If the nintendo 64 was a CD-ROM system, FF7 probably would fit in a single disc and have 30 fps videos

If the nintendo 64 was a CD-ROM system, FF7 probably would fit in a single disc and have 30 fps videos.
The thing is that unlike PS1 and Saturn, the N64 had the grunt needed to play full mpeg videos without the need of an external cartridge, by using it's retarded powerful coprocessor to pull that off.
And it was actually pulled off on Resident Evil 2 (it DO look like crap, but it's mostly because they compressed 30+ minutes of video in mere 50MB. not even modern codecs can do much better than that).

To put this in more detail, the Mpeg video format use three tricks to reach it's quite powerful (for the time) compression
The first is compressing the frames with DCT, that is a 64d rotation (!) that converts an 8x8 picture block into a 8x8 frequency table that is much more easily compressible (see picture for an animated demonstration of the technique).
This one the sony playstation actually supports with a dedicated hardware chip, but when compressing the frequency tables it uses a subpar compression.
The second trick is to store frames that are just the difference of the last frame and the next, thus getting you a much cleaner and easy to pack picture, and the last one is a "vector table", that instead of storing any picture, store a "deformation table" that smudges the current picture around to look like the next one (you probably saw some gifs/webms that show when this one fucks up).
Those last two save a shitton of space, but PS1 can't into em, but the N64 fucking did.

Other urls found in this thread:

gamasutra.com/view/feature/131556/
twitter.com/SFWRedditVideos

>essay

Tldr bitch

Just read the first line, click the gif and watch magic happen in front of your eyes.

>(it DO look like crap, but it's mostly because they compressed 30+ minutes of video in mere 50MB. not even modern codecs can do much better than that).
To be fair they also cheated it further by reusing duplicate portions of FMVs and only switching them on-the-fly to Leon/Claire-specific sequences when required, where-as the PS1 original had completely separate videos for each character, reused scenes included.

The N64 can play .mpeg videos without needing an extra accessory by taking it into chunks + smudging frames, allowing a Disc Drive N64 to store FF7 in a single disc and with higher FPS videos to boot.

Makes you wonder of a future where the Nintendo-Sony console really did become a thing

...

The magic of not having to wait the disk head to take half second to find the data you need.

Am i retarded brainlet so thats some next level stuff for me. Have a free bump

>that instead of storing any picture, store a "deformation table" that smudges the current picture around to look like the next one
That's how fucking video compression works you numbnut. I-Frames for storing whole key frames, and B-frames for storing interpolated motion of the pixels from the previous frame to the next.

N64 has nothing magical hardware-wise about it that the PS1 couldn't do video-wise, outside of PS1 having a better sound chip for that sort of task.

gamasutra.com/view/feature/131556/
A good read on the RE2 N64 port btw for anyone interested.

Lemme see if i can translate from techshit to brainlet.
All the first thing 64D bullshit i'm talking about do is to get all the 64 tiles on this picture and mix em up until getting the final result.
The trick is that you can just "neglect" the noisier ones, making the data easy to smash up.

That's how modern video compression works, you know, the one made for powerful processors.
PS1 had a 386 tier CPU and no chance in hell of supporting B-frames, with the I-Frames being handled by a dedicated hardware chip.
Also the I-frame was compressed with RLE instead of huffman.
You're indeed correct on the N64 not having a "special magical hardware". All it had was a 480 Mflop DSP coprocessor that could grunt out all the kinds of frames.

Thanks anyway user. Hopefully more techsavvy ppl see your thread

It's fun to write a massive block of techshit.

>The first is compressing the frames with DCT, that is a 64d rotation (!) that converts an 8x8 picture block into a 8x8 frequency table that is much more easily compressible (see picture for an animated demonstration of the technique).
And that's not even getting into chroma subsampling to save even more on the associated colour data. Whoever came up with the old video compression algorithms of yores were real mathemagicians.

Sometimes Sup Forums is more techsavy than Sup Forums could ever be
I actually thought there was some weird tech demo out there of the PS1 doing mpeg decompression, it just wasn't feesabile when trying to run anything else on the system without major issues

>and mix em up until getting the final result.
So for the example OP gif, + means it's adding to the picture during its cycles, and I presume the yellow number is how much it's scaled before being applied to the portion of the picture it's assigned to, but what's the right-hand side of the 64d rotation for which the yellow number is over?

The chroma subsampling idea is probably inspired by the NTSC/Pal color handling, where they're encoded on a much smaller sample rate, but you don't notice because eyes can't into color resolution.

Well, unless there were audio issues, that would be quite fucking nice for the PS1, as you don't do anything else other than playing video during playing videos anyway.

Actually FFVII did have sections where the video was used as a background itself, while game logic and entities were still active in the current scene

before the POORLY CODED memesters come in, I want to say that I really hope one day we get Higan levels of cycle accuracy for N64

The + - is just to indicate that the tile is being added or subtracted from the final picture.
Also i don't have any clue on the "right-handedness" of a 64D bullshit.
The matrix order is actually a zig-zag pattern so is ordered from the blurriest to the noisiest.

this kind of knowledge make me feel like a genus get my dick hard make me smart inside my own soul

Yes, and on those you don't use real Mpeg.

The Cen64 guy is trying this very thing, but you know, stuff walks slowly.

Guess this explains why PS1 videos had that 'look' to them (limited framerate, cropped, pixelated speckling in spots that didn't look like your typical video compression), similar to how most N64 games look vaseline smeared courtesy of its hardware AA.

It's a bunch of low bitrate "jpegs" strung together.

I believe BINK player had a similar setup. Videos were broken into chunks (rectangular), and then were updated independent of one another. On older PCs that couldn't keep up with it in ye olde days (with the game logic running), you'd see the later rectangles (it worked top to bottom, left to right) start falling behind the rest of the video.

It's quite different.
PS1 always update the whole frame, and use jpeg like compression.
Bink i believe is probably some variant of the cinepak codec, where you use low color tiles to make a multiply free video encoder.

On the cinepak specifically, every 4x4 tile of the picture was a 2bit - 4color tile, with two 16 bit colors attached to it and the 4 colors being (color A, color B, 25% col A 75% col B and 75% col A 25% colb ..).

But bink did the small retangle trick you said to not need to pack the whole frame at once.

I just wish the modern method BINK uses for compression didn't result in needing a fuckton of bitrate bloat for something resembling 720p Youtube quality.

Shit plagued last generation for macro-blocked compressed vids, but from what I hear, BINK was both cheap to license, and very low overhead for processing, and Unreal engine supported it, so it ended up being the go-to standard.

It need at least some basic deblocking.
I bet they could even make it with shaders to save CPU time.