NeoGeo Metal Slug conversion to Amiga 500 - madman task ;-)

[zero]

Originally Posted by Hewitson

It often slows down to well below that.

Yeah, which surprises me a little. The Neo Geo graphics hardware can cope with it - there is no sprite flicker and no bitmap to render, so it must be entirely the CPU being overloaded. The stuff on-screen isn't that complicated, especially for a 12MHz 68000. There is even hardware acceleration for sprite animations.

I couldn't see any hardware support for collision detection though. Maybe it's there, I didn't dig very far.

Maybe the coders just suck. Midnight Resistance is similar - the arcade board slows down more than the Amiga version, despite massive hardware acceleration of the graphics. Often there isn't even much going on.

[roondar]

Well, I didn't count the number of objects on screen in Metal Slug, but it seems to have tons of them. This could end up making collision detection a fairly hefty process to complete (though this doesn't go for all methods, most collision detection algorithms scale up poorly).

There's also the CPU load for keeping track of what object is where and which frame/scale factor to use. This could also take up quite some time given the large number of objects and elaborate animations/movement patterns.

Another idea: they may have opted to use C for the main code. Which will work just fine, but if the compiler wasn't great it might end up slowing everything down.

Of course, it's also possible the code is less than stellar - but that's hard to verify

On a related note, Metal Slug is not locked to 30FPS (i.e. it's trying to do 60 but fails). Metal Slug 2 apparently runs@30FPS and has fewer slowdowns as a result.

[Master484]

The Neo Geo development wiki has a page about slowdown:
https://wiki.neogeodev.org/index.php?title=Slowdown

Here is a quote from that site:

---

" Slowdowns are perceived during gameplay as slow picture updates (frameskip). On the NeoGeo, this happens when the CPU is not able to update the VRAM often enough to animate sprites smoothly.

If a game wants to run at a constant 60 FPS (no frameskip), the calculations for a frame need to last less than 1/60 = 16.7ms. "

---

Also the wiki says that Metal Slug 2 used a slow method to update the VRAM, which caused lots of slowdown.
https://wiki.neogeodev.org/index.php...Vehicle-001/II

So VRAM update speed seems to be the "bottleneck" that is the most common cause of slowdown in Neo Geo games. So bad programmers could easily make a slow game, despite having a million hardware sprites.

And on Neo Geo everything that moves has to be made of sprites, including the scrolling backgrounds. So I would imagine that VRAM needs lots of updating all the time.

[sandruzzo]

it seesms to slow down allways! Very disappointig

[idrougge]

Originally Posted by roondar

Whether a game is 'complete shit' or not is almost always subjective. Facts generally do not come into it.

Case in point: you keep picking examples of games that in your eyes are poor and pretending that these games are objectively poor when they often happen to highly regarded and thus clearly not objectively poor.

Actually, he expressively picked games that are held in high regard, so as to make a point. He could easily have picked a Tiertex game, but he chose not to.

Leaving objectivity aside, if we step back from the Amiga perspective for a while and just look at the games from an "objective" perspective, most Amiga games just don't fare very well. Frame update is a major factor — take any game released on more platforms than the Amiga and the ST, and the Amiga version will perform poorer. That is not just a matter of development resources, but the amount of effort needed to make a game with equivalent performance on the Amiga.

Yes it could, it had a number of platformers running at 50FPS, including a bunch that utilized two layers. That your example isn't part of that list doesn't change the facts in any way.

Again, Roondar never said that Amiga games couldn't ever be done in 50 fps, but the fact that even Andy Braybrook couldn't reach 50 fps in a game where the Amiga was the lead platform should tell you more about the Amiga's hardware capabilities than about the capabilities of the individual programmer.

Absolute nonsense. Incorrectly utilized maybe. But not pathetically underpowered. Specs & results wise, it eats the competition that was available when it was released for breakfast.

Unfortunately, the competition at the time was the ZX Spectrum, the C64 and the Atari ST. When the Amiga reached its prime, that was no longer the competition.

That consoles released several years later do better is to be expected and claiming the Amiga specs are poor because of the MD & SNES is, IMHO, extremely silly. You don't strike me as someone who doesn't understand these basic facts of how technology works, so I'm rather surprised you keep doing this.

I think Roondar does understand, but a lot of Amiga owners (my younger me included) do not. Making an Amiga game according to 1991—1992 standards is an entirely different affair than making a game according to 1991—1992 Megadrive standards.

The same goes for all platforms. To drive this point home: the vast, vast majority of Sega Megadrive & Super Nintendo games are junk. There are a few really good games on there, but like all systems the rest of them are utterly forgettable.

I won't question you on this point, but even the shitty games were even shittier on the Amiga, if they were ever ported. And I think that's part of Roondar's point — given the amount of passion necessary to make a truly great Amiga game, you could make an even better console game. Given that most console programmers weren't as passionate, the console hardware really is that much more powerful.

[roondar]

Originally Posted by idrougge

Actually, he expressively picked games...

@idrougge & @Hewitson - I've moved my reply to a separate thread because this is getting far too much off-topic.

You can find my reply to idrougge here: http://eab.abime.net/showthread.php?t=96392

@thread: no more off-topic about this point here from me. Carry on

[Tigerskunk]

back to topic:
What people also forget when discussing a port of Metal Slug is, that the background is almost non repeating. There are almost no tiles, everything is custom.

Means, you would need loads of RAM or streaming content from disk all the time. You would have reloads every few seconds, which would make this the shittiest of shitty ports.

So, I don't believe in an OCS version that would do the game justice.

[Tigerskunk]

Originally Posted by Amigajay

But yes, crazy to attempt NeoGeo ports, but the NeoGeo Pocket versions which still are good games would be better suited imo.

That one would be very possible to port over...
Seems to use around 16 colours.
The parallax seen in the first stage could be difficult to implement, but would be possible with a sprite layer.

[ Show youtube player ]

But then, people would complain about the lousy graphics for the sprites...
Maybe somebody could do a makeover on them..

[zero]

Originally Posted by Master484

Also the wiki says that Metal Slug 2 used a slow method to update the VRAM, which caused lots of slowdown.
https://wiki.neogeodev.org/index.php...Vehicle-001/II

That makes me think that either it was written in C or the devs were not the best... But their other games seem more competent. I guess it was their second game as an independent studio, the first being Neo Turf Masters which isn't exactly demanding.

[roondar]

It may be better to try and create a Metal Slug 'like' game that is not Metal Slug at all, but plays and feels (in so far as that feeling isn't based on the strength of NeoGeo hardware) similarly.

That way you can make something that works within the limits of Amiga hardware and not something that will always be considered a poor port. I know the NeoGeo Pocket version kinda is that, but it's still Metal Slug so like Steril said people will start to complain.

[sandruzzo]

Game boy advance version

[ Show youtube player ]

This make me think that an Amiga Version cold be possible, whiout being a direct conversion

[zero]

Originally Posted by sandruzzo

Game boy advance version

[ Show youtube player ]

This make me think that an Amiga Version cold be possible, whiout being a direct conversion

That doesn't actually look too bad. The sprites are on the large side but there isn't too much on screen. They seem to have swapped the massive number of objects on screen for less linear level design and more exploring, to fit with the GBA hardware limits.

[sandruzzo]

Even game boy advance version suffer of a lot of slow down, even with all that Hardware!

[donnie]

Originally Posted by Hewitson

Not opinion, fact. Time to take off the rose tinted glasses and see most Amiga games for the complete pieces of shit that they were.


Right. Thats why most games run at 25 fps, or in some cases even lower... The A500 couldn't even manage to run a basic platformer like Fire & Ice at 50fps for christ's sake. Pathetically underpowered.

Don't get me wrong though. There are some AMAZING games on the Amiga. There's just not very many of them.

No its your fucking opinion and not fact.


There are more games on amiga than mega drive and snes combined. There are so many games in genres you would never se on a 16bit console.


Alot of amiga(and other microcomps) games development have more incommon with modern indie games development than 16bit consoles.


Just take amiga games like lemmings, another world, moon stone

[Tigerskunk]

Originally Posted by sandruzzo

Game boy advance version

[ Show youtube player ]

This make me think that an Amiga Version cold be possible, whiout being a direct conversion

Maybe an AGA version, definitely no OCS.

[sandruzzo]

Originally Posted by Steril707

Maybe an AGA version, definitely no OCS.

Too easy on AGA

[DamienD]

Yeah right, crack on then sandruzzo

Edit: oh, you are talking about the Gameboy version, ok then.

[zero]

Yeah, the GBA version doesn't look too bad. Ditch the parallax, 16 colours + copper, target 25 fps...

Such conversions are kind of a lost art. A good example is Midnight Resistance for the Spectrum. The arcade game is way beyond what that system is capable of. Just 48k RAM, no hardware acceleration for anything... But it's rated as one of the best games on the system, because it made changes to suit the machine and maintain the spirit of the original.

The GBA version of MS is the same.

[sandruzzo]

Originally Posted by zero

Yeah, the GBA version doesn't look too bad. Ditch the parallax, 16 colours + copper, target 25 fps...

Such conversions are kind of a lost art. A good example is Midnight Resistance for the Spectrum. The arcade game is way beyond what that system is capable of. Just 48k RAM, no hardware acceleration for anything... But it's rated as one of the best games on the system, because it made changes to suit the machine and maintain the spirit of the original.

The GBA version of MS is the same.

Mybe parallax can be left, maybe using sptites on ocs. On Aga no problem to have it

[sandruzzo]

Now we could try to talk about how we could have a Great Amiga 500 version of Metal slug.

I would try different layout, but with 256*192 screen size. Even though it seems small, with that we have a lot free dma cycles to use to have a lot of moving object on screen.

We could try to have dualplayfield one for actuall gfx, and other for bobs. And if we interlace 2 background we can easly archive 16 colors, but we have to have stable 50hz