Colecovision Vs. NES in graphics.

[NIAD]

 

That's because you probably grew up using your right hand while holding the magazine in your left...

Damn... you're right! It's been so long since I had to look at a magazine for THAT.

[KAZ]

Necrobump!

 

When talking about Donkey Kong (colecovision) vs. Donkey Kong (nes), I think it is VERY important to include this game for comparison:

 

I know it is unreleased, but that will change very soon (August or so)

 

 

I think this version of donkey kong is higher quality than Nintendo's version. One reason being that it is code-perfect for emulation. It has more animations also, such as after you complete the 1st stage opcode Colecovision's donkey kong shows Donkey Kong grabbing Pauline while going up the ladder.

Edited May 10, 2016 by KAZ

[Polybius]

CPU wise, back in 1997 - four years before the first MiniGame compo - there were a series of programming battles between the Commodore 64 (1 MHz 6510, running 6502 code) and the ZX Spectrum (3.58 MHz Z80) to see which computer/CPU would be most efficient on various tasks and algorithms, counted in clock cycles. If I understand the conclusion right, it ended with the C64 had a 3:1 advantage, which evens out due to the ZX running at a higher frequency. You can find some of the results here: http://www.ffd2.com/fridge/speccy/

 

Of course many of those programs utilize the vast RAM of respective computer, perhaps to a bigger extent on the C64 as the 6502 has much fewer registers than the Z80 has. The ColecoVision has the same 3.58 MHz Z80 and 1K RAM, while NES has a 6502 near clone from Ricoh running at 1,66 - 1,79 MHz and 2K RAM + possibility to add RAM on the cartridge. Properly coded, the NES would even do calculations a bit faster than the ColecoVision, ignoring everything that has to do with graphics and sound output.

 

It is a common mistake to just look at the MHz when comparing processors of different architectures. Just look at how Apple desperately tried to convince buyers that a 1+ GHz G4 was at least as powerful as a 2+ GHz Pentium 4, although the numbers themselves would suggest otherwise.

 

 

True, the big reason for the commodores advantage, despite the slower clock speed, was due to its graphics proccessor and a slew of other chips that worked alongside it, where Sinclair computers used the main processor for everything, one chip handled everything on screen. If Sinclair had additional chips for the graphics and text, it would have been a much faster machine than a lot of machines, but also vastly more expensive than it was.

[matthew180]

The F18A can give you NES graphics on your CV today. I actually modeled some of the F18A's enhance features on the NES design, mainly the multiple name table layout for scrolling, and the way patterns are stored to support 2-bit and 3-bit color. You can match the NES's capabilities exactly, or if you want you can have better than NES graphics.

[digress]

What about memory? Is all the extra graphic info stored on the f18a board. And if so does that then free up the regular coleco video ram for extra ram usage?

 

 

 

The F18A can give you NES graphics on your CV today. I actually modeled some of the F18A's enhance features on the NES design, mainly the multiple name table layout for scrolling, and the way patterns are stored to support 2-bit and 3-bit color. You can match the NES's capabilities exactly, or if you want you can have better than NES graphics.

[mozartpc27]

Yeah, Donkey Kong was ported to the Atari 5200 by some hackers a few years back. I had the game briefly, back in the days when I actually had the money to spend on such frivolities. It's pretty much identical to the Atari 400/800 game, but that makes it hugely improved over its ColecoVision counterpart. It's as ugly as a sack of butts, but it PLAYS so much better and feels so much more complete.

 

I've gotten into collecting classic computers (and to some degree systems, in that I have the Expansion module to play 2600 games on the ColecoVision - for which I use an ADAM these days), and this comment raises an interesting point to me. Whenever I am playing around with a new system, my go-to game is always to get Donkey Kong. I enjoy comparing the ports (the only classic system I had as a kid was CV, so that is sort of my "default").

 

I have a TI-99 and an Atari 800XL, and the DK port for both. Both were programmed by Atari. As good as the Atari 800 port is (it has the cement factory screen and some of the cut animation), the first screen is oriented like the CV, with Donkey Kong on the right side of the screen. This means there is one less girder to climb, and I assume this was done, as it was on the CV, since the expected orientation of the CRT screen for both was a television, 4:3, that is, wider than they were tall. The arcade game is of course taller than it is wide.

 

But the TI-99 port, which is mostly inferior to the Atari 800XL port, and which should also have been created with the expectation that the computer is reporting video out to a "standard" 4:3 screen, i.e., wider than it is tall, like all home systems that use TVs, nevertheless has DK on the "correct" left side of the opening screen, and Mario has the full number of girders to climb. Since Atari did both ports, there must have been a reason that wasn't just difference between manufacturer. My guess is it has something to do with screen resolution of the machines, but I haven't looked it up, and I'm not anyway convinced that alone would account for it. It's a mystery to me why Atari would have this difference in its ports of the same game to two different machines.

 

If nothing else, both of these ports underscore what a crappy job Coleco did porting the game to the 2600 and Intellivision.

[matthew180]

What about memory? Is all the extra graphic info stored on the f18a board. And if so does that then free up the regular coleco video ram for extra ram usage?

 

 

 

 

The F18A does not provide any memory over the original 16K of VRAM, except for 2K of GPU accessible RAM. The extra pattern data to provide more colors per tile or sprite must be stored in the 16K VRAM. However, the F18A does have options to reduce the stride between patterns, so if you are not using all 256 patterns for tiles or sprites, you can reduce the size and distance between the multiple tables, thus using less memory for patterns.

[Pixelboy]

But the TI-99 port, which is mostly inferior to the Atari 800XL port, and which should also have been created with the expectation that the computer is reporting video out to a "standard" 4:3 screen, i.e., wider than it is tall, like all home systems that use TVs, nevertheless has DK on the "correct" left side of the opening screen, and Mario has the full number of girders to climb. Since Atari did both ports, there must have been a reason that wasn't just difference between manufacturer. My guess is it has something to do with screen resolution of the machines, but I haven't looked it up, and I'm not anyway convinced that alone would account for it. It's a mystery to me why Atari would have this difference in its ports of the same game to two different machines.

You're looking for a technical or business reason for such differences but the answer is actually quite simple: These different versions were programmed by different people, each with their own background, programming habits, vision and varying levels of passion for the source material, and for target machines that had different capabilities and limitations. And these people probably worked in different buildings. Please remember that back in the early 80s, the computer world was very different. In many ways, it was like a "far west" with no rules (except that the games needed to be as "debugged" as possible) and no framework to make development easier. Also keep in mind that programmers, especially in that era of video game history, were pure artists. They carved video games out of bits and bytes just like sculptors carved statues out of large rocks. And each one had their own style. Would Pitfall have been the same if it had been programmed by someone else than David Crane? It would have looked different, and perhaps not as polished (or memorable) as it turned out to be.

 

If nothing else, both of these ports underscore what a crappy job Coleco did porting the game to the 2600 and Intellivision.

I agree they could have done a better job, but there's the issue right there: The word "job". I've seen awesome homebrew versions of DK done on the Atari 2600 in recent years, but these homebrewers benefitted from all the technical documentation and programming tricks available on the internet, and they had all the time they needed to tinker with their code, discuss their tech problems with friends over the net, and achieve the best result possible. Did the "commercial" programmers back in the early 80s have these luxuries? They definitely did not. They had to program their games on limited hardware, with limited tools, limited documentation, and limited development cycles (which had to include testing and debugging) with a hard release deadline.

 

Myself, I tend not to judge programmers for the jobs they did BITD. I judge the games themselves. If the games are good and memorable, then the programmers can reap the rewards. Otherwise, I don't blame the programmers, because I know they did what they could with the limited time, knowledge and resources they had. Again, programmers are artists, some artists are simply more talented than others, and the "best" artists are hard to come by, especially in an industry that was booming in an exponential way in its early days.

[Noah98]

Comparing the Opcode SGM version is not a fair comparison. Donkey Kong on the nes was a launch title using the NROM 128 board. If the game was released later in the console's life cycle on one of the more advanced mapper boards like the MMC3, it could look amazing. Especially, if it was coded 20 years later. I made a cartridge with the Donkey Kong Pie Factory rom for my nes, and it is hands down the best home version of Donkey Kong. I own the Colecovision version, and they aren't even close. I also own a Mame arcade machine, and there is no way the Colecovision versions of Donkey Kong and Donkey Kong jr. Look more like the arcade. I don't even think they look "cleaner" although maybe the screenshots gove a false impression.

Edited September 26, 2016 by Noah98

[matt2d2]

When the NES came out I remember not caring one bit. Lots of little kids had NES, but the good games for it came out later. Sure SMB was out not long after release, but aside from arcade ports, (which I had many of on other systems), not much on NES that I wanted to play until '87/88 was when stuff like Metroid, Tyson Punch Out, Castlevania 2 hit. I did enjoy some of the games in this era.

[atarilovesyou]

I'm not sure if the OP is being serious. If he was, had both systems, or lived through the era, he'd know: there is no comparison. The NES wins hands down on all levels. The Coleco couldn't scroll. I'm not here to make excuses, but you can't pick on the Coleco for not doing something that the NES could...the Coleco games had no need for scrolling, anyways (most of them). Looping is a painful example of what scrolling on the Coleco was like. Sure, we loved it then, but what did you have to compare it to? The sprites looked terrible in comparison. Look at Buck Rogers on the Coleco. Then look at something even like Exerion on the NES. The Coleco just didn't have the necessary power to do those types of games. I can't think of one game that the Coleco did better than the NES. But who cares? It was the king of systems until it was dethroned. Just like the NES was dethroned by the next wave of better technology, catering to the changes of gamers as the years went on. I know, and can speak to these issues, as I owned both and lived through both eras (but lacked the money to ever own much for either system, lol..thank goodness for trading amongst friends in those days, and rentals of course).

[kaz321123]

Play pac-man collection (pacman) on colecovision, then play the nes version of pac-man.

 

There is no question which is better. Pac-man on colecovision is code-perfect. It also does not require a super game module either.

 

I really wish someone would make a 'code-perfect' pac-man for nes, so I could better compare. There is a lot of flicker in coleco pac-man that I don't think nes would suffer from quite so drastically.

 

 

However, I am thinking that the colecovision isn't suited for a game like 'mike tyson's punchout' or 'super mario brothers'. Big lovely sprites. I wish I had a perfect side by side comparison of 'mega man 2' on both consoles. Those giant hopping things in mega man 2 might not be possible on coleco.

 

I love collecting for colecovision these days, the recent games really are arcade flawless (Mario Brothers for example). I wonder if 'code perfect' arcade games are possible on nes in fact. If you have an example of a home-brew that is code-perfect for nes, PLEASE tell me.

 

 

In the 80's, if colecovision released opcode's flawless donkey kong (unreleased), we might have an interesting discussion. Bottom line is: what system does better regardless of the era it was in? (and the games released at those times).

 

I would rather play Dintar's 8k pac-man for Atari 2600 than Nintendo's pac-man also (nes). I thought I'd put that in. The sound effects are amazing in that. It is far closer to arcade than nes.

Edited May 8, 2017 by kaz321123

[Noah98]

Play pac-man collection (pacman) on colecovision, then play the nes version of pac-man.

 

There is no question which is better. Pac-man on colecovision is code-perfect. It also does not require a super game module either.

 

I really wish someone would make a 'code-perfect' pac-man for nes, so I could better compare. There is a lot of flicker in coleco pac-man that I don't think nes would suffer from quite so drastically.

 

 

However, I am thinking that the colecovision isn't suited for a game like 'mike tyson's punchout' or 'super mario brothers'. Big lovely sprites. I wish I had a perfect side by side comparison of 'mega man 2' on both consoles. Those giant hopping things in mega man 2 might not be possible on coleco.

 

I love collecting for colecovision these days, the recent games really are arcade flawless (Mario Brothers for example). I wonder if 'code perfect' arcade games are possible on nes in fact. If you have an example of a home-brew that is code-perfect for nes, PLEASE tell me.

 

 

In the 80's, if colecovision released opcode's flawless donkey kong (unreleased), we might have an interesting discussion. Bottom line is: what system does better regardless of the era it was in? (and the games released at those times).

 

I would rather play Dintar's 8k pac-man for Atari 2600 than Nintendo's pac-man also (nes). I thought I'd put that in. The sound effects are amazing in that. It is far closer to arcade than nes.

Pac-man for nes came out in 1984, so it's not fair to compare it to the 2008 Colecovision version. Also, it's clear that Pac Man on the nes wasn't coded that well. How a game is coded does matter. Look at the Tengen version of Ms. Pac Man for the nes vs. the Namco version. Or, look at how much better Ms. Pac Man is for the 2600 vs. the original Pac Man.

 

A fair comparison would be versions of games that came out around the same time in the same era. Examples: Galaga / Rampage / Double Dragon/ Xevious for the 7800 vs. the nes. Colecovision Gorf vs. 5200 Gorf.

[youki]

Pac-man for nes came out in 1984, so it's not fair to compare it to the 2008 Colecovision version. Also, it's clear that Pac Man on the nes wasn't coded that well. How a game is coded does matter. Look at the Tengen version of Ms. Pac Man for the nes vs. the Namco version. Or, look at how much better Ms. Pac Man is for the 2600 vs. the original Pac Man.

 

A fair comparison would be versions of games that came out around the same time in the same era. Examples: Galaga / Rampage / Double Dragon/ Xevious for the 7800 vs. the nes. Colecovision Gorf vs. 5200 Gorf.

 

you can compare with PACMAN (Atarisoft version) for Colecovsion. It never has been commercially realised, but the prototype is fully playable and has been done in 1982 or 83 . That is a really good version without flickering. (but not arcade Perfect code as is the opcode version).

 

 

 

I love collecting for colecovision these days, the recent games really are arcade flawless (Mario Brothers for example). I wonder if 'code perfect' arcade games are possible on nes in fact. If you have an example of a home-brew that is code-perfect for nes, PLEASE tell me.

 

 

 

code perfect arcade games on NES would be possible , if the source arcade game you "port" use a 6502 processor. Pacman (opcode version) is code perfect because he re-used the original code that was already running on a Z80 , just the graphic rendering and sound has been adapted to the colecovision hardware.

 

A code perfect PACMAN on NES would be very difficult as it is not the same processor. you recode the same behavior in 6502 , even if the final result will be very close it won't be code perfect.

[+darryl1970]

Yeah, Donkey Kong was ported to the Atari 5200 by some hackers a few years back. I had the game briefly, back in the days when I actually had the money to spend on such frivolities. It's pretty much identical to the Atari 400/800 game, but that makes it hugely improved over its ColecoVision counterpart. It's as ugly as a sack of butts, but it PLAYS so much better and feels so much more complete.

 

By the way, the observation that the ColecoVision version of Donkey Kong was "one of the best" only demonstrates just how lacking all the home console conversions of the game really were. Have you ever stopped and asked yourself why homebrewers and even Nintendo itself have released new ports of Donkey Kong for the ColecoVision, 2600, and NES in the 21st century? The answer's pretty clear: the originals on these three systems fell far short of arcade perfection and were in dire need of revision.

Update: http://atariage.com/forums/topic/249885-atari-a8-donkey-kong-hack/?p=3497095

Edited May 9, 2017 by darryl1970

[zzip]

Update: http://atariage.com/forums/topic/249885-atari-a8-donkey-kong-hack/?p=3497095

In 1982, it was amazing because nothing had looked so close to the arcade graphics before. But I agree it doesn't play that well and the Atari 8-bit version outshines it, even if the graphics are a bit less faithful.

[Major_Tom_coming_home]

This is a very interesting point to me, even though it was made 5 years ago

 

...I agree that It wasn't hardware capabilities that made the NES the more successful console. IMO On many it came due to innovation: The flagship and pack-in game for the Colecovision was an arcade port. It was a great port, but it was yet another attempt at selling systems based on what was popular in the arcades. The Pack-in for the NES most people got was Super Mario Brothers and holy shit...it blew away preconceived notions about what a video game could be. Sure there were platformers before SMB, but they could not compare or compete. Nintendo had arcade ports available, but it also provided us with an experiences that weren't anything like the arcade or what was available at home at the time. Also, I think people around my age would say the scrolling capability on the NES was a huge deal. AFAIK no other popular console or computer could scroll like a Super Mario, and it was a game changer. Then there were the controls. The 5200, Intellivision, Coleco, and even Emerson Arcadia 2001 all had very similar controllers that were not comfortable for a lot of people, were not usually well made, and could actually be difficult to use with their shitty keypad you needed to look at instead of the screen. The NES somehow got by with 4 buttons including start and select and could be used by feel, and the directional pad was superior to a joystick for 95% of games. Nintendo also innovated the design of their console, but the engineering was very flawed and I have to give that point to Coleco. Nintendo got it right otherwise, and changed everything in the process. Gaming at home was no longer all about arcade games and ports of arcade games at home, instead 'home'. replaced the arcade as the focus.

I personally like the colecovision screenshot of donkey kong better than the nes version. I think the dk jr. screenshot of the nes version looks better than the colecovision. The wii and ps3 have competed against each other for years. The ps3 is capable of better graphics than the wii. Both systems have great games and are better at different things. I see no reason that both the nes and colecovision couldnt have competed against each other as well.

 

[opcode]

A more in deep technical analysis:

Famicom video was inspired by the CV Texas video chip. They use similar concepts. And in fact they have the exact same video memory bandwidth. So why is the Famicom better? Because all the sprites attributes are mapped inside the chip instead of in memory. That saves a lot of memory bandwidth that is then used to improve sprites mostly.

 

Then we have hardware scroll.

 

But the secret weapon is using VROM inside cartridges instead of VRAM inside the console. That allows for instantaneous graphic bank switching, producing very complex animations not possible otherwise.

 

 

Sent from my iPhone using Tapatalk

[mr_me]

The sprites in both the nes and coleco vision are similarly limited with 64 total pixels per scanline. Coleco vision sprites are 16px wide where nes sprites 8px. Therefore the nes can have twice as many separate but smaller sprites per scanline. They are both similarly limited with sprite flicker. That's how they are similar.

 

The famicom/nes and coleco vision came out about the same time. Coleco like sega in japan used existing technology which was several years old. Nintendo improved it, like others have said, adding more colours and hardware scrolling. And remarkably, Nintendo did it with less RAM as Opcode explains. An innovation possibly resulting from cost constraints. It's Nintendo's R&D that stands out over others.

 

Further Coleco depended on others' IP for their games where Nintendo created their own. Nothing wrong with going with existing arcade IP in 1982. Nintendo did the same thing in 1983, arcade games were hot. But when the arcades died a couple of years later Nintendo came up with new game ideas. You couldn't just depend on third parties for games. Prices in computer memory were continually coming down every year in the 1980's. It's what allowed the coleco vision and Atari 5200 to happen and then later, the larger rom cartridges for games with more content like super mario brothers.

Edited June 2, 2018 by mr_me

[opcode]

I could do the math here showing how the saved memory cycles from sprites attributes helped to significantly improve the famicom graphics, but that probably would bore some.

 

And indeed, Nintendo transferred to high costs of VRAM into VROM, producing a console that has only 4KB of RAM total and yet is very efficient and inexpensive to produce.

 

 

Sent from my iPhone using Tapatalk

[phoenixdownita]

...

 

And indeed, Nintendo transferred to high costs of VRAM into VROM, producing a console that has only 4KB of RAM total and yet is very efficient and inexpensive to produce.

 

 

Sent from my iPhone using Tapatalk

Why didn't Coleco (or Sega for that matter) replaced the VRAM with a bus on the cart slot to allow all sorts of fancy features? (including VRAM if need be)

Is the VRAM interface of the TMS9928 that hard to work with that it would not make sense?

Is the 9928 ever writing to VRAM when not instructed by the control registers under CPU control?

If not then ROMs could be used to replace the VRAM, right? Aside from the pesky RAS to CAS decoding if the VDP does not write anything in VRAM outside the control of the CPU replacing VRAM with bankable ROMs would do the trick.

I understand it would represent a monumental hack as one can tell the 9928 was designed with VRAM in mind but hey!

 

Another scheme would be to have dual ported VRAM to allow direct injection from a secondary bus (on the cart) but that would have been likely too expensive way back when (the 9928 does not seem to allow for a VRAM arbiter so that a simple inhibit signal would get it off/tristate that bus).

 

EDIT: for computer usage VRAM was not replaceable as usually the "load" medium was tape or disk hence one needed actual RAM/VRAM but for consoles whose support was carts with mask roms it is/was a viable path. Keep in mind using ROMs would preclude all games that requires CPU based rendering (as opposed to pre-drawn gfx) .... that is to say no Doom

 

[opcode]

Why didn't Coleco (or Sega for that matter) replaced the VRAM with a bus on the cart slot to allow all sorts of fancy features? (including VRAM if need be)

Is the VRAM interface of the TMS9928 that hard to work with that it would not make sense?

Is the 9928 ever writing to VRAM when not instructed by the control registers under CPU control?

If not then ROMs could be used to replace the VRAM, right? Aside from the pesky RAS to CAS decoding if the VDP does not write anything in VRAM outside the control of the CPU replacing VRAM with bankable ROMs would do the trick.

I understand it would represent a monumental hack as one can tell the 9928 was designed with VRAM in mind but hey!

 

Another scheme would be to have dual ported VRAM to allow direct injection from a secondary bus (on the cart) but that would have been likely too expensive way back when (the 9928 does not seem to allow for a VRAM arbiter so that a simple inhibit signal would get it off/tristate that bus).

 

EDIT: for computer usage VRAM was not replaceable as usually the "load" medium was tape or disk hence one needed actual RAM/VRAM but for consoles whose support was carts with mask roms it is/was a viable path. Keep in mind this would preclude all games that requires rendering (as opposed to pre-drawn gfx) .... that is to say no Doom

 

Those are interesting ideas I never actually considered. Thanks for sharing!

 

 

Sent from my iPhone using Tapatalk

[Tursi]

Why didn't Coleco (or Sega for that matter) replaced the VRAM with a bus on the cart slot to allow all sorts of fancy features? (including VRAM if need be)

Is the VRAM interface of the TMS9928 that hard to work with that it would not make sense?

Is the 9928 ever writing to VRAM when not instructed by the control registers under CPU control?

If not then ROMs could be used to replace the VRAM, right? Aside from the pesky RAS to CAS decoding if the VDP does not write anything in VRAM outside the control of the CPU replacing VRAM with bankable ROMs would do the trick.

 

That's more or less how the NES worked. You can do a lot when the video memory is on the cartridge.

[phoenixdownita]

That's more or less how the NES worked. You can do a lot when the video memory is on the cartridge.

I know, I was just wondering why the other 2 (Coleco/Sega) didn't go for the same pattern. My guess is that the VRAM bus of the 9928 was too DRAM centered at the time and they didn't want to invest in glue circuitry to hide that and present it to the cart devs like a simple ROM bus (no RAS to CAS mess exposed to the cart) and they didn't want to take on the task to create a mapper chip for it either.

Wild speculation at this stage, we'll likely never know for sure, it was clearly the easy path as that was what the chip designers had in mind at least for basing a computer design on, I am not sure TI thought it would be used in home consoles.

[Tursi]

I know, I was just wondering why the other 2 (Coleco/Sega) didn't go for the same pattern. My guess is that the VRAM bus of the 9928 was too DRAM centered at the time and they didn't want to invest in glue circuitry to hide that and present it to the cart devs like a simple ROM bus (no RAS to CAS mess exposed to the cart) and they didn't want to take on the task to create a mapper chip for it either.

Wild speculation at this stage, we'll likely never know for sure, it was clearly the easy path as that was what the chip designers had in mind at least for basing a computer design on, I am not sure TI thought it would be used in home consoles.

 

Yeah, apologies, I only just read the earlier messages now.

 

I think the main reason was just that nobody had done it before. TI's documentation says "hook the VDP up to dynamic RAM using this circuit", and that's what designers did. Like you suggest, doing anything else would mean taking on the technical risk yourself rather than using the circuit that the manufacturer would support. But I would wonder if anyone even considered it.

 

A secondary constraint would be the wider cartridge port needed, possibly adding cost...

 

We can certainly figure out how it could have been done, even how it might have been cheaper to build. But I'd put my money on 'nobody thought of it'. I certainly wish they had, once I learned how the NES pulls off some of its magic I was seriously jealous.