Missing Display modes

[Gorf]

My question here is, what display modes the Amiga Chipset is missing.
In the sense of: what could have been implemented back than without much trouble/changes to the rest of the design and could be provided now by an emulator or FPGA.

(I know: this is probably not useful and the Amiga is what it is - this is just a thought experiment.)

My Ideas:

1. Alternating palette every other pixel.
   The color index gets shifted back and forth every other pixel - the Apple IIgs provided such a mode. This allows to have more colors in high-res mode, although dithered.
   eg with 2 bitplanes
   palette 1: black, white, red, green
   palette 2: black, white, yellow, blue
   So text (b/w) would be crisp high-res - background, icons can be made of a combination of all colors via dithering.
   The IIgs does this as stripes, but I think a scheme where the alternation is switched every horizontal line in addition would be preferable providing a more checkerboard like appearance.

1. Mixing high-res and low-res in dual-playfield mode.
   eg. 1 bitplane in highres for text or gfx-outlines in foreground and a colorful 4 bitplane low-res background.
   (the pixel-clock would run in high-res mode, but simply repeat each value of the low-res playfield a second time, and thereby lowering the memory bandwidth)

your ideas?

[Hewitson]

I'd have liked a better range of resolutions available for 15khz displays. 640x480, 800x600, etc. AGA should have done this.

[Gorf]

Originally Posted by Hewitson

I'd have liked a resolution similar to VGA available for 15khz monitors/TV's.

This your Amiga can already do: Interlace gives you VGA-resolution at 15kHz.

[bloodline]

Really, once Commodore we’re using the faster dram chips for chipram (the ECS chipset), it became realistic to fetch a byte of data per pixel (hence 8bitplane mode of AGA), it was ridiculous that the chipset didn’t include a 256 colour chunky mode. This is the biggest omission from the chipset, and that should have been present around ‘88 or ‘89.

[Gorf]

Originally Posted by Hewitson

I'd have liked a better range of resolutions available for 15khz displays. 640x480, 800x600, etc. AGA should have done this.

since you edited your post, a second reply:

you could do that in theory - but 15kHz is the result of the horizontal refresh rate (50/60 Hz) and the number of lines.
So 600 lines would give you 15.000/600= 25 frames per second

(PAL is officially 625 lines with 576 of them visible)


(horizontal resolution is not affected - see super-high-res mode)

[Hewitson]

Originally Posted by Gorf

This your Amiga can already do: Interlace gives you VGA-resolution at 15kHz.

Interlace doesn't count. Also, it's only capable of 16 colours.

Originally Posted by Gorf

since you edited your post, a second reply:

I always edit my posts. You should probably wait 10 minutes before replying :P

Originally Posted by Gorf

you could do that in theory - but 15kHz is the result of the horizontal refresh rate (50/60 Hz) and the number of lines.
So 600 lines would give you 15.000/600= 25 frames per second

(PAL is officially 625 lines with 576 of them visible)

I realise this, but PC video cards manage to output to a TV at almost any resolution so it's possible.

[Gorf]

Originally Posted by bloodline

Really, once Commodore we’re using the faster dram chips for chipram (the ECS chipset), it became realistic to fetch a byte of data per pixel (hence 8bitplane mode of AGA), it was ridiculous that the chipset didn’t include a 256 colour chunky mode. This is the biggest omission from the chipset, and that should have been present around ‘88 or ‘89.

This would need a double-fetch per time-slot (which AGA provides).
But I agree - once fast page ram was available this should have been done - your timeframe is probably right.

[Gorf]

Originally Posted by Hewitson

Interlace doesn't count. Also, it's only capable of 16 colours.

no - it is capable of all bitplane modes including EHB and HAM

I realise this, but PC video cards manage to output to a TV at almost any resolution so it's possible.

no they are not - best thing they can do is scale it down to a lower resolution, what might give you the impression.

[meynaf]

What could have been implemented without much trouble i do not know, but what's badly missing is clearly chunky and true color modes.

[Gorf]

Originally Posted by meynaf

What could have been implemented without much trouble i do not know, but what's badly missing is clearly chunky and true color modes.

chunky with color LUT is quite a different approach to planar, so it is not so easy to implement with the original hardware design. you would probably end up with a second pixel-pipeline in Denise and almost double the transistor count.
So not so easy (or expensive) back than - but straight forward today (see Vampire)

true color: a enhanced HAM mode where you can change all values every pixel would have been easier - again at least with fast page ram ("double CAS")

[Gorf]

Here a example how the IIgs provides a colorful GUI with sharp text, using the alternating palette trick on two bitplanes:

edit:
pay attention to the Apple-logo in the menu on top of the screen and the same logo on the "system" folder.
The menu-bar is in a different screen mode (non-alternating), and instead changes the palette every line. The rest of the screen in using the alternating mode, so the logo further down is dithered.

Also take into consideration that these stripes are probably less visible on an analog CRT.

[Hewitson]

Originally Posted by Gorf

no - it is capable of all bitplane modes including EHB and HAM

Not in WB.

Originally Posted by Gorf

no they are not - best thing they can do is scale it down to a lower resolution, what might give you the impression.

Obviously they are being scaled down to PAL/NTSC resolution.

[Gorf]

Originally Posted by Hewitson

Not in WB.

you mean not in High-Res?
Interlace gives you always the same number of bitplanes as the non-interlace mode. On OCS/ECS in High-RES this is of course only 4 (=16 colors).

More would only have been possible with faster RAM, as mentioned above, and/or a wider data-bus (AGA).

This is not exactly the small improvements or missing modes I had in mind...
I was more looking for things that could have been done without changing too much. Some clever ideas the original design could have used.

Obviously they are being scaled down to PAL/NTSC resolution.

Resolutions the Amiga can do natively.
Not sure what you mean here...

[bloodline]

Originally Posted by Gorf

chunky with color LUT is quite a different approach to planar, so it is not so easy to implement with the original hardware design. you would probably end up with a second pixel-pipeline in Denise and almost double the transistor count.

I have thought about this, I don’t think a chunky mode would have required as much as you are suggesting... as long as the fetch can keep up with the DAC it would be rather simple (it would use the existing planar palette), chunky doesn’t need any shift registers, no fancy playfield modes, no sprites, just a DMA read/write to a single register.

I could be wrong and missed a whole stage of the process, but I’m sure chunky would be so much simple than planar.

[Gorf]

It depends on how Denise resolves the colors internally - so I am not sure If it could even use the same color-registers...
It seems to do the "look up" stepwise and not in one go - at least, that might explain, why super-high-res and productivity mode do have a limited palette of 64 colors in total: there is only time to get the fist two values of red, green and blue (6-bit) and not all four (12-bit).

So this is not only a question of getting the display data fast enough from RAM, but also resolving the according palette-values fast enough.

How does AGA do that? simply faster clock?

[Solo Kazuki]

Originally Posted by Gorf

no - it is capable of all bitplane modes including EHB and HAM

Originally Posted by Hewitson

Not in WB.

Really?

This is in WB...

[bloodline]

Originally Posted by Gorf

It depends on how Denise resolves the colors internally - so I am not sure If it could even use the same color-registers...
It seems to do the "look up" stepwise and not in one go - at least, that might explain, why super-high-res and productivity mode do have a limited palette of 64 colors in total: there is only time to get the fist two values of red, green and blue (6-bit) and not all four (12-bit).

So this is not only a question of getting the display data fast enough from RAM, but also resolving the according palette-values fast enough.

How does AGA do that? simply faster clock?

Ok, good point. I don’t know, I guess we will have to wait until Visual6502 have decapped and worked out their Denise before we will know

[Leffmann]

A 256-color chunky mode obviously. Even OCS has the bandwidth for it in lowres.

I understand that extending the palette register file and LUT mechanism from 32 to 256 entries would've increased the cost, but would it really have been prohibitive?

And I don't understand the arguments that the whole chip design would have had to be remade from scratch. If you have all the data required read into one or more holding registers, why is it difficult to select a different set of bits for the LUT? I'm sure these chip designs already solve more complex problems than that.

[Gorf]

:d

[bloodline]

Originally Posted by Gorf

:d

: the days you could build computers out of 74s!