The Neo Geo was just a glorified Genesis.

[Joe Redifer]

Is there a Neo Geo game out there that line scrolls?

[StealthNinjaScyther]

Riding Hero, I think.

[Kamahl]

I understand that. It's just that I spent years hearing about SNES's mode 7 and Santurn's VDP2. I thought there were things you couldn't do without tiled backgrounds.

That depends on the capabilities of the specific VDP.

[Sik]

Is there any advantage to use tilemaps aside saving memory?

Saving money? Not on the player side, but on the developer side - the more you can reuse, the less you have to make, and thereby that reduces development costs. Since the whole point of tilemaps is repeating data, you can see how it can be useful

[Kamahl]

Saving money? Not on the player side, but on the developer side - the more you can reuse, the less you have to make, and thereby that reduces development costs. Since the whole point of tilemaps is repeating data, you can see how it can be useful

That's not true, nothing prevents you from copy-pasting tiles to draw a bigger image.
Even on bitmap displays, like on the Amiga, games just draw tiles in software.

[sheath]

If anything, I would think trying to reuse tiles rather than just draw new ones would take more thought and development time.

[Sik]

Even on bitmap displays, like on the Amiga, games just draw tiles in software.

Software tilemaps are still tilemaps...

If anything, I would think trying to reuse tiles rather than just draw new ones would take more thought and development time.

Drawing takes time though. If you know something you've drawn before would fit perfectly then why not just reuse it?

[Kamahl]

Software tilemaps are still tilemaps...

Not in the context of his question, he meant (I think, based on his mention of the VDPs), hardware tilemaps.

[sheath]

Drawing takes time though. If you know something you've drawn before would fit perfectly then why not just reuse it?

If I originally drew the tile to fit numerous ways I might think of that. Odds are I would come up with about as many new tile ideas in the process of trying to find another one that fits though.

[kool kitty89]

Saving money? Not on the player side, but on the developer side - the more you can reuse, the less you have to make, and thereby that reduces development costs. Since the whole point of tilemaps is repeating data, you can see how it can be useful

Not just from the developer's standpoint either, but from the publishing/manufacturing standpoint too. The more you make, the more you have to store, so the more ROM you need. (though optimizing for compression is another variable there)

So in the neo's case, you'd still be able to conserve ROM space since sprites are built from tiles anyway . . . but, since they're using 16x16 tiles, that would tend to allow less optimization than tilemaps (or sprites) using 8x8 cells (like the SNES and MD -PC Engine uses that for the BG too iirc, and several older systems).

In the Neo Geo's case, you need to leave all graphics uncompressed in ROM too since you don't load textures into VRAM (same with the NES), though some systems that do use RAM still might not allow for much compression. (the SMS has so little VRAM -relative to color depth- that decompressing a head of time is very limited and and limited CPU resource means little to no on-the-fly decompression -simply using fewer bitplanes would probably be the most practical)


And, as far as overall system design/engineering choices, there's more than just ROM cost effectiveness too. (but general system performance and cost trade-offs -as well as programmability, flexibility, and ease of programming)
And systems that use bitmap/blitter architectures could still have better ROM/mass-storage efficiency than tile/sprite based systems, but it would depend more on how much RAM and CPU/blitter resource was available (for buffering decompressed textures as well as blitting tiles and decompressing on the fly) . . . RAM efficiency tends to be less on those types of systems (at very least due to the framebuffer -but also depending on how the unpacked graphics are stored), but most of those systems also used cheaper RAM than contemporary consoles. (the Amiga used cheap/common DRAM compared to VRAM/PSRAM/SRAM used by many consoles of the late 80s/early 90s, plus it used a single-bus architecture -sans fastRAM- which would also cut cost significantly -not that it wouldn't also have been possible to design a tile/sprite based system around cheap DRAM on a unified bus -which is pretty much what the C64 and A8 chipsets did, though the A8 also supported bitmap modes; likewise, a blitter based system could be designed around using fast/exotic SRAM/VRAM/etc too)

If anything, I would think trying to reuse tiles rather than just draw new ones would take more thought and development time.

Optimizing for that takes skill just like anything else, but that sort of art design had been industry standard for a long time already, so it was a common and necessary skill to have for most arcade/console/computer game designers.

And as Sik points out, once you know what you're doing (for tile-oriented graphics), you can save a lot of time over drawing a fully unique scene too. (especially for good quality/detailed art)

I understand that. It's just that I spent years hearing about SNES's mode 7 and Santurn's VDP2. I thought there were things you couldn't do without tiled backgrounds.

There's no fundamental advantage for tilemaps as far as graphical effects are concerned (scaling or scaling/rotation or warping can be done equally well in framebuffer or line buffer/tile/sprite based architectures -not to mention other things like blending/translucency/shading effects).

There are other engineering, manufacturing, and performance trade-offs for those different types of systems (blitter/framebuffer takes more RAM -for the framebuffer- and takes more bandwidth to do some things, but has other flexibility; using all sprite or object based systems over tilemap playfields has flexibility advantages but could be harder to program for or take more chip space to manufacture -and/or take more space for lists/tables in RAM)

[Joe Redifer]

Not just from the developer's standpoint either, but from the publishing/manufacturing standpoint too.

Not just from a developer, publisher and manufacturer's standpoint, but from a shipping standpoint as well. You see, the more unique tiles you put in a game the more the game weighs. The more a game weighs, the more a game costs to ship. If you repeat a tile, it's weight does not increase with it, so the game gets a free tile which is not at all affected by gravity. Thus, if you shipped a game with only one unique tile but repeated 20,000 times and used for every graphical instance in a game, you could save upwards of $300 on shipping worldwide. And $300 can buy a pretty nice hooker!

[Sik]

Not just from a player, developer, publisher, manufacturer and shipping standpoint, but also from a pirate and a ISP standpoint! Using tiles reduces the filesize of the ROM and thereby copying the game will take less time, which results in reduced electricity costs at the very least, and possibly also reduced internet costs as well as better use of bandwidth caps. Moreover, the reduced filesize means it takes up less physical space in storage media, which could be used for something else.

And not just from a player, developer, publisher, manufacturer, shipping, pirate and ISP standpoint, but also an arcade owner standpoint. I mean, imagine if the player had to deal with completely different things as he advances and he keeps failing because he can't tell what it does! He will eventually quit in frustation! By making things predictable, you make the player keep coming back to the game and thereby you earn more money!

[andlabs]

And not just from a player, developer, publisher, manufacturer, shipping, pirate, ISP, and arcade owner standpoint, but also a consume standpoint! Recently leaked official Neo-Geo programming documentation points out that SNK intended MVS (the arcade) and Neo-Geo (the console — their terminology) to interact with each other: players used memory cards and rented [sic] the Neo-Geos to play their games at home and transfer their status to and from the arcade machines.

ALSO KEEP IN MIND: in the 80s, framebuffers were expensive: a 256x240 8bpp framebuffer with a palette would be 256*240 bytes of RAM — slightly under 64KB of video RAM (which has to be fast). 320*240 brings you over. And that's just for 8bpp — monochrome might be cheaper (I wonder how much a Blit, Macintosh, or something else cost to make back then). Tilemaps, on the other hand, take far less memory (it's just pointer indirection!), making them cheaper, but do require more transistor logic and can be slower... Again, I don't know any specific measures, but if someone (knowledgeable!) would like to back me up on this, please do. The Neo-Geo was developed in the late 80s (1988-1989 at least) and is probably based on older 68000-based arcade hardware made by SNK in the latter half of the decade, so... I don't think arcade systems moved away until the world of 3D stepped in :/

I'm guessing the Neo-Geo saved money by not having the same level of customizability as the Mega Drive: there's only one sprite plane and one tile plane and the tile plane is ALWAYS drawn over the sprite plane; sprites are vertical strips and a flag in VRAM determines whether or not horizontally adjacent sprites move together when scrolled; scaling is rather restricted (I believe you can only compress vertically up to 32 levels; the documentation is not well worded). (Neo-Geo cartridges are groups of PCBs and ROMs and the sound chip is a YM2610B, so I think SNK only made reservations on the video side.)

[Sik]

I'm guessing the Neo-Geo saved money by not having the same level of customizability as the Mega Drive

Except it didn't, ROMs were huge, which is the safest way to make the price skyrocket.

scaling is rather restricted (I believe you can only compress vertically up to 32 levels; the documentation is not well worded)

256 levels vertically, 16 levels horizontally.

[bpguimaraes23]

There's no fundamental advantage for tilemaps as far as graphical effects are concerned (scaling or scaling/rotation or warping can be done equally well in framebuffer or line buffer/tile/sprite based architectures -not to mention other things like blending/translucency/shading effects).

There are other engineering, manufacturing, and performance trade-offs for those different types of systems (blitter/framebuffer takes more RAM -for the framebuffer- and takes more bandwidth to do some things, but has other flexibility; using all sprite or object based systems over tilemap playfields has flexibility advantages but could be harder to program for or take more chip space to manufacture -and/or take more space for lists/tables in RAM)

Thanks! That's what I was curious about. The money part I understand.