SPECTRA Advanced Graphics

[cheveron]

Ohm, I see emotions!!!

No, I'm just fed up of people repeating stuff they've heard about ULAplus without bothering to check the facts.

SPECTRA is a SCART interface.

That' like saying, Multiface one is a Composite Video or Kempston Joystick Interface.

Then why does Paul Farrow himself describe it as a SCART interface? The aim of SPECTRA is to give a decent video signal on a 48K machine without the need to perform an internal modification. It does this very well.

Its extended colour scheme is worse than ULAplus in every single respect.

So ULA+ has 4x1 Attributes mode too?

No, because that mode takes too much RAM and too much CPU to make much use of. Static pictures look better in HAM8x1 mode.

In fact ULA+ is only good for recolouring existing games, like the Spectrum emulation on SAM in Mode 1.

That's the sort of nonsense that I'm tired of hearing. People haven't even begun to scrape the surface of what's possible with ULAplus. For instance 24 color backdrops with no CPU overhead, palette cycling, and I'm sure there are others. Also ULAplus is fully supported in SE Basic.

ULAplus is available now in clones and is emulated by almost everything. The only thing remaining to sort out on the plug-in board is the audio levels for LOADing and SAVEing. Everything else works.

So when it will be available for 48K Issue 2, Issue 3 and Toastrack? I was waiting years for it and nothing happened.

It will be available when it's available. How long were we waiting for a fully featured OS for divIDE? It's only now that esxDOS is really coming into its own. The development board made by Miguel works in any 48K machine from the issue 1 on because it is a direct replacement for the original ULA. Only minor changes to the core will be required for the timing on the toastrack and the grey +2. Miguel is even talking about doing an overrride for the ASIC in the +3 and +2A models at some point. But right now he's doing his PhD. He's put all of the source for the project online so anyone who knows a thing or two about FPGAs could build one now. In a way I wish he'd put it out now with the broken tape loading, because frankly we're all using divIDE and esxDOS now, right? But he wants to make sure even if all you have is a 16K issue 1 that it works completely. And remember you don't just get the palette modes with Miguel's board. You get all the Timex video modes too. I expect the plug-in will be released not long after he gets his doctorate.

SPECTRA will only ever work on the 48K machine and is emulated only by ZXSP.

"Subject to change". I know that 128K has no Y on the bus, and no Composite video, but why not take the signals from RGB Socket? SPECTRA for 128 is not impossible.

The hardware design of the SPECTRA is incompatible with the 128s and Paul himself has said he has no intention of making a 128 version.

And about the emulators: if software for Spectra will start to appear, it will be supported by Emulators.

It doesn't work like that. Emulator authors support what they want to support. The Spectrum SE is real hardware (admittedly only a single prototype but still real) and only two emulators supported that. ULAplus was emulated before the FPGA hardware was finished because it was trivial to do (except in FUSE which is why FUSE doesn't support it except as a patch), and authors could see the benefits of it. ULAplus has a fair amount of original software now, but I doubt Jon Needle will ever add support for it to Spectaculator even when Miguel's board is generally available. If there are two SPECTRA titles, most emulator authors simply won't bother with it. The reason ULAplus was so successful is because it can be used to recolor existing software. That means there are immediately thousands of titles available for it. Recoloring existing software wasn't the point of ULAplus, but it helped to get it emulated. SPECTRA is too niche. It will only be added by people who are interested in the technical challenge of implementing its screen modes. To date that's kio. I wouldn't hold my breath waiting for a Windows based emulator to support it.

Why recolour existing games if coding new games can be such a fun.

Because coding new games is niche, and playing existing games with more colors has mass appeal. Most users are gamers, not coders. If you want a device to be adopted you have to appeal to the biggest market.

And why to hack Palette if SPECTRA does not require any hacking?

Did you mean ULAplus? Most Spectrum software doesn't make extensive use of BRIGHT, so setting a palette prior to loading a game usually only gets you 32 colors on screen. If you edit the attributes in the game to use the FLASH bit then you can get the full 64 colors on screen. No coding required, just need to locate the attribute data. SPECTRA really doesn't support existing software without a rewrite and as I said, you really need to support existing software if you want mass appeal.

ULAplus has a carefully chosen 256 colour palette that works well for converting 24-bit images. In 8x1 attribute mode using the HAM8x1 software you can display a static 256 colour screen with 32 colours per row and 8x1 attributes.

Can this 8x1 HAM be done fullscreen on 48K or is it Timex-Only?

Yes it works on the 48K machine. Miguel has implemented the full ULAplus specification which includes the Timex video modes.

Carefully choosen sounds like standard 3-3-2 Encoding as in MSX2 Mode 8.

It's not just 3-3-2 encoding (because the eye is less sensitive to differences in blue), it's also the GRB arrangement which means that if you turn down the color then the palette becomes 256 levels of gray from 0 to 255.

SPECTRA is limited to a palette of 64 colours, many of which are barely distinguishable.

barely distinguishable? And ULA+ 256 Colours (let me guess: 3-3-2 RGB Bits?) are easier distinguishable?

Yes. Just compare a picture of both palettes side by side. The 3-3-2 encoding and larger palette makes a big difference. Miguel had to turn the color and brightness up on the television while making his SPECTRA video for YouTube so you could actually see the difference.

Its hi-res attribute modes are incompatible with existing Timex software.

Assuming many people have Timex 2068 software. How many Timex Software tilles are using Hires colour mode, apart from some demos?

More than there are SPECTRA titles. It's all about giving people the most reasons to want one.

Its maximum colour resolution still gives worse results than HAM8x1 and uses considerably more RAM.

And how many games use HAM8x1 mode? Oh wait... Not a single one... How many are in development? Is there still CPU time left if using HAM8x1 mode? Yes, Spectra has less colours, but the features can be really used in games and not only in slideshows.

As the name implies, HAM8x1 is for static images. There are already plenty of slideshows in circulation. A lot more could be done with ULAplus in games, particularly with palette cycling. So far only John Elliot's patch for Chaos has made use of that. The problem is that even after you deliver new hardware, it's insanely difficult to get anyone to write anything to support it. There just aren't many people coding for the Speccy who are interested in new hardware. And there are probably more people coding for the speccy than any other contemporary system.

ULAplus supports the Timex modes. Eventually the plug-in should bring these features to the 128K machines, including the ability to run SE Basic 4.01 with its 80-column hi-res screen mode.

And the release date is which year planed?.

Talk to Miguel. When I designed the specification for ULAplus with Chris Smith, the launch of a project by Chris was imminent. It got delayed. If I'd known it would take this long for the hardware to appear I probably would have kept ULAplus to myself for longer. But FPGA clones with ULAplus are becomming the norm. It's even been adopted in the Russian clone scene. I think it's inevitable that it will eventually arrive on real hardware.

Personally I think SPECTRA is a dead end and I wouldn't waste time developing for it.

I respect your opinion, but I don't share it. You don't need to "waste" your time developing for SPECTRA. I don't waste my time coding Slideshows for ULA+. Simple!

I certainly wouldn't want to deter anyone from coding for SPECTRA. If you enjoy it, then that's reason enough to do it. But there are some cool games that could be done with ULAplus, ZX BASIC and fourspriter. If I thought anyone was interested I'd write a couple of libraries for ZX BASIC. However, I'm not sure anyone is interested. You'll note that I'm not bothering with ULAplus for ZXodus. I simply don't think enough people care about this stuff to make it worthwhile and I had my fun designing it.

The ZXBC has already librarys to support ULA+, so I wonder why this hostility against SPECTRA. I myself think, developing for ULA+ is a waste of time, unless I want to make a slideshow.

I'm not hostile to SPECTRA. I am annoyed about the misinformation that goes around about ULAplus, much of which you've repeated in your post. It has plenty of gaming applications. It's just that no-one has used them yet, despite it being available for years. Getting people to write the software is the hard bit.

Oh, and sorry that I was too harsch to ULA+, but I waited for it for years as hardware implementation, and it still not available except in Emulators and clones. And it has no real use for gamemaking, only for slideshows.

I am sorry that people have been waiting for it for so long, but that's not something I had any control over. I'm almost tempted to put together a Defender like game with a 24 color backdrop and palette cycling effects to show what you can really do with ULAplus, it's just that I'm more interested on working on ZXodus at the moment.

Here's a screenshot of a very simple hack I did to do something a bit better than a simple recolor with ULAplus: