Atari v Commodore

[irwin]

Atari: 160x119 - 256 colors - interlace

but

I find this:

C64: 320x200 - 16 colors and dithering - and no interlace!

or

C64: 296x200 - 16 colors and dithering - interlace

 

 

Looks great, I've never seen before. But 1st image haven't 16 colors, maybe you can send the exact C64 pictures, or better send a link to the executable to view on the real C64.

 

Yes, 1st image - only 14 colors.

I attach image:

dreamworld.zip

Edited April 8, 2009 by irwin

[atariksi]

I don't see how "more apparent colors" is a rational statement. More apparent colors than what? You only get more than 16 colors if you use interlace and if you employ interlace on Atari, you get many more combinations.

 

I provided two links that show it is rational. Yes the C64 can only output 16 actual colors from it's hardware and yes PAL and NTSC quirks and page flipping can be used to generate the appearance of more colors than that. And yes I know that if versions of those techniques are employed on an A8 then the apparent color depth goes up past anything a C64 will be able to do.

 

Nonetheless without extensive register diddling, the A8 320x200 screen has count 'em two colors. This color resolution can be vastly increased vertically but one still only has two colors per scanline. Tricks to get it up horizontally are at least as onerous as the C-64 restrictions on their modes if not more so. Indeed, software mode 320x200 pix (C8 mode say) aren't terribly common.

...

 

My first point was there is no 320*200*16 mode on C64-- that's an exaggeration. Both machines have a 320*200*2 mode. They use various methods to enhance the total colors. So complete the statement "more apparent colors" than what?

 

>I think making ANY of these old 8-bits exceed what their original designers thought possible is very very cool.

 

Yes, that's good thing.

[atariksi]

I think my original point fully valid: The C64 can do some amazing things in the right hands and it won't kill us to admit it. And I've already said that I prefer A8s over Commodores but that isn't a need to ensure the A8 must be the winner of every possible comparison.

 

 

The same is true of the Atari 8 bit. In the right hands some amazing things can be achieved with the amazingly flexible hardware of the Atari 8 bit. It is just that back in the day, more often than not, the Atari was in the hands of lazy?/time/cost constrained developers content only to use the Atari's vanilla setup and code for the lowest common denominator 16k machine.

 

Frogstar stated what you replied to not me. I think you got the "/quote"s mixed up.

 

Atari has more flexibile hardware overall so however creative one is it's upper limit is what the hardware will allow for (documented or undocumented).

[atariksi]

i like those TIP pictures with lot of colors.

Atari: 160x119 - 256 colors - interlace

but

I find this:

C64: 320x200 - 16 colors and dithering - and no interlace!

or

C64: 296x200 - 16 colors and dithering - interlace

 

Actually, temporal dithering (interlace) at 25/30Hz does generate new colors and is different from spatial dithering. So in the Atari case you get 256*256 combinations of colors minus some redundancies and minus high flicker combinations. For example in GTIA mode (graphics 9), showing shade 1/16 on even frames and shade 2/16 at odd frames produces shade 1.5/16 w/minimal flicker. So I think that 256 colors should actually be more above for Atari and C64 should also get some more using interlace.

[atariksi]

Ok, some one list some specs:

 

Number of sprites per scanline (hardware and lowest color count - i.e not pairing)?

Number of sprites per screen?

If the total number of sprites can be extended (by changing the registers on the fly), what's the CPU resource ratio?

What res are the sprites available in?

Can you mixed hi-res and low-res sprites, and can you mix them with a different background res?

How many colors per sprite (both for pairing or non pairing)?

Native/max sprite sizes?

 

In the past 100 or so posts, we just discussed most of the above. Atari is always going to win when it comes to extending the sprites via replication-- it can set all sprite data base via a single register and it's sprites are full screen height so you only have to worry about HPOS setting. C64 has the higher resolution sprites (horizontally) and more horizontal coverage by default w/o any CPU tricks-- 24*21 in monochrome or 12*21 in 3 color mode. Resolution in Y direction is the same for both. Atari has 4 players and 4 missiles while C64 has 8 players called sprites (not 7-up).

 

Total number of sprites per screen, Atari wins since you can have thousands of 8*1 sprites flooding the entire screen including overscan using less cycles since it's sprites are 8*248 and missiles are 2*248. Once you get into overscan screen modes, Atari gets more of an advantage. You can read more if you skim through previous few pages...

 

thousands??? now we are calling every scanline of a player stripe a sprite? come on... and of course our superfast cpu can move around "thousands" of sprites in 50/60 frames... i have done this several times in more than 20 years... called "stafield" effect...

 

How many sprites did you use for the stars?

 

He was asking max possible number of sprites on screen so technically 8*1 from each player (5 players) * 240 visible scanlines = 1200 sprites and by setting HPOS multiple times for various players every scan line you get >2000 sprites easily. I am using it for image enhancement (Antic mode K) so it's useful for something.

[atariksi]

Yeah - IFLI mode looks incredible. It seems to be their version of our HIP mode (higher res of course).

 

 

If we learn to fulfill complete sentences, the upper one could be written as:

 

Yeah - IFLI mode looks incredible. It seems to be their version of our HIP mode (higher res of course but less colours).

 

 

But. IFLI is interlace. It means that the shown picture is not what the hardware produces for real.

With interlace you could do 256 colours in hires on the A8...

 

For pictures, very high color content makes up for half the resolution so I would just go with having one plane of 80*200*16 (with replication of GITA mode 9/11 as per required for image at hand) interlaced with 80*200*16 (mode 10) w/color clock shift w/two kernels running full length of the screen. That gives you 256*128 combinations of colors - redudancies as explained before. Then make a second pass through the image and take whichever scanlines have low frequency data and replace those lines with Graphics mode 9/11 instead of 10 and those lines increase color content to 256*256 combinations - redundancies as explained.

[Fröhn]

Total number of sprites per screen, Atari wins since you can have thousands of 8*1 sprites flooding the entire screen including overscan using less cycles since it's sprites are 8*248 and missiles are 2*248.

The same trick can be applied to C64 sprites aswell.

[atariksi]

Total number of sprites per screen, Atari wins since you can have thousands of 8*1 sprites flooding the entire screen including overscan using less cycles since it's sprites are 8*248 and missiles are 2*248.

The same trick can be applied to C64 sprites aswell.

 

You are more restricted since your HPOS requires 0..319 (two registers) to set in general and you also have to keep setting Y every 20 or so scanlines and you have a slower CPU (less cycles during at least during HBLank/VBlank) and your overscan mode is more problematic.

[atariksi]

So don't tell me it's always the same bandwidth/clocking.

 

Well, you can't tell me why the clocking should be different- just that you believe it to be so. Let's see a cycle by cycle breakdown of what happens on the bus and where the savings are. I am an electronic engineer by trade and have never encountered this planar clocking advantage.

 

Planar data doesn't make the ST read data any faster and it doesn't make the Amiga read data any faster (after all, we can calculate the size of the screen and the cycles used and it all adds up as it should), but apparently it makes old PC cards defy physics.

 

Yeah, you cannot rearrange the data to create bandwidth (obviously) but if your circuit is set up to handle multiple bitplanes, planar is superior.

 

If your circuit is set up to handle planar, then the only difference is how you program for it. There are some software tricks that are more easily done on planar data and there are many things that are easier with chunky data. That's the only performance difference you will ever see.

 

Anyway, until you present me with a technical description of clocking for chunky vs. planar, I am done with this discussion.

 

I'll have to look for a schematic; but if you look at EGA/VGA programming you can see that 3c0, index 12 lets you enable/disable bitplanes (0..3) and mode 0f/10h select the 640*350 monochrome, 2 or 4 bitplanes depending on memory in the video card....

[Fröhn]

You are more restricted since your HPOS requires 0..319 (two registers) to set in general and you also have to keep setting Y every 20 or so scanlines and you have a slower CPU (less cycles during at least during HBLank/VBlank) and your overscan mode is more problematic.

2 registers if you only split one sprite, but if you split more that extra register still only needs to be written 1 time for all sprites. And concerning Y-positions: You can stretch C64 sprites to full screen height. Since that X-position splitting isn't much useful for anything else than moving dots (starfield) you can simply stretch 1 sprite rasterline of a dot to full height. Lot's of demos do that.

[+Stephen]

i like those TIP pictures with lot of colors.

 

 

Where did you get those? I thought I had a full collection, but I haven't seen those before.

 

Stephen Anderson

[irwin]

i like those TIP pictures with lot of colors.

 

 

Where did you get those? I thought I had a full collection, but I haven't seen those before.

 

Stephen Anderson

 

http://tinyurl.com/dkf5os

[youki]

Prove that A800 is better than C64 or the opposite...

 

http://www.atariage.com/forums/index.php?s...t&p=1722385

 

[atariksi]

You are more restricted since your HPOS requires 0..319 (two registers) to set in general and you also have to keep setting Y every 20 or so scanlines and you have a slower CPU (less cycles during at least during HBLank/VBlank) and your overscan mode is more problematic.

2 registers if you only split one sprite, but if you split more that extra register still only needs to be written 1 time for all sprites. And concerning Y-positions: You can stretch C64 sprites to full screen height. Since that X-position splitting isn't much useful for anything else than moving dots (starfield) you can simply stretch 1 sprite rasterline of a dot to full height. Lot's of demos do that.

 

Actually, the thousands of sprites I was talking about were using new data every line so stretching sprites to full screen height would be a different animal. However, what register are you talking about on C64 that behaves like GRAFn registers of replicating sprite data to full height of screen?

[Rybags]

You can reset/set a sprite's bit in the Y-expand register and Vic will think it's still on the first instance of the line.

 

So, cost = CPU cycles for every line you want to do it. Benefit = good for stretch effects.

For a simple large object, it's obviously cheaper to just change Y-Pos and the data pointer.

[Fröhn]

Actually, the thousands of sprites I was talking about were using new data every line so stretching sprites to full screen height would be a different animal.

Sprite data every new rasterline is just one LDA STA in 21 rasterlines to multiplex the sprite.

 

Btw, you ignore another problem of your "thousands of sprites": Since you modify several sprite X-positions per rasterline, most of these X-pos changes will be in the middle of the rasterline and cause bugs if you move those stripes over it.

[R6502A]

I think my original point fully valid: The C64 can do some amazing things in the right hands and it won't kill us to admit it. And I've already said that I prefer A8s over Commodores but that isn't a need to ensure the A8 must be the winner of every possible comparison.

 

 

The same is true of the Atari 8 bit. In the right hands some amazing things can be achieved with the amazingly flexible hardware of the Atari 8 bit. It is just that back in the day, more often than not, the Atari was in the hands of lazy?/time/cost constrained developers content only to use the Atari's vanilla setup and code for the lowest common denominator 16k machine.

 

Frogstar stated what you replied to not me. I think you got the "/quote"s mixed up.

 

Atari has more flexibile hardware overall so however creative one is it's upper limit is what the hardware will allow for (documented or undocumented).

 

Oops! Sorry.

[atariksi]

Prove that A800 is better than C64 or the opposite...

 

http://www.atariage.com/forums/index.php?s...t&p=1722385

 

 

No reason to call the discussion here sterile-- it has been a lot more productive, informative, etc. than some other previous threads I have read. Your approach of experimental knowledge is inferior to a rational, logical (deductive) approach. You may make a game that is superior on Atari 8-bit now but later some one may make a better one on C64-- so it's just a relative truth rather than an absolute one. By taking a rational deductive approach as to what is achievable by the machine's chipset, the truth is established once and for all and there's no need for any change regardless of what software may be produced by the machines in the future. Going by the experimental approach is the reason why many mistakenly think C64 is the better machine-- they observe much more software with various features on C64 than they do on A8 (mostly due to non-hardware reasons).

[youki]

Prove that A800 is better than C64 or the opposite...

 

http://www.atariage.com/forums/index.php?s...t&p=1722385

 

 

No reason to call the discussion here sterile-- it has been a lot more productive, informative, etc. than some other previous threads I have read. Your approach of experimental knowledge is inferior to a rational, logical (deductive) approach. You may make a game that is superior on Atari 8-bit now but later some one may make a better one on C64-- so it's just a relative truth rather than an absolute one. By taking a rational deductive approach as to what is achievable by the machine's chipset, the truth is established once and for all and there's no need for any change regardless of what software may be produced by the machines in the future. Going by the experimental approach is the reason why many mistakenly think C64 is the better machine-- they observe much more software with various features on C64 than they do on A8 (mostly due to non-hardware reasons).

 

 

I disagree the only way to prove that one is better than the other is to experiment. Theory is often not the truth.

 

So take a game do it on both (keeping in mind to do show your machine is the best) and let 's compare.

In that specific case. There is no time limit, commercial factor, political or economical reason that will make one version being inferior due to externel constraint .

The only factor except the hardware iwould be the skill of the programmer.

 

But get your best programmer here for the Atari and the best for the c64 . Seing your discussion it seems very is very good expert in both camp here.

 

so let start with the vertical shooter i propose. and then let's go with an horizontal one. and then may be a kind of 3d game.

 

And i think we will have a fair view of what machine is the best. no blabla , no theory , just fact.

 

 

But i agree we learned lot of thing in this thread. But surely not what is the best.

[atariksi]

Actually, the thousands of sprites I was talking about were using new data every line so stretching sprites to full screen height would be a different animal.

Sprite data every new rasterline is just one LDA STA in 21 rasterlines to multiplex the sprite.

 

Btw, you ignore another problem of your "thousands of sprites": Since you modify several sprite X-positions per rasterline, most of these X-pos changes will be in the middle of the rasterline and cause bugs if you move those stripes over it.

 

Original question was which machine can show the maximum number of sprites. And on Atari you can show 2000+ sprites simultaneously spread out all over the screen (including overscan) even if some are restricted somewhat on left or right side of screen. This restriction is bigger on C64. And then C64 also has to set Y positions and shape ptrs every few scanlines. This is time where C64 won't be able to replicate more sprites horizontally. Now it gets worse when C64 enables overscan and running into 16K graphic limit isn't the only obstacle.

[Fröhn]

So take a game do it on both (keeping in mind to do show your machine is the best) and let 's compare.

In that specific case. There is no time limit, commercial factor, political or economical reason that will make one version being inferior due to externel constraint .

The only factor except the hardware iwould be the skill of the programmer.

I doubt that anyone is willing to spend months of coding on this. Even a graphical demo showing a bit sprites + background gfx movement similar to the game would need some work.

 

so let start with the vertical shooter i propose. and then let's go with an horizontal one. and then may be a kind of 3d game.

Why not take the existing ones?

for example.

[jvas]

Prove that A800 is better than C64 or the opposite...

 

http://www.atariage.com/forums/index.php?s...t&p=1722385

 

 

No reason to call the discussion here sterile-- it has been a lot more productive, informative, etc. than some other previous threads I have read. Your approach of experimental knowledge is inferior to a rational, logical (deductive) approach. You may make a game that is superior on Atari 8-bit now but later some one may make a better one on C64-- so it's just a relative truth rather than an absolute one. By taking a rational deductive approach as to what is achievable by the machine's chipset, the truth is established once and for all and there's no need for any change regardless of what software may be produced by the machines in the future. Going by the experimental approach is the reason why many mistakenly think C64 is the better machine-- they observe much more software with various features on C64 than they do on A8 (mostly due to non-hardware reasons).

 

 

I disagree the only way to prove that one is better than the other is to experiment. Theory is often not the truth.

 

So take a game do it on both (keeping in mind to do show your machine is the best) and let 's compare.

In that specific case. There is no time limit, commercial factor, political or economical reason that will make one version being inferior due to externel constraint .

The only factor except the hardware iwould be the skill of the programmer.

 

But get your best programmer here for the Atari and the best for the c64 . Seing your discussion it seems very is very good expert in both camp here.

 

so let start with the vertical shooter i propose. and then let's go with an horizontal one. and then may be a kind of 3d game.

 

And i think we will have a fair view of what machine is the best. no blabla , no theory , just fact.

 

 

But i agree we learned lot of thing in this thread. But surely not what is the best.

 

What do you think is better:

low res with high color depth

or

high res with low color depth?

 

The result will be subjective. There won't be an ultimate answer.

[Fröhn]

Original question was which machine can show the maximum number of sprites. And on Atari you can show 2000+ sprites simultaneously spread out all over the screen (including overscan) even if some are restricted somewhat on left or right side of screen. This restriction is bigger on C64. And then C64 also has to set Y positions and shape ptrs every few scanlines. This is time where C64 won't be able to replicate more sprites horizontally. Now it gets worse when C64 enables overscan and running into 16K graphic limit isn't the only obstacle.

I don't see the point in this? You cannot call those 8x1/2x1 stripes "sprites". They are useless. If you have 2000 of them, you don't have the time to actually move them. And even if you do move them, you cannot move them freely in X because of the X-position update bugs, you cannot move them in Y at all, half of them are just 1 or 2 pixel wide dots etc etc.

[atariksi]

Original question was which machine can show the maximum number of sprites. And on Atari you can show 2000+ sprites simultaneously spread out all over the screen (including overscan) even if some are restricted somewhat on left or right side of screen. This restriction is bigger on C64. And then C64 also has to set Y positions and shape ptrs every few scanlines. This is time where C64 won't be able to replicate more sprites horizontally. Now it gets worse when C64 enables overscan and running into 16K graphic limit isn't the only obstacle.

I don't see the point in this? You cannot call those 8x1/2x1 stripes "sprites". They are useless. If you have 2000 of them, you don't have the time to actually move them. And even if you do move them, you cannot move them freely in X because of the X-position update bugs, you cannot move them in Y at all, half of them are just 1 or 2 pixel wide dots etc etc.

 

You don't see your own point. Your first reply stated C64 can ALSO do the same. Now, you first admit that it can't before changing the subject. They can all be 8*1 using zoom on the missiles or they can be other combinations: 4*1, 2*1, 16*1, 32*1. They all move in x-direction fine and depending on use there's no restriction. At least 5 get updated during HBLank. Don't label them as bugged. You know the restrictions so you design the product accordingly. They can also move in y-axis during VBlank time.

 

You are now trying to promote the grapes are sour philosophy because you LOST the argument. "They are useless" YOU SHOULD HAVE STATED THAT FIRST and not stated that C64 can also do that.

[atariksi]

Prove that A800 is better than C64 or the opposite...

 

http://www.atariage.com/forums/index.php?s...t&p=1722385

 

 

No reason to call the discussion here sterile-- it has been a lot more productive, informative, etc. than some other previous threads I have read. Your approach of experimental knowledge is inferior to a rational, logical (deductive) approach. You may make a game that is superior on Atari 8-bit now but later some one may make a better one on C64-- so it's just a relative truth rather than an absolute one. By taking a rational deductive approach as to what is achievable by the machine's chipset, the truth is established once and for all and there's no need for any change regardless of what software may be produced by the machines in the future. Going by the experimental approach is the reason why many mistakenly think C64 is the better machine-- they observe much more software with various features on C64 than they do on A8 (mostly due to non-hardware reasons).

 

 

I disagree the only way to prove that one is better than the other is to experiment. Theory is often not the truth.

 

So take a game do it on both (keeping in mind to do show your machine is the best) and let 's compare.

In that specific case. There is no time limit, commercial factor, political or economical reason that will make one version being inferior due to externel constraint .

The only factor except the hardware iwould be the skill of the programmer.

 

But get your best programmer here for the Atari and the best for the c64 . Seing your discussion it seems very is very good expert in both camp here.

 

so let start with the vertical shooter i propose. and then let's go with an horizontal one. and then may be a kind of 3d game.

 

And i think we will have a fair view of what machine is the best. no blabla , no theory , just fact.

 

 

But i agree we learned lot of thing in this thread. But surely not what is the best.

 

What do you think is better:

low res with high color depth

or

high res with low color depth?

 

The result will be subjective. There won't be an ultimate answer.

 

That's your subjective view.

1.79Mhz > 1.0Mhz unless you need to extract the 1.0Mhz crystal from your C64 for your other projects.

256 colors > 16 colors unless you are color blind.

BYTE wide joystick reads are better than 4-bit reads unless you are interfacing to 4-bit bus devices (when it doesn't matter).

4 DACs > 1 DAC unless you are deaf.

Having 56+ graphics modes is better than having a graphics mode that's an extension of a text mode.

Having a Dlist to do scrolling, bitmap ptrs, etc. is better than using a register to set the mode for the entire screen.

Having overscan with one register setting is better than wasting many CPU cycles to set it up and harder to even address the pixels.

etc.

etc.

 

Don't make absolute statements like "There won't be an ultimate answer" since that's self-contradictory.