Zaxxon, should have been better for 1985

[zeusdaz]

Zaxxon on the ZX has been chosen this week by youtuber "Tyrantabyss" to be featured on the "S**t Game Time" series, slow, choppy with bad collision detection,...is that forgivable for 1985 ZX standards condidering it was the year of Dynamite Dan which was a piece of coding art?!, well it may have been acceptable of 1983 standards but surly not for 1985.

ZAXXON SGT Video Game Review:

[Gears of Games]

I actually bought this back in the day. Man was I disappointed! Especially after having played the rather decent C64 version...

[Sokurah]

Zaxxon on the Speccy was indeed a terrible and not well-executed game, but I honestly struggle to think of anyone that could do a real good conversion of it in 1985, because isometric and scrolling on the Spectrum is no mean feat.

[zup]

Well, there are not many games to make a comparison.

The only other shoot-em-up with isometric scroll that I can think of is H.A.T.E., but it has some different mechanics). H.A.T.E., and Highway Encounter demonstrates that isometric scroll may be done, and the first is also a 1985 game.

[StooB]

While Zaxxon was 1985, it was only just 1985, released in early January. To put it into context, Ghostbusters was at number one, and Ultimate were in-between Knight Lore and Alien 8. Vortex had released Cyclone around 6 weeks earlier and Highway Encounter didn't come out until July 1985, around the same time as Dynamite Dan.

[Pobulous]

The second stage in Raid over Moscow has similar gameplay, although with a different perspective, and that style might have been a better option.
Could have been mixed up with some into the screen levels as used on the Atari 2600 version to provide some variety.

Often it's better to do an interpretation that plays well, even if it looks a lot different to the original. These days I much prefer the looser arcade conversions, as they offer a unique experience vs the actual arcade or arcade-accurate ports on newer systems which are readily available.

[Rorthron]

H.A.T.E., and Highway Encounter demonstrates that isometric scroll may be done, and the first is also a 1985 game.

HATE was 1989. Highway Encounter was 1985, but does not scroll.

[+3code]

Mean Streak is from 1987.

[Ralf]

Zaxxon is one of these games that have cult status on other platforms but were always nothing special on Zx Spectrum.
Just like Boulder Dash, River Raid or Donkey Kong.

And looking at the video - yes it's bad. Very slow and jerky. Certainly could have been done better.

[Inky]

And looking at the video - yes it's bad. Very slow and jerky. Certainly could have been done better.

I had to have a quick nosey It could have been sped up a little quite easily. Basically, it seems there's a general reluctance by the coders to unroll any loops. For example, the scroll screen routine:

Code: Select all

$96CA  	ld      hl,$cfff
	ld      d,h
	ld      e,l
	ld      c,$80
$96D1  	xor     a
	ld      b,$20
$96D4  	rld     
	dec     hl
	djnz    $96d4
	dec     c
	jr      nz,$96d1

And the sprite drawing code is similarly pedestrian. I'm certainly not suggesting I've done anything remarkable, but I'd always unroll where it mattered

If I can find a place to put unrolled code, I might have a go at doing a .pok file patch at some point to see how much it can be sped up. I quite enjoyed doing the Ghostbusters copy-screen optimisation.

[akeley]

Certainly could have been done better.

It's easy to say from the safe and cozy vantage point of 2023, with nearly 4 decades of hindsight under the belt. But could it really? It's a somewhat unfair assertion, since the realities of game development were really much different back then, and also the situation and skills of various coders/publishers varied wildly.

This is why I detest @zeusdaz 's videos, which are unfortunately being spammed regularly all over retro forums. Apart from the juvenile angle (wow, I called an old game $hit and some other invectives, how edgy and funny, much clout) they are usually framed as a judgemental dump on some hapless, long-gone devs, who are not here and so can't really defend themselves, plus maybe they really were trying their best but had to deliver something in a few weeks, or whatever the case was in those Wild West days. Not to mention disingenuous apples/oranges comparisons with a game from completely different genre being somehow a proof that Zaxxon had to have stellar coding.

There's nothing wrong with poking fun and criticizing old games, as long as it's done with at least some modicum of understanding and objectivity. Unfortunately this series is anything but, in fact quite the opposite - definitely on the picking not even the the low hanging, but rotting on the ground fruit end of spectrum (pun unintended). And the fact that somebody actually pays to request these videos completely blows my mind.

[EdToo]

It was only recently that I began to appreciate the difficulties of cross-development back then when the developer might be presented with the finished game on a different platform and no source code and asked to replicate it in a short space of time. I think it was an article about Elite I was reading and the differences on each platform, for some reason I always assumed they would have been presented with full source code and clear notes.

[Lethargeek]

It's easy to say from the safe and cozy vantage point of 2023, with nearly 4 decades of hindsight under the belt. But could it really? It's a somewhat unfair assertion, since the realities of game development were really much different back then, and also the situation and skills of various coders/publishers varied wildly.

weren't the realities the same for Commando (or, if you want any-direction scrolling, for Critical Mass) in the same 1985?

[Stefan]

It was only recently that I began to appreciate the difficulties of cross-development back then when the developer might be presented with the finished game on a different platform and no source code and asked to replicate it in a short space of time. I think it was an article about Elite I was reading and the differences on each platform, for some reason I always assumed they would have been presented with full source code and clear notes.

When I ported Sophistry from the Spectrum to the SAM Coupé in 1994, I was given a disk with the devpac source code and a nice wad of print outs. Both were not final versions. So I wrote a disassembler, disassembled the Spectrum version and worked from that. This was all done on the SAM itself, which, while with a disk drive, legible 80 column text and 512KB of RAM was way above developing on a Spectrum, is still so primitive compared to today's tech.

[StooB]

weren't the realities the same for Commando (or, if you want any-direction scrolling, for Critical Mass) in the same 1985?

No. Commando and Critical Mass were released in November 1985 and were, with the exception of Nigel Alderton's work on Street Hawk, the developers only releases of the year. In contrast, Platinum Productions released eight titles in 1985 from Zaxxon in January to any-direction scrolling Rambo in November, so they would have had around six weeks to spend on each game.

[AndyC]

Isn't H.A.T.E. basically cheating a bit, by severely limiting the places where anything is covered by the background? Whereas Zaxxon (the arcade machine) pretty much revels in every opportunity to show off it's 3D-ness. I also suspect that if H.A.T.E. had been the Speccy port of Zaxxon, the same video would exist today bemoaning how much slower than the arcade game it was.

Also I'd second the idea that, if you can't contact the programmer for comment, it's not a great look to just flat out slag off old games. Most of the coders of the day we're overworked, underpaid and often barely more than children themselves. Working with barely any documentation, very little in the way of tools that a modern developer would expect, they mostly did the best they could. Sometimes it falls short, but not necessarily because they were lazy or incompetent.

[SteveSmith]

It was only recently that I began to appreciate the difficulties of cross-development back then when the developer might be presented with the finished game on a different platform and no source code and asked to replicate it in a short space of time. I think it was an article about Elite I was reading and the differences on each platform, for some reason I always assumed they would have been presented with full source code and clear notes.

I seem to remember in various SU articles where they spoke to programmer, it always seemed to be the case that, if they were lucky, they'd get an arcade machine to copy from. But more often than not they didn't even get that, never mind actually sprite data. Am I right in thinking that in Bob Papes RType book, he had to go to the arcades to actually see R Type?

[Lethargeek]

No. Commando and Critical Mass were released in November 1985 and were, with the exception of Nigel Alderton's work on Street Hawk, the developers only releases of the year. In contrast, Platinum Productions released eight titles in 1985 from Zaxxon in January to any-direction scrolling Rambo in November, so they would have had around six weeks to spend on each game.

what you say aren't different realities but a different choice in the same realities
it wasn't like somebody forced them at the gunpoint

[SteveSmith]

Also I'd second the idea that, if you can't contact the programmer for comment, it's not a great look to just flat out slag off old games. Most of the coders of the day we're overworked, underpaid and often barely more than children themselves. Working with barely any documentation, very little in the way of tools that a modern developer would expect, they mostly did the best they could. Sometimes it falls short, but not necessarily because they were lazy or incompetent.

I see what you're saying, but you could level that argument at any criticism of anything, e.g. we can't criticize the Marvel films because we don't know what circumstances that scriptwriter or special effects people were working under. I don't think there's neccessarily any direct criticism of the programmer (although they are bound to perceive it as such); the publisher should probably be the one that receives the criticism for being in overall control and deciding to publish it. In addition, Zaxxon did only get 58% in Crash (the only UK mag that reviewed it) so it wasn't that well received even back in the day. Worrying about the programmers circumstances certainly didn't cross the minds of the mags back in the day!

[Van_Dammesque]

I had to have a quick nosey

I know something of programming but never heard of the term unrolling a loop, can you explain it like I'm a dummy please?

[SteveSmith]

I know something of programming but never heard of the term unrolling a loop, can you explain it like I'm a dummy please?

It means to perform a function manually a certain number of times instead of putting it inside a loop. For example, to print the numbers 1-10, you could put it in a loop, or you could just write "Print 1", "Print 2" etc... 10 times.

[Van_Dammesque]

It means to perform a function manually a certain number of times instead of putting it inside a loop. For example, to print the numbers 1-10, you could put it in a loop, or you could just write "Print 1", "Print 2" etc... 10 times.

Got it, thank you.

Would this practice on not rolling out be an artifact of designing a code base for multiple platforms?

[Pobulous]

It would be interesting to see a full disasembly of this one.

They are doing a 4 pixel horizontal scroll of an offscreen buffer, and then presumably copying it to the screen 2 pixels lower for every scroll.
But then at some point, even with an oversized buffer, they need to adjust the whole buffer downwards as well.

Would doing a circular buffer help, so (also combine with unrolling the inner loop):

Code: Select all

buffer_end
        dw    $5cff
$96CA  	ld      hl,(buffer_end)
	ld      d,h
	ld      e,l
	ld      c,$80
$96D1  	xor     a
	ld      b,$20
$96D4  	rld     
	dec     hl
	djnz    $96d4
        bit      4,H         ;if H has crossed the 4K boundary, move to end of buffer
        jr       z, checkc
         ld      hl, $cfff
checkc:
	dec     c
	jr      nz,$96d1

Then before the next scroll,
zero bottom 2 lines of buffer
add new tiles on right edge of buffer
decrease buffer_end (and check for wrapping)

[Lethargeek]

Would doing a circular buffer help,

I don't think a game looking like Zaxxon even needs an actual scrolling. Eg Commando has no real scrolling nor a buffer, but lots of unrolled separate subroutines for almost every type (even a typesize) of bg object, and lots of specialized subroutines for the same sprites - one for moving in every direction, for enetering and leaving the screen, you get the idea.

[AndyC]

Got it, thank you.

Would this practice on not rolling out be an artifact of designing a code base for multiple platforms?

Not really, for the most part unrolling a loop is going to offer similar advantages on any machine using the same CPU. But unrolling loops does generally take more space, so might be avoided if space is tight (or if the code isn't time critical). Of course in the early days, most coders weren't overly familiar with instruction timings (it was more about making it work than making things work fast) so unrolling loops.