Why did you NOT upgrade to a CPC in the 80s?

[PeterJ]

Thanks @AndyC.

I did look at the Plus range last year, but although more colours are available, from what I understand the resolutions and number of colours per resolution (on screen at the same time) are the same? ALthough obviously, having sprites was an advantage.

https://acpc.me/#ACME/LITTERATURE_MANUELS/[ENG]ENGLISH

[1024MAK]

The CPC was released way after the Spectrum, and had multiple graphics modes. Imagine if one of those modes could replicate the graphics memory of a Spectrum. I.e. 256x192 (with the same weird layout of the Spectrum graphics memory) followed by an attribute area.

Altwasser received a patent for the ZX Spectrum video design. I presume this was transferred to Amstrad when they bought the Sinclair computers business. Until then, Amstrad would have been unable to sell a computer that used the same technique (well, they could sell computers but would open themselves up to legal action).

Mark

[Joefish]

I seem to remember the 464 and 6128 Plus range added the features of the GX4000 console (which was based on the CPC range), but officially they were only accessible to games run off the cartridge. But people quickly found software hacks to make them work.

[AndyC]

Thanks @AndyC.

I did look at the Plus range last year, but although more colours are available, from what I understand the resolutions and number of colours per resolution (on screen at the same time) are the same? ALthough obviously, having sprites was an advantage.

https://acpc.me/#ACME/LITTERATURE_MANUELS/[ENG]ENGLISH

Yes, but also sort of no. One of the added features is programmable line interrupts, so where the original CPC was limited to 6 fixed interrupts per frame the Plus models can have them on any lines it wants. That makes it a lot easier to do things like colour or Mode changes mid frame so putting a lot more colours on screen is a lot easier to achieve. There are also the sprites, which have their own set of 15 colours, which means having 31 colours on screen is easier.

It's may not seem like a massive improvement in the graphical capabilities, but there really is only so much that could be achieved when the hardware design is kind of fixed into reading one byte every four "square" pixels. I'm not sure there was a better approach that would've improved graphics in reality - I guess you could have had 256 colour super wide pixels, but I think that'd look uglier, or 4 mono pixels with the other four bits being used as a foreground palette entry?

I seem to remember the 464 and 6128 Plus range added the features of the GX4000 console (which was based on the CPC range), but officially they were only accessible to games run off the cartridge. But people quickly found software hacks to make them work.

Yeah, the Plus machines are effectively the same as the GX internally. There is a hardware unlock sequence that was provided to developers under NDA that prevented accessing the features normally. Once the GX flopped, those same Devs just went ahead and used it in disk games anyway (or at least a few of them). Not that it mattered as hackers had already disassembled the pack in game which was on the same cartridge as BASIC and started figuring out how to access the functionality anyway....

[catmeows]

Altwasser received a patent for the ZX Spectrum video design.
Mark

I don't see much difference between Zx Spectrum gfx mode and C64 bitmap hires mode.

[AndyC]

I don't see much difference between Zx Spectrum gfx mode and C64 bitmap hires mode.

There are lots of differences at a technical level, not least that the C64 has a fixed background colour, supports individual characters being interpreted in multicoloured mode and allows the bitmap to be relocated in RAM.

Patents are very specific, it's not just what you do but how you do it. As long as the method is different, achieving the same result doesn't matter.

[Lethargeek]

Well there are 27 colours rather than 16, so yes the brightness levels between some are further apart than they would be on the Speccy, but there are also more intermediate shades that can be used.

no, it isn't "there are also intermediates" but "even low/intermediate brightness difference is too much"

But I still think a lot of it comes down to emulation accuracy. The bright colours aren't quite that bright on a real monitor and bright and dark colours near each other bleed into each other in a way that doesn't happen in emulations. You have to see it on a real CTM to give an accurate judgement.

if "bright colours aren't quite that bright on a real monitor" then lowest brightness shades are almost invisible
or the other way around, there's less difference BUT black is not black but rather a light grey

[AndyC]

if "bright colours aren't quite that bright on a real monitor" then lowest brightness shades are almost invisible
or the other way around, there's less difference BUT black is not black but rather a light grey

It doesn't work like that, black is black, the bright colours aren't quite as bright as emulator screenshots and, in general, the emulator colours aren't really reproduced as well as they could be. It's a lot like the early days of spectrum emulators, where the authors just assumed non-bright colours were 50% of the bright ones.

[1024MAK]

So, this is rather going further off-topic, but the CPC computers output analogue video, and the Amstrad colour monitor is an analogue type with a CRT.

Just like the ZX Spectrum, there is no definitive correct level for any of the colours (other than black).

On a real CRT you can fiddle with the brightness and contrast controls (though on a monitor these may be internal, designed only to be adjusted by a technician).

Mark

[Matt_B]

I'd think that you could probably get closer to a definitive palette for the CPC, given that most people were using the same monitor, where it's far more TV-dependent with the Spectrum.

As ever though, it's the case that everything looked far less saturated, more blurry and full of chroma aberration back in the day, and all the more so if it was going through an RF modulator. Stick your emulator though the heaviest CRT filter it's got and remove any corrective eyewear before passing judgement, I suppose.

[Lethargeek]

It doesn't work like that, black is black, the bright colours aren't quite as bright as emulator screenshots and, in general, the emulator colours aren't really reproduced as well as they could be. It's a lot like the early days of spectrum emulators, where the authors just assumed non-bright colours were 50% of the bright ones.

first, "black" was even less black on the old CRTs than it is today even on the cheapest LCDs
second, if it's all about the emulator then why spectrum brightness levels look ok even in emulators

It's a lot like the early days of spectrum emulators, where the authors just assumed non-bright colours were 50% of the bright ones.

so what makes you think all amstrad emulator authors couldn't get things right even this late?
are they too hopelessly stupid compared to the spectrum people or what?

On a real CRT you can fiddle with the brightness and contrast controls

...aaand this is exactly that makes "black" looking even less black

[1024MAK]

first, "black" was even less black on the old CRTs than it is today even on the cheapest LCDs
second, if it's all about the emulator then why spectrum brightness levels look ok even in emulators

...aaand this is exactly that makes "black" looking even less black

Most LCD (including LCD with LED backlight) also have contrast and brightness controls. Hence the user can set up the picture how they want, within limits.

The black level is defined in video standards for the electrical signal, and this is one part of the signal where the vast majority of manufacturers of home computers did follow the standard. As otherwise they risked some colour TVs not working correctly with their computer.

How black the screen looks is highly dependent on the make/model and technology of the TV/monitor, as well as the setting of the contrast and brightness controls and the ambient light levels in the room.

I get that you don’t like the video output of the Amstrad CPC computers, you are entitled to your opinion. But let’s not get carried away with this, as both the ZX Spectrum and the Acorn BBC Micro can, when connected to some monitors (that have their controls set up for high contrast, high brightness pictures) also produce images that are unpleasant on the eye.

Do emulators get it correct? No. Because there is no correct output. As I have said, with analogue video, there is no magical 100% correct value for the contrast and brightness of the colours. Emulators should include user controls to adjust the picture to how they prefer.

As a practical example, do you remember as a child ever going to a shop that sold CRT TVs, did you notice that every single TV that was switched on, connected up and working produced a slightly different coloured image on screen? You also sometimes see this with LCD TVs as well.

I will end it there, as this is off-topic. I’m sure there is another topic on here where there has been a discussion about the colours that the ZX Spectrum produces. That, or better yet, a new topic would be a better place to discuss this further.

Mark

[AndyC]

first, "black" was even less black on the old CRTs than it is today even on the cheapest LCDs
second, if it's all about the emulator then why spectrum brightness levels look ok even in emulators

Well yes, but you know what I meant. As black as black ever got on a CRT, just like on the Spectrum.

so what makes you think all amstrad emulator authors couldn't get things right even this late?
are they too hopelessly stupid compared to the spectrum people or what?

The Amstrad is many orders of magnitude more complicated to emulate than the Spectrum. When you're trying to get everything running and working as it is supposed to, often perfect colour reproduction isn't high on the list and just algorithmically generating the RGB values is just easier.

There is some investigation on the Grimware site into what the "correct" colour values should be, calculated by measuring the voltages on the RGB out (and including the nuance that the Plus produces slightly different values). I don't think many emulator authors have necessarily picked those values up though and there is still the issue that they're based on ideal values rather than what the monitor looked like.

Other things that almost every emulator gets wrong: the aspect ratio of pixels - in an emulator Mode 1 pixels are invariably square, which isn't accurate. And since Mode 0 pixels are double the width of Mode 1, emulators tend to have a 2:1 aspect ratio, again wrong. Likewise, if you look at the high res Mode 2 an emulator has discrete rectangular pixels, the real monitor doesn't actually have quite enough resolution to pull that off and so what you actually see is very dependent on neighbouring pixels - a black and white checkboard pattern will give a kind of fuzzy grey colour (this is also why individual pixels in other modes kind of blend a bit). And in Mode 2 the hardware is ever so subtly bugged such that pixels start a fraction of a second earlier than they should, which is exploited by some techniques for faking a higher colour depth.... etc etc

[Lethargeek]

I get that you don’t like the video output of the Amstrad CPC computers, you are entitled to your opinion. But let’s not get carried away with this, as both the ZX Spectrum and the Acorn BBC Micro can, when connected to some monitors (that have their controls set up for high contrast, high brightness pictures) also produce images that are unpleasant on the eye.

that's the problem, as "able to produce an UNpleasant image" is quite different to "UNable to produce a pleasant image"

The Amstrad is many orders of magnitude more complicated to emulate than the Spectrum. When you're trying to get everything running and working as it is supposed to, often perfect colour reproduction isn't high on the list and just algorithmically generating the RGB values is just easier.

i think you're misusing "orders of magnitude" but whatever

There is some investigation on the Grimware site into what the "correct" colour values should be, calculated by measuring the voltages on the RGB out (and including the nuance that the Plus produces slightly different values). I don't think many emulator authors have necessarily picked those values up though and there is still the issue that they're based on ideal values rather than what the monitor looked like.

well, after all these years spectrum, atari and c64 emulators got it acceptable, but not the cpc ones somehow
a strong hint that the root of the problem lies elsewhere

[AndyC]

i think you're misusing "orders of magnitude" but whatever

I think you're vastly underestimating how simple a Speccy is in comparison.

well, after all these years spectrum, atari and c64 emulators got it acceptable, but not the cpc ones somehow
a strong hint that the root of the problem lies elsewhere

Because they are all much, much larger scenes? Because most people think the current approach is good enough? Because you actually need at least a 4K screen to get even close to approximating the actual monitor display?

[TheMartian]

Well yes, but you know what I meant. As black as black ever got on a CRT, just like on the Spectrum.

Usually there are two levels of "black", one of which is the "black" colour, and the other happens when the picture beam is disabled, as in vertical retrace. You can get stripes of disabled picture beam in the middle of the screen because the sweep is controlled by the CRTC, and this is programmable.

The Amstrad is many orders of magnitude more complicated to emulate than the Spectrum. When you're trying to get everything running and working as it is supposed to, often perfect colour reproduction isn't high on the list and just algorithmically generating the RGB values is just easier.

Not really. If something, there is a lot less documentation on the CPC than on the Speccy. Where the ZX Spectrum has the ULA book, the CPC only has schematics from a 40010 Gate Array decapped, and, aside the Amstrad CPC CRTC Compendium by Longshot/Logon System, all the info on the different CRTC models is pretty much conjecture (as in "guessing from how the hardware behaves").

There is some investigation on the Grimware site into what the "correct" colour values should be, calculated by measuring the voltages on the RGB out (and including the nuance that the Plus produces slightly different values). I don't think many emulator authors have necessarily picked those values up though and there is still the issue that they're based on ideal values rather than what the monitor looked like.

I'm using these, I believe, in SpecIde.

Other things that almost every emulator gets wrong: the aspect ratio of pixels - in an emulator Mode 1 pixels are invariably square, which isn't accurate. And since Mode 0 pixels are double the width of Mode 1, emulators tend to have a 2:1 aspect ratio, again wrong. Likewise, if you look at the high res Mode 2 an emulator has discrete rectangular pixels, the real monitor doesn't actually have quite enough resolution to pull that off and so what you actually see is very dependent on neighbouring pixels - a black and white checkboard pattern will give a kind of fuzzy grey colour (this is also why individual pixels in other modes kind of blend a bit). And in Mode 2 the hardware is ever so subtly bugged such that pixels start a fraction of a second earlier than they should, which is exploited by some techniques for faking a higher colour depth.... etc etc

I'm trying to be extra careful with this, so I'm adjusting the pixel aspect ratio according to the standard PAL clock (@15,625MHz) vs the emulated machine pixel clock.
In the ZX Spectrum 48K, the pixel clock is 14.000MHz, so pixels are something like 1.11/1.00.
In the 128K models, pixel clock is 14.1876MHz. That gives 1.10/1.00 (slightly less wide)
In the CPCs, pixel clock is 16MHz, so pixels are more vertical. I'm using 0.9765/1.00

As I say, the CPC emulation:
- Is not so difficult "per se"
- We just lack some accurate information (especially on the CPC+ series)
- Heck, we're trying...



Edit: Pictures or it didn't happen.


Edit 2: Amstrad CPC CRTC Compendium by Longshot/LS, not Overflow/LS. Overflow is the author of Logon's Run demo.

[equinox]

The Amstrad CPC computers were available in [...country list]
Lists are not intended to be exhaustive.

And here is some brilliance generated by Meta's (now axed) Galactica science AI attempt:
[https://pbs.twimg.com/media/FhvY6g8WYAI ... ame=medium]

[1024MAK]

Mark