 

The "Quake" engine is certainly not a terrible long-term goal, but I'd recommend looking at the goal of making a Wolfenstein3D style ray-caster engine before even thinking about a full 3D engine. All you need is the ability to draw a pixel on the screen and obtain keyboard(possibly mouse) input.  The maths and programming skill required is slim and is a much more manageable project.

 

I'd learn(in order) C++, Software Development/Engineering, Maths(Trig & Algebra) and WindowsAPI(if you are using Windows).

 

Thanks for the response, that's very useful to me and has helped me a lot.

 

I guess I messed up my initial terminology in my earlier posts, I guess I was, in my head, referring to a "Game engine", not a generic 3d engine. A game engine being something like Shadowcaster, Source Engine, etc.

But I'd love to try what you suggested. My first game goal would be to make the unfinished Wolfenstein 3d sequel "Rise of the Triad" -> http://www.3drealms.com/rott/originalspec.html

In looking at the source code for wolf3d, Carmack notes some interesting areas I'd love to explore, such as variable wall heights etc.

[alnite]

But I'd love to try what you suggested. My first game goal would be to make the unfinished Wolfenstein 3d sequel "Rise of the Triad" -> http://www.3drealms....iginalspec.html

In looking at the source code for wolf3d, Carmack notes some interesting areas I'd love to explore, such as variable wall heights etc.

 

I'm not sure where you are in your learning, but I think this is too much of a goal if you just started.  Here's my two cents about the whole questions about learning programming.

 

Learning programming up to the point of mastery can be divided into four major parts:

1. The first part for the Average Joe is to learn the actual programming.  For people who have never programmed in their life before, it actually is hard for them to grasp the concept of writing instructions.  Computers follow instructions down to the exact steps.  It doesn't question, it doesn't disobey.  So programmers, must write the instructions correctly, line by line, and this is hard for Joe.  If you ask Joe to write a code that "Ask for a number and print that number.", don't be surprised if Joe asked back "Shall I use for-loop?"  I have experienced this first-hand teaching newbies programming.
   
   Programming is a weird concept for Average Joe.  Being able to compose instructions together that make sense, and make up a usable program, actually takes time to learn and master.  Actually, I have observed that some Computer Science graduates can't still write a program from scratch.  They can read code, they can change code, but they can't write it from scratch.  Yes, the so-called programmers can't program.
    
2. Joe's next hurdle is to master programming languages.  This is all about "learning C++/Java/C/Python" and all that stuff.  Knowing the syntax, what they do, and when to use them is no easy feat.  This, by itself, may take a year or two for any programmer to master a language.  Over the course of Joe's career as a programmer, Joe will find himself learning multiple programming languages.
    
3. Then there's the third part, the library/API part.  Learning this is similar to learning a programming language.  As a matter of fact, some programming languages come with extensive range of API.  Some others (like C/C++) don't.  This is where Joe learns how to draw stuff on screen, how to play audio files, etc.
    
4. The fourth part is to master software architecture.  This require considerable amount of experience, and I'm pretty sure vary by person and experience.  I have programmed for more than 20 years, and it's only just recently that I got *it*.  By the time I got that "aha" moment, I have learned about 8 different programming languages, completed about a dozen personal games, and released about two-three commercial games.

Obviously, these four parts don't necessarily have to happen before one another.  They are all learned simultaneously.  You could go back to part 2 if you just picked up a new language, for example.

 

I think, only after you have mastered all four, of a particular programming language of a particular API, you are qualified to make an "engine".  Knowing C++ inside out doesn't not make you qualified to make an engine in Java.  I'm not saying you shouldn't make one.  I'm not stopping you from making one as long as it's for your own personal use and learning purposes.  It's a great learning experience.  Just keep in mind that whether it's a "game engine" or a "3d engine", making it is no easy task, especially for the Average Joe.

Edited by alnite, 25 January 2013 - 10:58 PM.

[James Miller]

But I'd love to try what you suggested. My first game goal would be to make the unfinished Wolfenstein 3d sequel "Rise of the Triad" -> http://www.3drealms....iginalspec.html

In looking at the source code for wolf3d, Carmack notes some interesting areas I'd love to explore, such as variable wall heights etc.

 

I'm not sure where you are in your learning, but I think this is too much of a goal if you just started.  Here's my two cents about the whole questions about learning programming.

 

Learning programming up to the point of mastery can be divided into four major parts:

1. The first part for the Average Joe is to learn the actual programming.  For people who have never programmed in their life before, it actually is hard for them to grasp the concept of writing instructions.  Computers follow instructions down to the exact steps.  It doesn't question, it doesn't disobey.  So programmers, must write the instructions correctly, line by line, and this is hard for Joe.  If you ask Joe to write a code that "Ask for a number and print that number.", don't be surprised if Joe asked back "Shall I use for-loop?"  I have experienced this first-hand teaching newbies programming.
   
   Programming is a weird concept for Average Joe.  Being able to compose instructions together that make sense, and make up a usable program, actually takes time to learn and master.  Actually, I have observed that some Computer Science graduates can't still write a program from scratch.  They can read code, they can change code, but they can't write it from scratch.  Yes, the so-called programmers can't program.
    
2. Joe's next hurdle is to master programming languages.  This is all about "learning C++/Java/C/Python" and all that stuff.  Knowing the syntax, what they do, and when to use them is no easy feat.  This, by itself, may take a year or two for any programmer to master a language.  Over the course of Joe's career as a programmer, Joe will find himself learning multiple programming languages.
    
3. Then there's the third part, the library/API part.  Learning this is similar to learning a programming language.  As a matter of fact, some programming languages come with extensive range of API.  Some others (like C/C++) don't.  This is where Joe learns how to draw stuff on screen, how to play audio files, etc.
    
4. The fourth part is to master software architecture.  This require considerable amount of experience, and I'm pretty sure vary by person and experience.  I have programmed for more than 20 years, and it's only just recently that I got *it*.  By the time I got that "aha" moment, I have learned about 8 different programming languages, completed about a dozen personal games, and released about two-three commercial games.

Obviously, these four parts don't necessarily have to happen before one another.  They are all learned simultaneously.  You could go back to part 2 if you just picked up a new language, for example.

 

I think, only after you have mastered all four, of a particular programming language of a particular API, you are qualified to make an "engine".  Knowing C++ inside out doesn't not make you qualified to make an engine in Java.  I'm not saying you shouldn't make one.  I'm not stopping you from making one as long as it's for your own personal use and learning purposes.  It's a great learning experience.  Just keep in mind that whether it's a "game engine" or a "3d engine", making it is no easy task, especially for the Average Joe.

 

As for prior knowledge, I've got a working knowledge and understanding of BASIC, and C. I designed and programmed the Command Line Interface of East-NSW TAFE college, and built a very very basic DOOM clone in QBASIC. The fourth image isn't my own work however, but that of Ken Silverman. I was toying around with his 3d engine written entirely in QBASIC which was quite fun to toy with, similar to the Wolf3d engine. Mind, the BASIC stuff like the Doom engine was done years ago when I was nearly a toddler so it's not fantastic, but I've re-invigorated my love for wanting to make an engine.

 

 

The API part is what I was curious about, that's what I always wondered about, as I never found anything in all of my C++ textbooks about how to actually *draw something* on the screen, even a single pixel. Regardless, I'm curious and eager to delve into it and see what I can come up with, but I'll be happy with getting one good solid program done, even if it's no more than a simple calculator. Progress motivates me, and getting started was always the hard part for me, but once I start making even a slight tinge of progress (ie, something actually works when I compile it!) then I tend to just steam along after that.

[swiftcoder]

The API part is what I was curious about, that's what I always wondered about, as I never found anything in all of my C++ textbooks about how to actually *draw something* on the screen, even a single pixel.

C++ is purely a programming language - it knows nothing about abstract constructs like 'screens' or 'pixels'.

 

You will need to use a library that exposes such functionality (something like SDL or pixeltoaster)

[James Miller]

The API part is what I was curious about, that's what I always wondered about, as I never found anything in all of my C++ textbooks about how to actually *draw something* on the screen, even a single pixel.

C++ is purely a programming language - it knows nothing about abstract constructs like 'screens' or 'pixels'.

 

You will need to use a library that exposes such functionality (something like SDL or pixeltoaster)

 

So what would've the early 3d game developers (ie again, Carmack working on Wolf3d) have used? Did they write their own libraries or were these kinds of resources available back then?

[Aliii]

If you really want to do it from zero, then it will be rather a long journey than a simple "project". If you have plenty of time and motivation then you`ll have nice results in 2-3 years(depending on how good you are at maths/3D maths.)

 

-You not just have to learn C++ ...but OOP prorgramming itself, lots of maths, 3D maths, physics, directX / openGL, (a sound API and networking ...optionally) etc.

If you dont wanna create your own 3D models ....just import them from an other program ...thats saves a lot of time. 

 

-It makes more sense to build up your engine incrementally and create actual games along the way, ....cause you can spend months with working on things that make your engine more solid but dont add anything to the graphics. .....it can be quite demotivating.

 

-And you need to have 2D projects too. ....with many things its better to have a good understanding of them in 2D before doing the dimension jump:)

 

....anyway Wolf 3D and Doom1 are not really 3D games, but 2D games with 3D(ish) graphics.

[swiftcoder]

So what would've the early 3d game developers (ie again, Carmack working on Wolf3d) have used? Did they write their own libraries or were these kinds of resources available back then?

Back in the days of DOS, the hardware was a lot simpler, and the OS didn't prevent user code from accessing the hardware directly. You could map the framebuffer into memory and write directly to it.

 

These days the hardware is infinitely more complex, and hundreds of processes may need to be drawing to the screen at the same time, so the OS and graphics drivers work very hard to insulate you from all that. You can't get any lower-level than the OS-specific drawing or 3D API - libraries like SDL just abstract across the many different OS-specific APIs that accomplish the same thing.

[James Miller]

If you really want to do it from zero, then it will be rather a long journey than a simple "project". If you have plenty of time and motivation then you`ll have nice results in 2-3 years(depending on how good you are at maths/3D maths.)

 

-You not just have to learn C++ ...but OOP prorgramming itself, lots of maths, 3D maths, physics, directX / openGL, (a sound API and networking ...optionally) etc.

If you dont wanna create your own 3D models ....just import them from an other program ...thats saves a lot of time. 

 

-It makes more sense to build up your engine incrementally and create actual games along the way, ....cause you can spend months with working on things that make your engine more solid but dont add anything to the graphics. .....it can be quite demotivating.

 

-And you need to have 2D projects too. ....with many things its better to have a good understanding of them in 2D before doing the dimension jump:)

 

....anyway Wolf 3D and Doom1 are not really 3D games, but 2D games with 3D(ish) graphics.

 

Sounds like an ideal plan, and yeah I'd begin with simple 2d engines, games etc at first, learn how it all slots together etc. I guess I wouldn't expect to be able to make a Wolf3d clone for a few years yet, but it's my goal. I dream of being able to make just some kind of generic 3d fps engine, even if it was never good enough to be sold or used commercially, just something to toy with and use to learn, but it's one of the things that's on the cards for the next decade of my life, or optimistically, 5 years.

 

As for 3d models, I should be able to build most of them myself, I've done a level design, 3d modeling and 3d animation before. I've done a bucketload of mod work over the last 10 years in Doom/2, Build engine, Source Engine and Unreal. Have also built a lot of special effects, so it would be fun to play around with some of these aspects from the coding side of things once I get the hang of it.

 

2-3 years sounds good, nothing good comes easy or fast, it just relieves me to know that it's not some 10 year endeavour, I don't want to end up making something half decent by the time my life is already more than halfway over.

Still, thanks for all the answers guys! I know I haven't replied to all of them but I've read all your responses.

Edited by James Miller, 26 January 2013 - 02:54 PM.