Are the NEXT's graphics 'authentic?'

[The_Guy222]

I got to thinking about this topic after looking at some NEXT games, and viewing discussions on Youtube.

There were a couple of comments on videos to the effect that the NEXT does not generate graphical output in the way a historical ZX successor would have.

I understand it's FPGA, but does the simulated graphical hardware resemble what would have been built for a late 80s/early 90s system?


The reason I ask this is that it's become apparent that NEXT games can lose the 'identity' of the Spectrum by straying outside of classical limitations. Games like Vradark's Revenge.

On the other hand, Delta's Shadow port for the NEXT was developed deliberately not to stray too far from classic ZX Spectrum graphics.



Right now, developers have to subjectively choose to mimic ZX-style output. Would this be the case if a different method of graphics output was used?


I guess I'm asking, "What would a canonical ZX-successor graphics card look like?" The Sam Coupe was a canonical successor.



But people were known to say that the Coupe was not enough of a departure / advancement from the Spectrum. So how does the NEXT graphics processing differ from what a hypothetical post-Coupe ZX-successor might look like? Did the NEXT go in the right direction?

[Pegaz]

Please don't start with such heresy... it's late and I'm out of popcorn.
But, let's say briefly - Everything on Next is unique and its the only natural successor of ZX Spectrum.
Sam Coupe doesn't count, because it doesn't have the Rick Dickinson fancy keyboard and Sinclair logo.
You see how easy it is, when you think logically...

[Nitrowing]

iMHO, the Coupe was the successor to the Spectrum, under the hood, similar in so many ways.

The issue of "what came next after" can't really be quantified as the jump to 16bit was enormous. I can only look at the speed, music and colours of Sonic to think that Sinclair would have cheaped out and anything they released to take it on would have only appealed to over 30's

The Next can do so much it does feel like the things stifling development are the lack of big companies wanting to spend any time/money on a tiny project and the limited amount of time our dearest enthusiasts have to devote to it.

All I expected from the Next was games without colour clash and good sound effects on my giant screen. The development of some gorgeous looking ganes is great but not "Spectrum" in my eyes. If I want gorgeous screens, I listen to my RTX howl while running around Liberty City.

I would like to see an RPG looking similar to the Movie/Knight Lore isometric though

[HEXdidnt]

This has been nagging at me recently. I commented on the Aliens: Neoplasma thread that I felt it was odd that the graphics weren't significantly upgraded versus the Spectrum version, and one of the development team revealed that their research indicated that's what people wanted out of the Next.

I have to admit that, when looking at some of the games that take full advantage of the machine's graphical capabilities, I'm still not impressed, but in a different way. Sprites in some of the early games were excessively anti-aliased, as if they'd been designed at double the size (or more) and shrunk down with a bit of blur applied. Background layers are often so busy the sprites become less distinct. On balance, I'd certainly agree that Neoplasma looks better, but still think it could be taken a little further without straying too far from the Spectrum 'feel' they're aiming for. To me, Vradark's Revenge looks ideal - clear, colourful backgrounds and well-defined sprites, and the Spectrum 'feel' comes from the fairly limited number of unique tiles making up the backgrounds, and their very regular size.

However, I've been more of a SAM Coupé user than a Spectrum user since that machine first came out. I started out adding patches of colour to my monochrome Spectrum work, and didn't really develop beyond that until recently, when I discovered the concept of the "optimised 16-colour palette", which was a real eye-opener. I look back at my earliest work now, and think it's terrible.

Coincidentally, I recently posted a poll in the SAM Users Facebook group, asking how other SAM users would define "the SAM Coupé feel", and the vast majority responded that full colour MODE 4 - the top-end graphics mode - was preferred, but that MODE 2 (8x1 pixel attribute mode) was sorely underexplored. The shoot 'em up, Sphera, would be an awesome showcase for what MODE 2 can offer... if it weren't such a terrible example of a shoot 'em up.

Thing is, the quality of graphics on the SAM was generally pretty appalling. In most games, you either have MODE 4 graphics that are barely distinguishable from MODE 1 or 2 (eg. Manic Miner), or pillow-shading everywhere (eg. Boing!). There are great games saddled with lacklustre graphics (eg. The Witching Hour), and lacklustre games with impressive graphics (eg. Defenders of the Earth). It never really developed its own 'graphical identity' because everything was so different. The Spectrum had the advantage of being at the forefront of videogame development for most of a decade... the SAM wasn't as lucky.

Currently, it feels as though the Next hasn't found a consistent identity, and there's no real consensus on what 'authentic' means. There's the same sense of mediocre games with excellent graphics, and great games that look awful... though your mileage may vary. Going back to Neoplasma again, I may not fully appreciate its halfway house graphical style, but I can't fault the smoothness of its animation.

I doubt the Next will ever be as successful as the original Spectrum and - perhaps I speak heresy - I don't think many of it's games have even superseded what would be possible on the SAM yet, despite being vastly more powerful and better spec'd... I'd love to see SAM conversions of some Next games... but I suspect that's extremely unlikely.

Development seems to move along quicker and more consistently on the Next than on the SAM, if only because it's maker hasn't gone into receivership yet, so it's still being supported and is still available (kinda?) six years after the first development boards arrived. By that measure, it's certainly taking longer than the original Spectrum to find its feet... but the pool of developers is much smaller and, unlike the original Spectrum, there's precious little money to be made developing for the Next because, ultimately, it's a niche product.

Give it another couple of years (maybe another round of Kickstarter), and I think the games will start to look a bit more consistent, with a distinct Spectrum Next style.

[AndyC]

The SAM Coupé was a natural extension to the Speccy, in the same way the Amstrad Plus was an extension to the CPC. Both were honestly disappointing in many ways, because trying to extend a restricted architecture and keep compatibility were conflicting goals that were inevitably going to reach a poor compromise.

The Next is too advanced in many ways, it represents what you could now do to extend the architecture, but not really something that would have existed in the late 80s.

[AndyC]

Fwiw I'd have been much more impressed by a SAM version of Aliens Neoplasma than the Next version, even if they'd looked very similar.

[1024MAK]

The ZX Spectrum uses 1982 technology and ideas. It’s a development of the ZX81 really.
The next step up is the Timex Sinclair 2068.
And the QL.

That was it. End of line. No more Sinclair Research graphical improvements.

Yes, there was the Sam. But there was also the Enterprise 64. And then the Atari ST. And the Amiga.

If Sinclair had produced a new home computer (as opposed to extending the existing architecture) after 1984 but before 1989, I suspect the graphics would have been more similar to that of the Atari low and medium resolutions.

Ultimately, as graphics/video generation technology gets better, it loses it’s character. This applies universally. Not just to the Sinclair computers. It’s the drive for better resolution, more colours and shades, and thus, nearer to real life (or at least “4K” video).

Mark

[Nitrowing]

Ultimately, as graphics/video generation technology gets better, it loses it’s character. This applies universally. Not just to the Sinclair computers. It’s the drive for better resolution, more colours and shades, and thus, nearer to real life (or at least “4K” video).

Mark

Only need to look at PS5/Xbox/PC to see that the screen output is identical whereas the 8bit machines had completely different palettes. Most Unreal titles look like a take on Fortnite, which is fine for that particular game but really hinders other games trying to stand out.

The issue I see with the Next is that early game development was as much about pushing the computer's capabilities and that was being done by a few people because it only needed a few people. Zoysa produces amazing graphics within the tiny power available but how many people would be needed to push a Next to it's limits? What are it's limits? Could it run Fallout 1? How pretty could Doom look? Sonic? Skyrim?

[Pegaz]

The issue I see with the Next is that early game development was as much about pushing the computer's capabilities and that was being done by a few people because it only needed a few people. Zoysa produces amazing graphics within the tiny power available but how many people would be needed to push a Next to it's limits? What are it's limits? Could it run Fallout 1? How pretty could Doom look? Sonic? Skyrim?

The Xilinx Spartan-6 fpga can barely implement full C64 hardware, not to mention more demanding 16-bit or even stronger machines and their games.
Its the similar chip used by the ten times cheaper ZX-Uno, which is quite enough to support most 8-bit systems, with the possibility of turbo mode up to 28Mhz.
Ofc, I like to see the enhanced Spectrum graphics, if it keeps its original look&feel.
But honestly, I've seen more of that with the EmuZwin emulator and its recolored original games almost 20 years ago, than on the ZX Next.
This technique uses the original Spectrum games under the hood, over which goes an additional layer of multicolor enhancements.
As you can see, the result is great.

[TMD2003]

All I'm going to say on this is: if you want to see a Next game that has "an authentic Spectrum feel", then get stuck in and make it, or at least make a mock-up. Get CSpect, run it, learn how to define palettes and layers, and show us what you want from a Next screen. I'm still getting to grips with it even now - the order that colours and layers are defined is important and sometimes I still don't get the border colour I want, given that the BORDER command doesn't work for every layer, and it involves trial and error rather than knowing exactly how it's done. I'm getting there, though - so can anyone.

Graphics with a "true Spectrum feel" (Ben Shapiro would have a field day with that), likely as not, will have to stick to the 16 (well, 15) regular colours - with the standard ULAplus-style palette levels, the "regular" BRIGHT 0 colours will be slightly darker than on a regular Spectrum (#B6 rather than #C0 or there abouts, which Fuse uses as its standard level), so adjust the brightness and contrast on your TV if you're really bothered. You can do something like Melkior's Mansion that's stuck rigidly to the Spectrum colour palette, just without any colour clash, and you'll have what you want.

But, no "Spectrum Successor" would ever have stuck to that. It'd have had an extended palette like an Amstrad CPC, which could do orange and dark colours, though not a particularly convincing brown (what is it, a C64?) By 1985 we had the 16-bits, with as many colours as a Next can handle, which is why there are so many purists whining that Next games look like an ST or Amiga game. But then, the SAM Coupé could do that as well. Where the Next trumps all of these 1980s predecessors is that it can have more than 16 colours on screen at once on layer 2, which is what any Next programmer worthy of the name is going to be using (although real EXPERTS will dive into machine code and use layer 3).

Even in the CSSCGC I've put 24 colours on screen all at once:

...I don't think the screen would necessarily have lost anything in the down-conversion to a 16-colour screen, at least, nothing that anyone would notice. The mower body is three shades of green that aren't used anywhere else (it's actually "regular" green with the colours of the Google logo slightly blended in) - those could easily have been the same as one of the greens in the Fighting Machine's eyes. The darkest red in the BOOM could have been the brown of the shed, the yellow in the BOOM could have been the same as the Martian sky, and so on until "unnecessary" colours are eliminated. But I had 256 to choose from, so I ran with it.

At the other end of the scale, I've been experimenting with colour palettes that are designed to look like CGA and EGA, and which can be selected at will. That's not "Spectrum-like", it's more like an IBM PC of the mid-1980s with a 286 and no capacity to run Windows 3.1. As with the regular Spectrum colours, these are slightly off-centre, brighter than they should be this time (#6D and #B6 instead of #55 and #AA). Again, adjust the brightness and contrast if you're bothered.

I might even be a real heretic and define a C64 palette, but I wouldn't want Sir Clive to spin in his grave.

[Timmy]

I really have no problem with customers asking developers to make graphics they see fit (whether it should look spectrum, or next, or msx2, or pc, or c64) as long as they are rewarded very handsomely.

Drawing in any style is a lot of work, especially for most people.

For example, a few years ago there was this multicolour craze for Spectrum games. After a while of trying to draw sprites with 8x1 colour, I just realise that I couldn't do it. It also takes a lot of processing time that I'd rather spend on making better controls than nicer graphics.

Every style takes a lot more time, and I feel the Spectrum feel is also important that games play really well. You know, the "playability" thing.

Although I can understand that if you view the Next as a small improvement on the Spectrum, then you'd not want big steps in graphical improvements. But I would definitely not telling other people to not make different looking games.

I remember the curse of MSX2 (and up), where their video processor is so advanced, and everyone wanted the best graphics, that it ended up with very small amount of games released for them (compared to the Spectrum), because it's just a lot more work and time to make them.

Also, PLEASE don't make mockups of games you want to see. Because 1) You'll be disappointed if the final result doesn't look like what you want. And 2) Now you've set up really high expectations that developers have to do. And usually a really nice colourful screenshot looks fine, but usually it just meant there's less processor time and memory left for the rest of the game.

[The_Guy222]

iMHO, the Coupe was the successor to the Spectrum, under the hood, similar in so many ways.

The issue of "what came next after" can't really be quantified as the jump to 16bit was enormous. I can only look at the speed, music and colours of Sonic to think that Sinclair would have cheaped out and anything they released to take it on would have only appealed to over 30's

Wow! I am amazed at the level and depth of responses here! I didn't know there's be such an enthusiastic discussion.

It feels like Sonic the Hedgehog destroyed all the 8-bits, and it's amusing to see the Commodore and ZX Spectrum communities recreating it for their systems.

Sonic was on store display TVs for a while after it came out. I remember it being eye-catching and clearly the new standard.

I do agree with those who say that games like Vradark's Revenge, while colourful and pretty, may lack distinction between foreground sprites and the background. And yes, I get the sense that many NEXT games would be possible on the Coupé. But I can see why the NEXT project didn't replicate the Coupé. It isn't too fondly remembered, and maybe does not have enough graphical potential.

[It should have been seriously considered, no?]




I find this comment by AndyC to be interesting:

The SAM Coupé was a natural extension to the Speccy, in the same way the Amstrad Plus was an extension to the CPC. Both were honestly disappointing in many ways, because trying to extend a restricted architecture and keep compatibility were conflicting goals that were inevitably going to reach a poor compromise.

This reminds me of the modern neo-8-bit systems. I thought it might be more helpful to the Commodore and Spectrum communities to create some kind of extended original [as you sort of get with the NEXT]. But Commodore people went with the MEGA65, the X16, and the C256, which are (real or imagined) successor systems.

Like, imagine what a Commodore 64 NEXT would be like. 99% compatibility with C64, but options to have processor ramped up to 8 or 12 Mhz or something, 512 KB RAM or more, dual SID or better, a VERA-like chip. In other words, an agreed-upon spec for an extended C64. So people could port their games to that, as well.

I guess a true extended-C64 would have been too incompatible, or too limited if compatible. So we end up with the X16 and the like.



With the NEXT, it seems as if the graphical/sound hardware is very powerful, allowing it to blend in with modern arcade games. More would be expected of a developer who wanted to push the hardware. Whereas a more meagre GFX processor that eliminated the need for colour clash, could handle 1-3 extra layers, 24+ colours, maybe even basic hardware sprites like the Coupe, would be fine for many. Needing more optimization, but more of an outgrowth from ZX limits.

I've been looking at the ZX Evolution, which seems to be a kind of bootleg successor, and several games look like 'enhanced Spectrum.' Maybe it would have been okay to limit the NEXT to its level. But NEXT is possible, so why not? I guess it's up to the developer. Props to the EmuZwin enhanced games, it's more in-line with what I expected upon hearing about the NEXT. The power of the NEXT may inspire more people, though.

[Lethargeek]

The reason I ask this is that it's become apparent that NEXT games can lose the 'identity' of the Spectrum by straying outside of classical limitations.

removing old limitations is not a problem, adding the new ones is what making the new gfx non-authentic
in some respects the new gfx is actually even less capable than the old standard spectrum gfx
even in 2d sprite gfx (eg all Nemesis the Warlock sprites just won't fit into NEXT's hw sprite memory)

[SaNchez]

In ZX Spectrum Next 16kb of memory for sprites means 128 sprites with 16 colours per point (https://wiki.specnext.dev/Sprites). That is, you can cover 2/3 of the screen with sprites. In Nemesis the Warlock, the game area is 2/3 of the screen. Killed enemies can be moved to the background in layer2. So there is no problem.

[helpcomputer0]

Also, PLEASE don't make mockups of games you want to see. Because 1) You'll be disappointed if the final result doesn't look like what you want. And 2) Now you've set up really high expectations that developers have to do. And usually a really nice colourful screenshot looks fine, but usually it just meant there's less processor time and memory left for the rest of the game.

I get your points but surely what the Next needs most is more interest?

My starting point (and I'd bet for a lot of game artists) is a mockup, and if done properly you can keep within expections for the actual game. If you don't start with a mockup then you can end up with a mishmash of styles.

My own thoughts on the Next's style or "feel" is that, as others have said, it's built over time. Artists/devs copy each other in a particular scene and that develops.

I expect that despite many of the hardware restrictions being removed, the "feel" of the Next will still be carried by the artists and developers that are mostly, if not all, ZX Spectrum fans. I think that's more important than how many colours or sprites you can use.

[Lethargeek]

In ZX Spectrum Next 16kb of memory for sprites means 128 sprites with 16 colours per point (https://wiki.specnext.dev/Sprites). That is, you can cover 2/3 of the screen with sprites. In Nemesis the Warlock, the game area is 2/3 of the screen. Killed enemies can be moved to the background in layer2. So there is no problem.

Nope! The amount of dead bodies isn't a problem (these are just 2 small still sprites anyway). But the need to have too many frames of sprite animations ready at any time is THE problem. There are 22 Kbytes sprites in total, masked, one-sided. Let's deduct the masks, this gives us 11 Kbytes or 88 Kpixels - almost all of these need to be ready in adjacent game frames, and most of them might be even in one frame!

Even at 4bpp the NEXT will have only 32 Kpixels in sprite memory - that's OVER TWICE LESS than necessary!

...and it's just a humble 48K Spectrum game

[AndyC]

But you don't have to preload every single possible sprite frame in advance. I mean, sure that's the easy way but you can just load what is necessary on the next frame, right? Almost every 8-bit machine with sprites has limitations that require the programmer to juggle things around, it's not like machines that had sprites automatically became trivial to program for.

And, if worst came to worst, you could just draw them with software as the original does. So it's not really reducing functionality in any way.

[SaNchez]

Nope! The amount of dead bodies isn't a problem (these are just 2 small still sprites anyway). But the need to have too many frames of sprite animations ready at any time is THE problem. There are 22 Kbytes sprites in total, masked, one-sided. Let's deduct the masks, this gives us 11 Kbytes or 88 Kpixels - almost all of these need to be ready in adjacent game frames, and most of them might be even in one frame!

Even at 4bpp the NEXT will have only 32 Kpixels in sprite memory - that's OVER TWICE LESS than necessary!

...and it's just a humble 48K Spectrum game

You don't need to store all graphics in VRAM, nobody does that (maybe only in early NES games). VRAM is reloaded from RAM every frame via DMA, so 16kb is only images for 1 frame, the amount of graphics is limited only by the main RAM (768kb or 1792kb if expanded).

[Timmy]

I get your points but surely what the Next needs most is more interest?

I don't believe making "possible" screenshots is a good way to generate interest, because many Spectrum users got burned with those.

You could probably ask many people here how Outrun's promising screenshots turned out. Or Barbarian 2. Or Toki. Or some other games.

I do understand the importance of making these screenshots for imagination, or Demos (that would be very nice), and in 2024 these screens might even lead to PC games too. So it's not all bad. But I fear some of these images will backfire, and cause more problems than they are trying to solve.

But it's obviously only my opinion and I'm not saying that you shouldn't make screenshots. My main point is that (and sorry because my English isn't that well, and I think I made the wrong point earlier) is that you might have to be prepared that those screenshots might not look like the actual product, especially when the screenshots look too good. And you can still make screenshots if you want to.

Lower your expectations, and so on. If a game over delivers, that's always better.

PS. Have you ever considered putting some of your pictures in a Demo? Because if you ask me, you would be very famous in that scene.

[Lethargeek]

But you don't have to preload every single possible sprite frame in advance. I mean, sure that's the easy way but you can just load what is necessary on the next frame, right?

with hw sprites you HAVE to provide for the worst case of the game frame
in the worst case there might be ALL the enemy gfx (besides the player gfx) necessary
and it's well over 50 Kpixels just for this, still MUCH more than all of the NEXT's sprite memory
so... GAME OVER

Almost every 8-bit machine with sprites has limitations that require the programmer to juggle things around, it's not like machines that had sprites automatically became trivial to program for.

but the original Spectrum didn't have these limitations
hence, adding them couldn't be called an "authentic" "natural" extension
...or just an "extension" even

And, if worst came to worst, you could just draw them with software as the original does. So it's not really reducing functionality in any way.

you can't as there's not enough cpu speed (you need more than 8x as the memory isn't as easily adressable, but on the NEXT you have even less iirc)

You don't need to store all graphics in VRAM, nobody does that (maybe only in early NES games). VRAM is reloaded from RAM every frame via DMA, so 16kb is only images for 1 frame, the amount of graphics is limited only by the main RAM (768kb or 1792kb if expanded).

with hw sprites (and dynamically generated picture in general) you have to store everything for the worst case, see above

[AndyC]

I've coded for the CPC+, where there are just 16 sprites, each of which has it's own graphics definition and the only way to change it is to reload all 256 bytes. It's still perfectly possible to get 50fps gameplay.

I find it hard to believe the Next, with a very generous 16K of potentially shared sprite frames, isn't enough.

[Lethargeek]

I've coded for the CPC+, where there are just 16 sprites, each of which has it's own graphics definition and the only way to change it is to reload all 256 bytes. It's still perfectly possible to get 50fps gameplay.

i'm not that familiar with cpc+ but i suspect that actual pixel source size is not in the dedicated (limited) sprite memory

I find it hard to believe the Next, with a very generous 16K of potentially shared sprite frames, isn't enough.

well, because if you think pixels, not bytes, then 16K is not "very generous", it's actually "very much not much"

...even compared to the old original humble Spectrum

[ParadigmShifter]

These days you need 64MB to get a graphic (4096x4096x4 bytes) from an artist

They don't trust PNG cos the alpha channel is undefined in the standard. So 64MB a pop then...

[helpcomputer0]

I don't believe making "possible" screenshots is a good way to generate interest, because many Spectrum users got burned with those.

In my experience most people want to see good graphics/animation and interesting gameplay, and it doesn't matter whether it's a mockup, demo, or game in development. I'd love to see more of this on social media (and this forum), even if no games were made from these, it generates interest.

PS. Have you ever considered putting some of your pictures in a Demo? Because if you ask me, you would be very famous in that scene.

I've considered making all the graphics for a small game in the Next palette and releasing them for free, and then see if anyone makes the game.

I released a lot of partial graphics and mockups for ZX Spectrum but I don't know that anyone used any of those.

[ParadigmShifter]

In my experience most people want to see good graphics/animation and interesting gameplay, and it doesn't matter whether it's a mockup, demo, or game in development. I'd love to see more of this on social media (and this forum), even if no games were made from these, it generates interest.



I've considered making all the graphics for a small game in the Next palette and releasing them for free, and then see if anyone makes the game.

I released a lot of partial graphics and mockups for ZX Spectrum but I don't know that anyone used any of those.

Well I can't draw and have no imagination so mockups and game ideas are fine by me, although I do agree with Timmy that they should remain private or have "this is a mockup, does not represent actual game" until you've ran it past a coder to see if it is viable

I'm only interested in 48K though but your graphic mockups are good.