new amiga 500 doom like !

[KK/Altair]

one question : if i'm not wrong i saw some kind of scanlines vertically on the gfx, is it done on purpose or is it a technical thing ??

If you mean the vertical pattern visible on the textures - it's color dithering. Internally pixels are 2x2 (160x100 viewport resolution), but I can color both vertical halves separately and use that for color dithering. The checker pattern would be much cooler, but the scanlines are copper-doubled, so it just has to be this way. I've already tried making proper checkerboard with blitter (much nicer, but adds blitter pass = slower) and jittering using hardware scrolling (blurs features across pixels) and the current dither method feels like a best tradeoff. I could also use full colors only (like in Wildcat wolf), but this would severely limit what I could do with textures, lighting and enemies.

And if you mean the full-height vertical stripes flashing from time to time in random places - it's a known engine bug I didn't bother to hunt down so far.

2nd question : do you think that your dev tool could make a game like : monster farm jump ? if you could sell your dev tool, i think i would try to do this :

Unfortunately, this would require completely different kind of engine. And considering the limited features there (mostly 2D textured plane with fixed camera), it could be quite doable if approached this way.

3rd question : many talented coders tried the impossible to make what you have done, why do you think they failed, what they've got wrong ?

I have completely no idea.

But keep in mind that I was experimenting with this engine from half a year - not counting previous experience with Wildcat wolf, or all the 3D engine prototypes I've done in my life. And here, I have rewritten code and changed assumptions regularly during this time, so it wasn't a matter of getting things right or wrong, but more like a constant improvement of details, which eventually added up. You simply didn't see how many things I got wrong and had to eventually reconsider.

I've watched footage for your presentation and I'm wondering if art specifically made and tailored for your engine would be possible to bring up the fidelity even further (like f.e. mask some of the scanline doubling or work alongside the engine's quirks and limitations).

Choosing proper colors, acceptable dithering combinations and using it properly in sprites and textures would definitely make worlds of difference here. I simply grabbed FreeDoom sprites & textures and tried to come up with an acceptable palette and dithering method. Just acceptable - not even perfect, so more time spent here would definitely help.

Also do you think that having original art for the game would contribute to your motivation?

It's hard to tell right now. The acceptable level of motivation is actually there. It's just the vision of amount of required hard work that's putting me off.
So, just to give a few examples of what's ahead:

1. Decouple map, textures and sprites from the executable and load them dynamically. Right now the game is just a big EXE with everything compiled into it. Much easier to code, but limits everything to just one map.

2. Complete sound system from scratch - right now it plays one MOD continuously, but the true dream would require some kind of custom music player (and probably a tracker :P) with dynamic allocation of channels, that would also be shared with SFX. SFX would really benefit from having accss to 2 channels or more (e.g. player shotgun + enemy dying sounds), but 2-track music is out of question. Ability to share instruments between tracks on same disk would also be awesome space saver. (yes, as I've said before - perfectionism)

3. Minor details required in each game - splash screens, menus, etc.

4. And mind you, the engine is not even half done with the fixed camera height and flat floors. That part requires serious rewrites, too (but I'm positive that it can be done).

And so on...

I'd love to see some kind of network connection where I could deathmatch against other guys with Amiga500s. Imagine the possibilities here..

This would be a bit tricky because the engine doesn't work in fixed timesteps (like Doom did). So, every connected Amiga would display and process a completely different set of frames. There should be ways of dealing with that to make the experience acceptable (or even fun), but it's better to focus on single-player first and make it as good as it can be on pure A500.

Also, I think it would be nice if you could make a youtube video playing your game, so more people can see this...

That's already on my list. Just cleaning my workplace is kind of obstacle in this... (yeah, I'm a messy person and having more stuff around than available drawers doesn't really help).

Get the basic game engine complete, a decent number of maps and enemies, play-test and put it out there for a decent price, then as i said to him if sales are good this will inspire to add more features for a possible sequel, the last thing we want from this is to end with another Doom clone tech demo that never went anywhere!

I'm realistic with my estimations, and even that would be years of work to pull by oneself. So I'm more inclined to make this a community project with a group of contributors and/or supporters to keep going and produce results in an acceptable time. And having done soooo many Doom maps myself, modding is probably the way everybody can get most fun from this project in foreseeable future.

1) Did I get it correctly from the livestream that the whole map is basically divided in sectors aka triangles where each triangle corresponds to a one or more visible walls? Thus avoiding costly raycasting altogether?

Yes, but not triangles. It uses BSP partitioning, so each sector can have any possible convex shape. Raycasting is not suitable for anything more than Wolf3D-like engines, and even in this case rendering walls directly was faster (Wildcat wolf was not a raycaster, too).

2) When you say A500+1MB you mean a 512K+512K Amiga 500 right?

Yes. I meant 1MB total and plus somehow got in the way without me noticing that it completely changes the meaning. So 512k Chip + 512k Slow is the target. But if you have a way to use Fast instead of the latter, it will REALLY help.

3) You already mentioned that level and texture design is important given the limited resolution and color depth. In the video some parts look really good while others have some vertical lines in them (like a trellis). Is this a result of the horizontal pixel doubling or an effect of the textures used?

No texture design was harmed in the process. Just grabbed the assets from FreeDoom and converted them.
I hope I've answered the vertical lines question above in this post already.

4) You say that your motivation is diminishing after working on it for so long. So recognizable . What is needed at this point to make it a complete game? From what you've shown it looks like the engine and game features are there. Even the levels you made look like fun. Music? More level design? Artwork?

A splash screen, and some memory optimizations to make some room for it. An original track using less than 64k samples would help, too (now it uses a placeholder from X-Ceed). Then I could call it Minimum Viable Product, because even with just a single level and completely no SFX, it's a playable pre-alpha version, kind of fun, and even probably something few people would consider buying for a few dollars (I certainly would!).

But what's needed to make a fully finished game? More or less the above, but TONS of it.

Does it mean that the existing PC wad editors could be used (considering, of course, the restrictions of the 'engine') ?

I'm already using Doom Builder 2 for all mapping stuff. You just have to keep all the restrictions in mind (or get bitten by the converter) and check frequently how your map is looking in the real thing (the textures are imported, but lighting is off and 160x100 final resolution really kills).

If the second, do you already have developed something, and if again so, do you plan to release it with the 'playable demo' you mentioned ?

The problem is that the map needs to be compiled directly into the EXE. This was simpler to do than the file/track loader, but now time has come to rewrite this part properly. But modding is what I'd like to do next and expect it to be a big focus of this project.

By the way, I also do believe that making a game with your engine will find his audience (not to be rich but... )

I have scripting engine already working (but compiled into the EXE right now, again), so as soon as I do this part properly as well, you should be able to really do a lot with this.

I don't know if it was mentioned on the stream, as I was not able to watch it entirely, so what is the maximum colour depth that the engine is accepting now for the textures ?

The viewport is 4bpp and that's fixed. But each texel has both halves separate, so you can pick two colors for a texel to be used for dithering.

I don't know how fast is the loading process of the textures, but maybe while the player is using an elevator/lift (high?), this could allow the loading of new textures ?

Textures are part of the EXE right now. But I'd love to have things loaded dynamically. Probably weapons would be first (load new weapon when player changes it), because of how much space weapon frames occupy, and also space for future SFX.

Impressive doesn't do it justice and to read 25 fps on a stock A1200 makes me wonder if I'd get 60fps on my 40mhz 030.

Don't expect much higher than 30 FPS. Chunky 2 planar is done on Blitter, which has the same speed and becomes limiting factor on faster platforms. This could be addressed later, but right now A500 is main focus. But I wouldn't expect 60Hz anyway, because I know that AGA/060 demo coders still do effects in 30Hz because C2P takes one frame.

Without textures on ceiling and floors it is not DOOM, it is only bad clone of Wolf.
And of course four times worse resolution (160x100) and only 16 colors after more than twenty five years after original DOOM looks bad.

I fully agree. But what I do here is still more than I've seen myself done on A500 ever - so I still consider this a fun challenge to try.

[Samurai_Crow]

Originally Posted by Overmann

Do you peeps (the programmers) have any tips on how to get into programming for the amiga. I've not programmed for anything but I've always wanted to. Is it best to learn something modern (python) and then go to the amiga once the concepts of programming are clear, or would you reccomend a different route?

I've spent years looking at the contributions of great coders thinking to myself "I whish I could do that..", and If I had just started when I first had that thought I would probably be contributing by now. It's never too late to start

This is off-topic but programming an Amiga nowadays is a way to show off. It's harder to program in C than Python.

[roondar]

Originally Posted by tolkien

Only a little thing. Do you know what an Amiga 500 is?
Know its limitations and specs? It's not a 199x machine...
Come on guy. See it with perspective.

Yeah, don't bother.

He's comparing an A500 to machines that have a CPU that is on the order of 10x faster (or more if he's talking about a 486) and come with much more memory. Strangely, he concludes the much faster system with the extra memory gets to have more impressive games.

Who would've thought it?

[roondar]

Originally Posted by KK/Altair

Don't expect much higher than 30 FPS. Chunky 2 planar is done on Blitter, which has the same speed and becomes limiting factor on faster platforms. This could be addressed later, but right now A500 is main focus. But I wouldn't expect 60Hz anyway, because I know that AGA/060 demo coders still do effects in 30Hz because C2P takes one frame.

I suppose on AGA you could use the Copper and BPLCON4 to have part of the screen in a form of 'compressed copper chunky' and the rest using standard C2P. Should be enough raster time to change 96 pixels per line (i.e. 48 colours at 2x2 resolution). And using BPLCON4 you should be able to use them in strips of 32 pixels (again, as 16 colours at 2x2) as long as some space is kept in between the changed colour areas.

Would be hard to implement though and probably won't do much that much terms of speed (given it's only 30% of the screen width).

Note: this is just a random thought I've been having about C2P, not a well thought out approach nor a tested one. Probably best to ignore it and continue on using your current way

[invent]

KK/Altair welcome to EAB and thankyou for sharing your insight into the game/demo.

[malko]

Thanks for your answers KK/Altair.
If you need someone to help testing, you can count with me
Just let me know.

[Overmann]

Originally Posted by Samurai_Crow

This is off-topic but programming an Amiga nowadays is a way to show off. It's harder to program in C than Python.

Yeah, I know. I was trying to derail the conversation away from the troll.
But I see it's back on track now. I'll seek an answer elsewhere

[no9]

Originally Posted by KK/Altair

SFX would really benefit from having accss to 2 channels or more (e.g. player shotgun + enemy dying sounds), but 2-track music is out of question.

Wolfchild music is actually 2 tracks only
[ Show youtube player ]

[spacey]

just watched the recording of the talk - very very cool!

[Mathesar]

Originally Posted by swinkamor12

I rather try to tell inventor of horse (bitplanes) to use cars (chunky pixels) as rest of world.

Yes! You are completely right. Bitplanes are so 1980's. I have also moved on to better (chunky!) systems. So much in fact that I never use my Amiga for any real serious stuff anymore. I am typing this on my Surface which uses an Intel x86 processor, chunky GPU and it runs windows 10. I would never consider using an 68K processor with 1MB ram and an outdated OS without memory protection for anything serious nowadays.

Now that is out of the way, let's talk about fun, cutting edge, creative programming on a 35 year old machine which I happen to be very fond of.

[Mathesar]

Originally Posted by no9

Wolfchild music is actually 2 tracks only
[ Show youtube player ]

Cool! And only 24 Kbytes in size

[Samurai_Crow]

Originally Posted by no9

Wolfchild music is actually 2 tracks only

[ Show youtube player ]

Another option is to target the CDTV so the CD tracks can play in parallel to the Paula voices. That would also help the texture mapping because the CDTV's 1 meg of memory is all chip RAM. The icing on the cake would be the 2 button wireless controller with numeric buttons to switch weapons with.

[Mathesar]

Originally Posted by Samurai_Crow

Another option is to target the CDTV so the CD tracks can play in parallel to the Paula voices. That would also help the texture mapping because the CDTV's 1 meg of memory is all chip RAM. The icing on the cake would be the 2 button wireless controller with numeric buttons to switch weapons with.

Only problem with that: Not a lot of people (including me!) own an CDTV.
In the spirit of creative programming, maybe what is needed is some creative music tracking as well. Using 2 channels and 64kb memory. The Wolfchild soundtrack shows that it can be done.
Also, IMHO, a game like this doesn't necessarily need an overwhelming soundtrack but more like an atmospheric, low key soundtrack. Maybe perfectly doable with 2 channels? Maybe a tracker artist could comment?

[no9]

I wouldn't bother with ingame music at all. Or with the least priority. 4 channels for SFX in dynamic gameplay with lots of shooting is not a luxury either.

[grond]

@KK/Altair:

About the dithering, have you tried how shifting the repeated line by one pixel looks? This should come for free and might break up the vertical line (but might as well look even worse, one can't know without trying). You could even consider shifting the first line but not the repeated line on even frames and the repeated line but not the first line on odd frames. This would be some kind of time-division dithering. But it might also flicker as hell.

[KK/Altair]

This is off-topic but programming an Amiga nowadays is a way to show off. It's harder to program in C than Python.

C is luxury here. Most of the code here is rewritten in asm for performance.

Wolfchild music is actually 2 tracks only

I don't say it can't be done - it just sounds kind of raw. And even with just 2 channels you could benefit from a batter replayer, with improved channel management. Also I'm really surprised, that wavetable* synthesis didn't get widespread adoption. Paula is ideal for this and it could result in smaller modules and more interesting instruments.

*) Having a few 1-cycle full volume waveforms, and changing volume and waveform every frame according to envelopes & LFOs.

Another option is to target the CDTV so the CD tracks can play in parallel to the Paula voices. That would also help the texture mapping because the CDTV's 1 meg of memory is all chip RAM. The icing on the cake would be the 2 button wireless controller with numeric buttons to switch weapons with.

Stock A500 upgraded with 512k slow is the target, and this is not negotiable. The system I spent half of my childhood on deserves it (I had A600 with 2MB chip, but it's close enough).

Also, the 1MB of chip RAM might sound cool, but this engine would benefit much more from Fast. Blitter, CPU and display DMA are running together most of the time.

In the spirit of creative programming, maybe what is needed is some creative music tracking as well. Using 2 channels and 64kb memory.

MOD trackers impose a lot of restrictions by themselves. I'd love to have a custom playback and tracker here.

4 channels for SFX in dynamic gameplay with lots of shooting is not a luxury either.

You know, maybe saving music for later is the good idea at this point...
Also, this way, we could pair it up to have 2 SFX stereo channels.

About the dithering, have you tried how shifting the repeated line by one pixel looks?

Yes. There are four approaches I can see here, and I've tried them all:
- just don't use dithering - worked in Wildcat, but very limited colors in the long run,
- shifting every odd line by 1px - blurs edges across pixels too much, reducing (the already low) image readability,
- perfect checkerboard - best image quality, but requires building odd lines with extra Blitter pass I can't afford,
- the existing solution with line doubling - acceptable tradeoff between the all three above.

The Cyberwolf preview by Dekadence uses the same approach.

[turrican3]

first thank you for the answers !
When do you plan to share a demo and tools to help you with the stages ?
what is your medium target ?? hard disk, df0 . how much disk do you think will be needed ?

[britelite]

Originally Posted by KK/Altair

Yes. There are four approaches I can see here, and I've tried them all:
- just don't use dithering - worked in Wildcat, but very limited colors in the long run,
- shifting every odd line by 1px - blurs edges across pixels too much, reducing (the already low) image readability,
- perfect checkerboard - best image quality, but requires building odd lines with extra Blitter pass I can't afford,
- the existing solution with line doubling - acceptable tradeoff between the all three above.

The Cyberwolf preview by Dekadence uses the same approach.

Yeah, I also tried shifting every odd line, and came to the same conclusion that it just looked a bit too messy.

[vulture]

could the dithering method be selected by the user?

[AJCopland]

KK/Altair really amazing work, hearing all these little technical bits is fascinating too