I find it very interesting when developers challenge themselves by not using existing engines, frameworks etc, this looks super difficult so props to you!
@@monad_tcp Doing a basic TLS implementation actually isn't that bad, mostly just a lot of learning about cryptography and reading RFCs and stuff (and I imagine you're no stranger to that). I'm doing a similar "full stack" project right now. I'm more or less done with TLS 1.3, but I still need to do an ethernet driver and TCP.
@@monad_tcp I wouldn't write anything like that myself to be honest, waaay too big of an opportunity for critical vulnerabilities when you're making something that is touching the internet at the lowest level.
The biggest jaw dropper in 1990's programming is the use of mathematical tables in memory and other tricks to cheat calculate results rather than relying on C or C++ math libs to do calculations in order to optimize for speed. Also, they were limited to using very little memory. Lastly, the display functions were mostly written in assembly to also optimize for speed, and that was a bitch to do. People like Ken Silverman and John Carmack were absolute beasts with their respective engines.
And the most jaw dropping thing for me is that people don’t know that you can rotate a vector using a matrix. And that quake used Taylor series to re-normalize said vector. And that you don’t have to waste memory or cache on this even on a 386 without 387.
But can people still write code like alllllll this (including the replies) without an IDE that highlights, suggests, auto-fills, and catches syntax errors like it really is 1995 like the title says.............. No, I don't think they can.
@@isaacyukon5869 at work we have a crappy system without IDE. Eclipse should work, but I could not get the matching versions of eclipse and plug-in and windows..
@@isaacyukon5869 I mean, yeah, people can. They just don't in a majority of cases, because it doesn't make sense to. Self imposed limitations as challenges, like this one here to write an engine that uses similar techniques to those used in 1995 due to technical limitations, are done for fun; whereas most people wouldn't consider being forced to look through a textbook or re-read through large portions of their codebase instead of having the IDE autofill 'fun'.
You have to consider that he definitely did his fair bit of research on how to do it. Back then, these 2.5D/3D games were groundbreaking for a reason. The maker of this video didn't invent the algorithm, he is just building what other people did before him and that information is widely obtainable through the internet. Not to play down what he did, it's still amazing work, but that doesn't mean he is a god programmer.
For me, there is comment of guy just below your comment who states that some part of this video looks familiar to him because it was something he already did. Thankfully name of that guy is mentioned in description and he was thankful for that.
@@Danuxsy Huh "dOiNg SoMeThInG UseFuL lIkE Ai RESeAcrch", AI research is super different then doing stuff with low-level system languages (like coding in C and C++) and bare bones hardware (Making my own GPU ect)
@@Danuxsy who tf said that we should all be focusing on birthing AI? the knowledge he has has nothing to do with AI, the guy is coding his own engine from scratch. but of course, it's clear that you don't have much experience in the area
"Programming a first person shooter from scratch like it's 1995" in 1995 Rare was creating Goldeneye for the N64 on SGI workstations, not a kind of 3d doom clone, that is insane to think about how quickly computer technology was moving at the time.
They were using SGI workstations, but the programming of the game engine was still mostly done in C and a bit in MIPS for optimization. SGI workstations came with their own C compilers on them, and the GCC was also installable.
hey! I've been a fan of yours I think since before that video was originally uploaded :) your code was a fantastic reference and obviously the tutorial visuals stuck in my head. IIRC my approach differs in that I ended up projecting walls according to a conversion of their angle relative to the camera (basically -HFOV/2..+HFOV/2 mapped onto 0..SCREEN_WIDTH) whereas I don't think I ever fully understood the method your code uses - it was a good reference in the start but I think the only way I could fully understand things was working out the math myself!
@@jdh IIRC, Biqswit uses the regular old z-divide to achieve 3-D perspective. Your approach and his work effectively the same way and are based on the same principle, the triangle congruency. Doom uses a similar technique to yours, to "project" the view-space angles of the wall endpoints to screen-space x-coordinates--albeit with the help of a pre-calculated lookup table for trig functions. Anyways, great content! Loving the aesthetics of your gfx engine.
that's part of the plan! I had originally even used fixed point Q16.16 math for the project just so that it could run on things without an FPU, but it got too tiresome when it started expanding. when the final game is done though I do want to backport (if you can even call it a "backport") it to MS-DOS!
It's really quite interesting to see you come up with all these fun challenges Especially when nowadays literally every game dev chooses to use a game engine, making a game from scratch is a topic that no one really talks about
You might be surprised by the abundance of games that roll their own engine. It's still a pretty common practice for a developer to make their own engine from scratch, especially when they have particular requirements like performance or rendering techniques. That being said, most custom engines are just your basic old 3D or 2D renderer, what jdh makes is on a whole level of its own.
@@bluesillybeard I am aware that many games are made with a custom engine It's just that people don't talk about them a lot and there's not many devlogs about them Then there's jdh of course...
@@BLAZE_GLITCH there's thinmatrix as well. Major game devs won't show off the internal working of their engine, that's trade secrets, tech demo are also rare.
I started to learn programming 3 months ago. I cannot imagine myself writing a single line of whatever black magic fuckery you did in this video. Seriously, to me, people that can come up with stuff like these are geniuses
I'm 2 years into a comp sci degree and honestly same. His code looks like math equations instead of programming to me. Love watching him do this amazing stuff though.
making 3D games or 3D software is the most difficult thing to code. It's the highest level of programming. Also at some point you start using OpenGL or DirectX functions.
@@tealtrim9747 Once you get into 3D programming and low level, everything is just math equations... Some of the videos I have made on 3D stuff just needs hours of explanation. If you want I can hook you up with an explanation I made of basic raytracing.
@@inlandish I am not very good at math, I should probably just stay away from the more mathy areas of programming. I appreciate your offer though, I just don't want to waste your time because I'd be too dumb to understand or properly utilize what you teach me.
As soon as you got the lighting and pallettization in I instantly grew jealous that I will likely never make something that nails this aesthetic so well in my lifetime
Incredible work. Here’s a tip you didn’t ask for. To get that smooth stepping texture mapping look, you’ll have to subtexel correct each vertical scanline. Just passing this along because it took me a while to get right
What a coincidence! Just a couple weeks ago, I started on my own doom-style engine using Bisqwit as a reference. For my engine though, I went entirely with vector based graphics. I also, in accordance to 90's standards, only used Borland graphics interface. I definitely have not made as much progress as you, but I think we're trying to reach different goals. Regardless, great job!
See, this is why I love being able to code. You don't need anything except a creative mind, an internet connection, and a laptop to build entire worlds.
You are a monster, man, really.. I was looking for that kind of content for a long time.. a real programmer working on a doom coding the old fashion way... Of course I learned something like 1% of what you said but I really like those kind of videos where people show the fundamentals. Thank you
More jdh! Honestly, I am sad to see the robot botanist go, but its nice that you can move on to some other projects, and continue to learn! Any chance we could have the Alpha of the robot game to play around with?
Haven't seen the code yet, but based on your explanation at 13:09, there is a chance that the player could clip through a vertex defined on an adjacent "block" if they're going fast enough. A quick fix would be to check every block's collision that the movement ray intersects, with a ray-intersection function similar to the one used in the Wolfenstein prototype. Looking forward to Spinach-Cat-Banana-Game getting its well-deserved release!
@@nikkiofthevalley Unfortunately it's not that simple. Let's say a vertical wall spans the height of an adjacent block about 1mm left of the right border of its block. The player could be **very** close to this vertex, but still stay outside of its block because it hasn't crossed the block's border yet. If the player is 1mm right of the border, and the wall is 1mm left, then if the player moves >2mm left in one frame (not very fast at all), it'll clip. That's why we need to check whether the movement ray intersects with any other blocks.
good catch! I didn’t mention in the video, but the code already does that - it uses the exact same DDA “line drawing”/graph traversal algorithm as the wolfenstein demo just like you mentioned!
@@jdh Great! I mention this because I was working on a similar engine (in ES3 javascript 🤮) and the player kept clipping through borders of where the BSP was set up. Took me about a week to figure out, and by that point I had lost interest. Love your content and your dedication to making your own game engines.
People are saying "don't program your own game engine, they already exist, it is waste of time" and then somebody actually creates one and makes a YT video about and everybody is like "holy shit, that is amazing, you are so good!" Nice work, btw!
For most intents, a off the shelf engine like unity, unreal or godot works just fine. Btw i am watching this as i was thinking about doing my own engine (for calculator games so that's why)
@@attilavs2 Yeah, for a production game a real game engine is a good choice. However, making your own engine for *learning experience*, or for obscure platforms like calculators :D, is what I want to do as well.
@@abuk95 Not only production games, if you have like a creative idea and don't want to spend too much time on the codey bit, or want fancy graphics, or even just limited by time. Btw if you want to try calculators, it's pretty cool with the latest ones you can run basically every 2d game and stuff like doom ect... Basically a 1993 computer in your pocket, and on which you can have almost full acces to the hardware, especially on Casios
Damn man this is so cool, I'm so jealous. I went into programming, starting with C, to one day be able to code games like Doom, Half Life and the likes, and the furthest I ever went was a Wolf3D clone. I suck at math and I wish I could understand 10% of what happens in a portal-based rendering engine like you wrote. That's awesome! And I'm so glad you used plain old C.
it actually isn't super complicated! I strongly recommend Bisqwit's video (ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-HQYsFshbkYw.html) and Fabien Sanglard's "DOOM Engine Black Book" (fabiensanglard.net/gebbdoom/) - they make it very easy to understand
The doom black book is great but it’s too high level and doesn’t go into details (at least with topics like how it does texture mapping) So you are left with a lot of questions unanswered
It will provide a formula but I had to actually look at the doom source code to see certain things related to texture mapping but I still get quite confused. It’s just drawing a vertical column but I haven’t been able to wrap my mind around how it decides what texture columns to draw. But then there’s a whole ordeal with texture pegging which I don’t understand either
When I watch jdh's videos I always end up wishing I was as good of a programmer as this guy is. This guy can just make whatever he wants. This guy is on a level that most non programmers think you can get after you study programming for a couple of months.
It's not really about programming skills, but mathematics skills in this case, I'm programming using C++ for years now, but when it comes to reinvent the wheel like this guy did, I'm pretty sure I will get quickly bored and just throw away the project in the trash can lol !
@@K3rhos Linear algebra wasn't hard for me in university. I will definitely try making 3D games in OpenGL but I highly doubt that I will be as successful as this guy. Because he seems to possess big amount of knowledge of so many different topics. He built his own PC ffs. He built his own OS to run Tetris. He made Minecraft from scratch in a couple of days. I can't make anything serious in a couple of days without introducing bugs.
to make the movement more smooth you should make the direction keys accelerate the player rather than set the velocity to a fixed amount (like it looks like you're doing). it really adds a lot to the feel of moving around! anyways, nice project, i look forward to seeing it develop!~
This is very impressive, I couldnt imagine trying to do all of this without a graphics api like DX, opengl or vulkan. Very cool, thanks for this awesome video
really blows my mind how much work and knowledge went into creating these games back in the day. and they did it all without google or modern code editors
You should implement this engine for your 8-bit CPU 😎, or even a 6502 computer, then you'll even be playing on similar hardware. But honestly, great video, and very interesting project. Usually people only make the original Wolfenstein style raycast engines, but not the DOOM one, so this was pretty informative and fun to watch.
a 6502 would have to be overclocked to oblivion to implement this (and you'd also have to write a full floating point emulator). I think you meant i386?
@@thezipcreator True, but I think I've seen demos of doom/wolfenstein like games running on the Commodore 64 and other 6502 hardware? (Though I may be wrong). I was just referring to that, as it was much more similar to the CPU he built than i386, and would present more of a challenge.
Doom got it's speed from several really nasty tricks, most notably modifying the executable code to draw walls as it was drawing. Not that I think jdh can't do it, but it's fairly different to what he's done so far.
> prev video: what went wrong: no gameplay, no prototyping, started with an engine; > next video: starts to create an engine, but now completely in software. Logic!
This was my adventure back in 1999, when I came across the book "trricks of the game programming guru", by Andre Lamothe. It was a little different back then, when you were writing your own interrupts, chunks of inline assembler, and good ol' mode 13h, where your resolution was 320x200x8.
"I gave our enemies possibly the world's worst AI." *Proceeds* *to* *literally* *describe* *Doom's* *own* *AI.* Yeah, I can confirm that sounds about right.
"I've never had to say its name out loud before" Relatable as hell. Especially frustrating if the people you're talking to don't speak english for some reason.
Thank you. This video appeared at exactly the time in my life that I needed it most. That being, the time I've started thinking a lot about how 2.5D shooters work under the hood. It's something I personally associate more with Dark Forces than with Doom, but it's got that nostalgic smell either way.
@@iambored1528 so then explain to me then how the doom source code which is publicly available is actually not 3d and explain how it is 2 dimensions and a half since clearly you think public source code that you can read and compile for yourself, isnt 3d.
@@iambored1528 That is completely false and can be proven by looking in the source code. So now I know that you have no clue what you are talking about and are making up stuff that goes against widely available facts. plus you can't have floor and ceiling height in ray casting. Doom uses binary space partitioning, btw.
@@iambored1528 No, i am saying that it is open source meaning you can read the code yourself and see that it is 3d. but you are instead to lazy to do that and are now lying in a youtube comment section
@@iambored1528 height would be another dimension, making three dimensions. you are literally proving yourself wrong but refuse to see it. and again you refuse to look at the doom source code which proves you wrong. why are you intent on lying to people?
Doom did not use portal technique. I just used a 2D BSP tree for solving the drawing order of walls, or rather, (not necessarily intuitive) cut segments of walls. There was no forming of enclosed sectors. The Build engine is more close to portal rendering (implicit by defining 2D sectors), from what I could gather & remember. One could argue that Quake, using a style of 3D BSP tree which _does_ enclose convex 3D "sectors" ("solid"-BSP), also "kinda" uses "portals", which are auto-generated, but it's not strictly a classical portal renderer either. Whereas in Doom, the BSP does not serve a portal function at all (which would typically also help deciding in what sector currently dynamic entities are etc). It does not even use the BSP for collision handling, something Carmack only thought of later.
You should try and get some sort of sound propagation / resonance / reverb system going, like steam audio! Way too modern for a 90s engine, but it seems like one of those cool areas where you could blend the technology together :)
at first i thought from scratch meant from scratch the game dev tool for kids basically and i was very intrigued but i was pleasantly surprised, really enjoyed the video :)
haven't seen you use Rust on your channel, have you ever considered making something with it or trying Rust? I think you'd feel at home thanks to it being fairly low and high level both, while also avoiding bugs early on with memory management :)
insane absolute gods that came up with this stuff. it took many decades to get to this point though. the first "3D" game was made by some nasa guys back in 1973 called maze war which was also the first multiplayer game. It's mostly wireframe though, but it looks 3D, and you probably know about this already.
i think i would never understand programming on sheer conception level. i can do simple baby code by myself, but programmists looks like a gods for me, like I'm trying to understand 4th dimensional creature
Dude I am so jealous that you are able to do this; I've so often wished I could program my own engine like this and build a game completely from scratch...
Sure you can’t? For most of us we just can’t beat the global competition. But it is not too hard to recreate this on legacy systems for nostalgia. GBA,
Huge respect to you. Being 15 and also with the information resources available in 95, managing this is an insane task Wasn't even born until 3 years later haha^^
As a long time fan of the channel and seeing as I’m learning C++ for my University degree, including SDL, this video was very interesting to me! Keep up the good work!
I always complain about RU-vid becoming oversaturated and "I miss 2019" and then RU-vid is like BAM, new jdh video. Chill, but still entertaining. Keep up the great work!
A quick hint. Declare ASSERT as "if (e); else { ... }". It will let you use ASSERT in conditionls like "if (x) ASSERT(...) else something". Moreover, the compiler will raise a warning if `=` rather than `==` is used in a condition.
It's a trip to see how video games were made before engines were really a thing. You gain perspective from seeing how developers solved simple problems that are now easily solved by block code. Also, you should get a pop screen for your mic, or move the mic.
Pretty cool to watch you coding essentially 2 decades worth of tech progress in a single video. Having lived through all of it, it's refreshing that those younger than myself can learn it all from published sources instead of the propriatary secrecy which was the 80s and 90s.