Enterprise 64/128 TOSEC

[TFM]

Yes right, and use moshrooms, shells, and pretty much anything than grass ,)

[MacDeath]

I up this topic.

Could anyone of you tell me where can I find proper renditions of the Enterprise colour palette and list ?
With names and so on ?

The "colour bias" is actually hard to understand if you don't have a proper list of the colours numbered and so on...

otherwise I guess the palette is a simple "8-8-4 values" RGB, so R=3bit+G=3bits+B=2bits...
As can be found with the MSX2...





So how can I be sure how to "port" graphics from CPC to Enterprise if i can't have a proper colour table for this system nor more specific informations on this system ?
My aim is to add this to the CPCwiki's enterprise page, but also start a page on "how to emulate CPC palette on other system and other graphic port concepts" (not sure about the name at the moment).


Post edition : ok after a long and desperate search I managed to find this page...
Enterprise Technical Documentation: Nick Chip Programmers Guide


Still not exactly what I am looking for, will search more and read more then.

[arnoldemu]

In the nick documentation, the definition of the colours is defined.
you can set 8 colours, these are actually composed of r,g,b values.
colour bias I believe is also r,g,b, but it's lowest 3 bits are an index from 0 to 8.

it is not possible to convert a cpc picture perfectly, some sacrifices must be made.

[arnoldemu]

In the nick documentation, the definition of the colours is defined.
you can set 8 colours, these are actually composed of r,g,b values.
colour bias I believe is also r,g,b, but it's lowest 3 bits are an index from 0 to 8.

it is not possible to convert a cpc picture perfectly, some sacrifices must be made.

"Of the eight bits, there are three each for red and green and two bits for blue, thus allowing eight levels of red and green and four levels of blue. The layout of bits is:

      
b7   b6   b5   b4   b3   b2   b1   b0 
+----+----+----+----+----+----+----+----+ 
| g0 | r0 | b0 | g1 | r1 | b1 | g2 | r2 | 
+----+----+----+----+----+----+----+----+

total RED   = ( r2 * 4 + r1 * 2 + r0 ) / 7
 total GREEN = ( g2 * 4 + g1 * 2 + g0 ) / 7 
total BLUE  = (          b1 * 2 + b0 ) / 3"

for each "line" in the line parameter table you can define 8 colours. 1 byte for each colour, each byte is a r,g,b.

Then in 16 colour mode, you use must use this register:

FIXBIAS
b0..b4 Colour bias to be used for top 5 bits of palette colours 8-15.
the low 3 bits come from pen index - 8.

so, what this is saying is that the red, green, blue define the total palette.
in each mode, the number of pens available, how the data is read and then shown is different. some are like speccy, some are like cpc.

in 16 colour pixel mode (just like cpc's mode 0), you can freely choose 8 colours, the others are defined with fixbias. So here, converting a cpc screen can not be done perfect.. careful choice of fixbias is needed to get close.

in 4 colour pixel mode (just like cpc's mode 1), you can define the colours almost perfect.

in 2 colour pixel mode (just like cpc's mode 2 I believe), you can define the colours almost perfect too.

perhaps I just confused things... ?

[MacDeath]

The bias concept if what I can't find proper explanation...
It is said somewhare that this value set a derogatory list of 8 colours from the 256 palette...

3 bits of fixbias = 32 values.
32x8 (inks) = 256 colours.

The problem I wish I could find a picture with the whole 256 palette with each colours numeroted/named, and the bias lists shown too.

This could be helpfull so Amstrad graphic artist could manage their ink set in order to also be "enterprise friendly" (or MSX2 friendly too)

Nice point with the enterprise is the "raster list".
It reminds me a bit the way the Apple//gs could get 16x16 palettes and assign a different palette (inside the 16x16 = 256 /4096) for each scanline, the way character attributes work.
those are somewhat "scanline colour attributes".

Even the PLUS despite sweet raster interrupts, may not have it that easy.

According to the documentation, the Fixbias can also profit from "sprites" from the extension port...
What is it ?
Is this for video incrustation or a way to get some peripheral to add some "maskedHardsprites" support ?


I like the way it also say that the number of colours must be set to 2 when using attributed mode, or else "imprevisible result"...
What would actually happen then ?

Can't we have some 2bpp attributes ?

The management for the video modes can seem a bit peculiar...
I mean you can set the number of colours or have another set of video modes.

I guess higher bpp means resolution limitations de facto, and to mix them is more a way to put less colours for similar video modes than to have 256 colours in high resolution (like 320x256...) ?
(unless you manage to mod the computer to overclock Nick perhaps... could be nice too)

So yeah.

There I made some stuff in order to compare CPC, EGA and the 8bit 3-3-2 RGB...
The 3-3-2bit RGB can actually be easily comparaired to a real 4x4x4 Cube like this, exept it is a "8x8x4 cube".

A funny aspect with the 8bit 3-3-2 RGB is that it actually lacks seriously in the "true grey" domain...

But searching well you can find somewhat 2 slightly bluish (cold) grey comparable to those for the EGA...
Also I could find 2 "gold greys".
But As the CPC also has a serious lack in Grey, it is not that important IMO.

To me, for an 8 bit computer, a palette such as the 256 colour palette is somewhat overkill...
Many colours quite look exactly the same (bright cyans or greens...), but of course it also allows a large selection of unusual colours too (the greyish ones per exemple or the crossbreed ones).
As the CPC palette is also quite limited, I don't think this is that hard to emulate properly the CPC, of course a few colours wouldn't be exactly the same, but this is the case on real hardware too...
Diferent CPCs on different monitors would get slight differences anyway.

To be fair, unless you want a picture in 16 shades of grey or specificities like this, the simple EGA palette is quite enough for most "game like" graphics in 16 colours.
Sadly the EGA itself rarely managed to have access to its real palette, being too often actually limited to the CGA/spectrum like palette.

But the 640x200x16 mode from the Amstrad PC1512 (or EGA too) could counter this due to the dithering capabilities of the "640" horizontal resolution.

[arnoldemu]

The bias concept if what I can't find proper explanation...

It is said somewhare that this value set a derogatory list of 8 colours from the 256 palette...

3 bits of fixbias = 32 values.
32x8 (inks) = 256 colours.

The problem I wish I could find a picture with the whole 256 palette with each colours numeroted/named, and the bias lists shown too.

I don't think there is one. I agree it's explanation is not clear and I agree a palette to indicate how it works would be very useful.

This could be helpfull so Amstrad graphic artist could manage their ink set in order to also be "enterprise friendly" (or MSX2 friendly too)

yes I agree.

Nice point with the enterprise is the "raster list".

yes

According to the documentation, the Fixbias can also profit from "sprites" from the extension port...
What is it ?
Is this for video incrustation or a way to get some peripheral to add some "maskedHardsprites" support ?

it is for a peripheral, i don't think it was used. I don't know if the image is an overlay or a replacement. it may be genlock/"incrustation".

I like the way it also say that the number of colours must be set to 2 when using attributed mode, or else "imprevisible result"...
What would actually happen then ?

Can't we have some 2bpp attributes ?

there are some "invalid" combinations. I don't know what the result is.

I guess higher bpp means resolution limitations de facto, and to mix them is more a way to put less colours for similar video modes than to have 256 colours in high resolution (like 320x256...) ?
(unless you manage to mod the computer to overclock Nick perhaps... could be nice too)

I think it's all about memory bandwidth.
a higher resolution, but less colours, equals the same number of bytes (or less) to read each cycle.
to get higher resolution and lots of colours does mean to read more data.
To do this the chip needs to run faster, and it needs to work better with the way the z80 reads the data.
They both share the same ram access.

For c64, cpu, then video, then cpu, then video, this is how the memory access is done for display.
On cpc, it makes the z80 wait a little.
With dedicated vram (on msx1) it is possible to do this and z80 remain same speed, but then you must have a special way to access the dedicated vram.

msx1 has a big problem here, later msx have blitter like extensions to make this easier.

The 3-3-2bit RGB can actually be easily comparaired to a real 4x4x4 Cube like this, exept it is a "8x8x4 cube".

A funny aspect with the 8bit 3-3-2 RGB is that it actually lacks seriously in the "true grey" domain...

interesting.

But As the CPC also has a serious lack in Grey, it is not that important IMO.

I often miss another shade of grey, I wish the cpc had it.

To me, for an 8 bit computer, a palette such as the 256 colour palette is somewhat overkill...
Many colours quite look exactly the same (bright cyans or greens...), but of course it also allows a large selection of unusual colours too (the greyish ones per exemple or the crossbreed ones).
As the CPC palette is also quite limited, I don't think this is that hard to emulate properly the CPC, of course a few colours wouldn't be exactly the same, but this is the case on real hardware too...
Diferent CPCs on different monitors would get slight differences anyway.

yes all true

To be fair, unless you want a picture in 16 shades of grey or specificities like this, the simple EGA palette is quite enough for most "game like" graphics in 16 colours.
Sadly the EGA itself rarely managed to have access to its real palette, being too often actually limited to the CGA/spectrum like palette.

ega is an evolution of cga, so it inherited some problems from it's ancestor?

But the 640x200x16 mode from the Amstrad PC1512 (or EGA too) could counter this due to the dithering capabilities of the "640" horizontal resolution.

yes dither and colour combinations often make a lot more.

splattr game on spectrum makes new "colours" with clever colour choices and a uniform dither.

[MacDeath]

ega is an evolution of cga, so it inherited some problems from it's ancestor?

may be out of topic but yes, the EGA was plagued with the legacy from CGA.

Only in 640x350 mode you could freely choose from the true EGA palette.
in vertical 200pixels modes you were limited to the "full CGA" palette only (unless some tweakings perhaps), but could get them all in 320x200, 160x200 or even 640x200...
There were also special modes like 320x200x4 (like CGA) where you could choose freely the 4 inks from the 16 colour palette, while CGA had to choose from fixed palette with only the black (first ink) to be set into another "custom" colour... which is quite retarded.

Concerning purely graphic modes, the CPC is actually more powerfull than a CGA...
Only CGA advantage is that it have attribute text mode in 16 colours... and it also has 2 greys...

IBM who designed those Video cards specification was known as "fucking mainframe IBM" and had the subtility of a mamooth at the time.

Games with EGA Graphics
but this site seem dead...


Back to topic.


I can do the palette picture for the Enterprise provided I can undertand how those colours are numeroted/numbered.
It would also be nice if I could understand the bias "clusters" a bit more...

Another funny stuff is that you can also simply "name" the teletext basic colours = pure blue, pure magenta and so on...

let's see...
3-3-2 bits = 8-8-4 values.

Blue can have only 0%-33%-66%-100% values I guess.
Concerning Red and green the value goes from 0 to 7... 0 being 0% and 7 being 100% I guess.

So each value add +1/7 = +14,285...%

0= 0%
1=14,285...%
2=28,571...%
3=42,857...%
4=57,142...%
5=71,428...%
6=85,714...%
7=100%

It's like mixing Atari ST palette and EGA palette...

As a result you can't be equal to the blue values, so actually no "equilibrated greys" possible.

But still you can have a few "almost greys" and a lot of unsaturated hues.
To emulate C64 graphics is indeed even more problematic on Enterprise than on CPC due to having no real greys while C64 has 3 true greys ... and 11 "almost greys" colours...  

Here is the sort of picture I would like to do... but for the Enterprise.
[attachimg=2]


I managed to find this, it seems it is a table to convert amstrad CPC colours into Enterprise (and so on) colours...
Sadly it miss a lot like the bias constraint so i want to do a better one too.
[attachimg=1]
I haven't figured out what GEC0 means.

Could someone tell me also what walue should I use as R-G-B on my modern PC to almost faithfully emulate or generate the Enterprise colours ?

I fail hard at hexadecimal...