Could sprites have been 8 high and 256 across?

[+Random Terrain]

When all of this stuff was designed, could the bits of the bytes that make up sprites have been displayed vertically instead of horizontally? So instead of tall skinny sprites, we could have had short, long sprites? If so, could things have been designed to display them either way on the screen by setting a register?

[Rybags]

Not really. 256 across, you're pretty much talking a med/hi res playfield.

 

8 bits per the 2 players means less circuitry is needed inside the TIA. Also remembering the CPU has to load the graphics data manually on 2600 so in any case the 6507 is running nowhere near the speed to keep up with that sort of demand.

 

I don't know how TIA fares for it's time insofar as cramming transistors onto the given sized die but I imagine there was probably not a lot of room left.

Given the playfield is 40 pixels across and only 20 bits were provided tends to suggest that also.

But then again, having to only deal with that amount of data and option of reflect or repeat means the CPU load is reduced in that area in some circumstances.

Edited November 14, 2013 by Rybags

[Mr SQL]

When all of this stuff was designed, could the bits of the bytes that make up sprites have been displayed vertically instead of horizontally? So instead of tall skinny sprites, we could have had short, long sprites? If so, could things have been designed to display them either way on the screen by setting a register?

 

We can set them 8xN because we have N scanlines worth of time; however a simple and practical engineering solution makes "8 high by 256 across" possible with the existing hardware:

 

1. Get a PAL Television.

2. Turn it on it's side.

[Csonicgo]

I've always wondered why more games didn't try that approach. given how small my TV was that I played my atari on, I would be more than happy to turn it on its side.

The real problem is heat dispersion....

[+Gemintronic]

 

We can set them 8xN because we have N scanlines worth of time; however a simple and practical engineering solution makes "8 high by 256 across" possible with the existing hardware:

 

1. Get a PAL Television.

2. Turn it on it's side.

 

Last time I got a pal and turned him on his side he screamed and never came back.

[+Random Terrain]

Thanks for the replies.

[YANDMAN]

I've always wondered why more games didn't try that approach. given how small my TV was that I played my atari on, I would be more than happy to turn it on its side.

 

The real problem is heat dispersion....

Remove the case , Drill a bunch of holes, Re-fit the case

[alex_79]

When all of this stuff was designed, could the bits of the bytes that make up sprites have been displayed vertically instead of horizontally? So instead of tall skinny sprites, we could have had short, long sprites? If so, could things have been designed to display them either way on the screen by setting a register?

There's actually only 1 byte that makes up a sprite in the 2600 from the hardware point fo view. The "display area" the TIA is working with is just 1 scanline, so the sprites are monodimensional and you can't rotate anything in a one dimensional space. So there couldn't have been an hardware function to achieve that.

To display bidimensional graphic objects on the VCS you must update the registers for each scanline (so they can be as tall as the entire screen), but this is made in software.

Larger sprites would have required larger register (i.e. more bytes for each sprite) but this also means more cpu cycle wasted to update the graphics for each scanline.

 

Edited November 14, 2013 by alex_79

[SeaGtGruff]

We don't really have tall, skinny sprites on the 2600, at least not if you measure the sprite's height by how many pixels it has in the vertical direction. Everything on the 2600-- background, playfield, ball, player 0, missile 0, player 1, and missile 1-- is only 1 pixel tall (1 really tall pixel extending the full height of the screen). It *seems* like they've got more pixels vertically, but that's only because the programmers are doing mid-screen register changes every 1 or 2 scan lines.

Edited November 14, 2013 by SeaGtGruff

[tokumaru]

The hardware could certainly have been designed differently to allow ultra wide sprites (nothing would limit their height to 8 pixels, though), but that would have made the console more expensive. The main reason the 2600 was designed the way it was is that televisions draw pictures from top to bottom. This means that a console has just a little bit of video memory (barely enough to represent a single scanline worth of graphics) can get away with it, because that memory can be changed over time to generate different looking scanlines.

 

You couldn't really take advantage of the scanlines being drawn from left to right (like we do from the picture being drawn from top to bottom) because that would require really fast CPUs not available at the time, so the only way to have wider sprites would be to put more video memory in the system, which would have made it more expensive.

[roland p]

Maybe they could have been wider, by just repeating. Now the max width is 48px (2 x 3 copies). This is set with the D0, D1, D2 bits of the NUSIZx register. D3 is not used. Maybe D3 could have been used for 'endless copies'?