Love to see videos about computer techniques from early 90's to mid 90's

Says the guy who hasn't probably written a line of code in his life and yet he feels the need to tell other people how to write their things.

Really, you have no fucking idea what Mode 7 actually is. Why should I value your opinion?

Contribute something useful or gtfo

because when I google information on how to build a basic engine I get angry redditors telling me to use unity?

I can't even find information on how to replicate capcom psx backgrounds.

Surprise, redditors suck. People in general suck. I'd help you if I knew enough about the topic. My intention was to give an inspiring alternative view to a common opinion on contemporary software development.

Have you watched any of GameHut's videos? He worked on some old videos games such as Mickey Mania and Sonic 3D and goes over some of the tricks they used to achieve the special effects.

I'd say it's closer to being a branch of mathematics.

You are in love with some aesthetics, not the techniques behind them. While the techniques behind some optimizations done back in the day sure are interesting, they were done that way because there were no better ways to do them. I'm not saying you should use Unity, but really, consider using libraries because the maths behind from scratch software rendering are some autistic shit and take quite a bit to implment correctly. If you want to make a game, that is; if you want to get into the demoscene, or just wrote a software renderer for the sake of eriting a software renderer then by all means, go ahead.

I think you are looking for parallaxing. There are many tutorials for that.

Sounds to me, you don't know how to use a search engine. Let alone which search engine to use.

This is where I leaned a lot (college is useless).
youtube.com/watch?v=2Op3QLzMgSY&list=PLAA97f8v5JX5WRZ6DUBSsogXlG4biWinL

Make something in TempleOS. You get functions for pixels, lines, circles, sprites. After that it's up to you.
Or write something for any 8-bit computer; it's about the same deal but you get very little cpu and memory, and you pretty much have to figure out the best way to use its graphics and sound chips. That's fun, if you can get the real hardware. And they're neat because they all have their own distinct personality due to the different hardware. Compare these two demos, both with Z80 cpu but entirely different graphics/sound chips. First one is ZX Spectrum, the other is MSX.
youtube.com/watch?v=7gmoPuF3_VM
youtube.com/watch?v=busiDuS94qY

Isn't the point here to push our current hardware to the limit with new crazy tricks? The old techniques are inspiration, not something to re-implement.

Crazy tricks implies exploring the hardware, not just coding to a library. You could theoretically do that and hit your GPU directly, but unless your platform is something like raspberry pi board with everything integrated, you run the risk that few other people will be able to watch your demo or play your game. And anyway, you'll never get anywhere close to the modern hardware limits. People still haven't fully explored computers from the 80's yet. Check out this demo from 2011, it completely turned the Amstrad CPC scene on its head and yet the authors admit that more experienced groups could do much better still:
pouet.net/prod.php?which=56761
And anyway you won't be able to do much exploring with modern hardware because the OS keeps you in ring 3. Before computers got all lame, your code was literally running the whole system, it was the OS so to speak. The closest you can get to that now is TempleOS. Of course this doesn't solve the incompatibility problem, where every modern PC has different sound and graphics hardware. But it's closer, more authentic to the old school because at least he enforces the same screen resolution, color depth, and limits you to the PC speaker only. So it's kind of a pseudo-compatible layer running in ring 0, to pretend you got a modern C64.
Well here's a VIC-20 demo (for stock machine with 5120 bytes of memory) where he actually uses lots of tricks he figured out himself:
pouet.net/prod.php?which=10626
Same thing with Trixter's 8088 demos:
pouet.net/user.php?who=819
Those guys are actually doing crazy tricks. Just using library in ring 3 isn't going to get you anywhere near that.
Bonus video: here's a recent production for Atari ST, just released at demoparty a couple weeks ago:
youtube.com/watch?v=IeyJkV8d8qc

get yourself Amiga, it's has shit ton of different screen modes and drawing anything on it is nothing like on pc.

If it comes to demos on retro computers the C64 and Amiga rule evetything. I like to see some cool stuff on the other platforms, but for the 8 bit c64 will always be the best and for the 16 bit the fucking Amiga, the consoles might be more powerful but nobody makes demos for them for obvious reasons, thought titan made a ground breaking demo for Sega genesis last year. Heres couple of my favourite c64 demos. I can actually explain on how some of these hardware tricks works. C64s video chip gets abused so hard to make this shit work. it's so great that you can manipulate it on the fly unlike on the NES for example where you can't do anything while the image is being drawn. Cpu is only usable during the vertical blank.

youtu.be/bIM4p0uL6Pw (2016) - killer samples with animations at the end. Amazing stencil effects
youtu.be/B4gC397tdlY (1995) - one of the first demos to really push the 4x4 chunky effects. includes real time wolf3d like maze.
youtu.be/rsZ3IzRBeQ0 (2001)
youtu.be/81qRtZfkYE8 (2001)
youtu.be/NJ3gEm4Xerg (2000)
youtu.be/SWdfD2Q-3Z8 (1997) - one of the best and craziest polish demos. Amazing sync to the music.
youtu.be/pSpVNWayZFY (1993) - first trackmo on c64
youtu.be/dyuN8XAn8KY (1994) - first demo with fucking fast 4x4 screen mode allowing these crazy zoom rotators and spheres.
youtu.be/o9eLNFD_AKk (1994)
youtu.be/Obhh0V0an5k (1990) - the first interlaced flexible line interpretation picture at 4:00 showing about 40 colors.

unless you're doing graphical computing like more efficient rendering n shit you aren't doing what could be considered "computer science", you're just gluing libraries together.
The science in computer science is building the tools that other developers use, basically. Or just doing the highly theorethical stuff (solving non-computable problems in non-constructable machines, that sort of shit)

Can't you just do you own OS and do the demo for that? I heard of a few guys that did something like that, but not for a demo, it was for number crunching. They did an OS that was mono-task but took the task and separated it into parallelizable chunks to take all the juice they could from a consumer processor.
Truth is, that kind of shit isn't really done outside academia because hardware has gotten that much complex to completely squeeze hardware dry. Thanks Intel for that.

You can add also all the functions that intel/amd hides from people and that will never be exploited besides some specific people.

You're the same nigger that made the thread about how he was too retarded to learn high school math and shit himself so hard a vol had to clean the thread up, aren't you?

That's engineering, not science.
That's mathematics

Computer Science is not a science by any coherent modern definition of the word. It's called a science for basically the same reason 'social sciences' are called sciences, misappropriation of the term for pseudo legitimization of its practice and communication to a philosophically illiterate public, and because it's different enough from the mathematics it came out of to warrant a new term, but 'Algorithmatics' is etymologically atrocious and sounds dumb.

Old scenecoder here. There aren't many crazy tricks these days as modern hardware is much less retarded. I used to have to do fast screen wipes by sending the data through the FPU on the 386 as while the hardware had a 64 bit bus the CPU was 32 bit only. You'd not expect to see hardware this wonky today. Getting lower-level access to GPUs could be interesting, but much of what you could hope to get has already been made available by OpenACC.
The main issue today is how so few people know how to use what's already available and straight-forward, no tricks.

Theoretical computer science is mathematics. Theoretical computer science is a branch of mathematics that is concerned with the mathematics of computations.

We already have a proper name for "computer science" (which doesn't make sense because it's not a science studying nature) and that name is "information technology" aka IT.

Dear OP, I know that feel but I don't know that feel. High-level getting you down because its proponents are retards who don't care about the art? Fine, but don't forget: low-level is when the machine gets in the way.

CS should be called Compebonics, because people can't be forced to write in proper machine language.

Computer science is not information technology. You can have one without the other. You can develop and operate IT that has no explicit regard to the fundamental mathematical logic of computer science. You can study computer science without ever touching or implementing any kind of IT.

You havin a laff mate? Those have no connections, it's a non-sequitur.

Reminded me of a source port for Quake called RetroQuad: youtube.com/watch?v=gh-lL2L4WvQ