pouët.net

I think that "remastering" really old demos is kind of a weird thing to do. Usually the whole point of a demo (especially if it is from the 80s or 90s) is to show off what the creators can do with the hardware of the time within the typical constraints and, well, to win a competition. Remaking a demo with the same content except with a higher fidelity and on a completely new platform leaves nothing to appreciate except for the aesthetics. I don't think these old demos have much to appreciate in terms of aesthetics, they're mostly technical things that you need a good understanding of to appreciate. I know that 90s-cyber-tech is an aesthetic, but you really don't need to stick with Crystal Dream II to explore it.

Not really. The first time someone did this was totally awesome. Everything after that was just trying to chase a trend. ;)

posting again in epic thread

I like this.

I understand that for many demos it used to be that 'showing off what you can do with the hardware' is what it was about. But still, many demos are not like that, and are just to be enjoyed.

So what we need now is just to run this tool on the entire pouet database.

If we have youtube captures for many demos, why not have browser ports for all of them.

Where's the harm in that?

Fuckings to LLMers.

I give a fuck :) I actually like these ports. Also, I shared links to some friends who know the 90s demoscene. I posted links to the ports on retro-forums. And I haven't seen any negative opinions. People seem to like watching old demos in their original form, without MPEG compression and without downloading a gigabyte of data. Not to mention that not all demos on YouTube were captured correctly and quality.

I don't understand what the drama is about.

- These guys make a "browser port". Yes, with the help of AI, but they pay for it themselves and do it in their free time. They don't claim it as their own original product. They don't enter competitions with it. They don't try to sell it.
And if the port plays the demo with perfect accuracy, pixel by pixel, then what's the problem? Well, AI remastering is of course debatable, but in this case it's just an option.

- The people who watch these things understand that it's nothing more than a convenient way to watch and remember their favorite demos, with the cool music and all those cute effects that amazed us back in the 90s.

- Why don't you complain about browser-players for tracker music, for example? Or something like XMPlay? It's not even the same as playing it on an Amiga, GUS, or Sound Blaster, and it even dares to enhance sound! :)

But wouldn't a proper browser emulator be the better option?

Make one :)

If it doesn't require configuring it for every single demo, finding and loading ROMs, doesn't run too slowly, starts the demo with one click, and shows it in the correct aspect ratio... Maybe that would be pretty nice.

Although some demos are a pain in the ass to run even on real hardware :) Some need a specific video card, some don't work correctly if the CPU is too fast. Some need EMS, others need XMS, and for some of them it's absolutely forbidden.

So I doubt you can make a universal emulator that wouldn't need to be configured for specific cases.

If someone wants to do that, there's DOSBox, including a browser version. And technically, isn't it really that different from these AI ports? Besides, I'm afraid that in the near future all emulators will have some AI code in them anyway :)

I'd vote for the online-emulator route. Or also easily installable emulators. Even with AI-based improvements or completely AI-developed emus (I've seen these usescases in reality and it feels better to have them than not).

Here for example is the archive.org's DOSBox running Second Reality. You need to select a Soundblaster variant for it to work.
https://archive.org/details/demoscene_SecondReality-FutureCrew

You compare.

Rationale for going the emulator way and also AI usage in developing/improving them:

1. Emulator coding is more like an enabler, but the involved ppl surely don't get as much attention and praise (as it seems to me) as e.g. demo coders.

2. The way of writing emulators can often be less of an interesting, creative work. Especially for all the fixing, adapting, improving little parts, just to get that one more software running. This might in the end kill the motivation for some emu devs.

3. Only a few emulators might have been technically impressive, especially if the delta between the emulated hardware and the hardware it's running on is smaller. And the more sophisticated the emulation gets, the more it might be overwhelming, up to the point, that someone can't handle it anymore. Or it just takes much longer to get the feature implemented and be made available. E.g. cycle exact emulation of CPUs with the whole chipset.

4. The emulator provides the right stage to "feel" the limitations of the hardware the demo is running on.

5. The improvements in one emulator will enable tons of demos to run, not just one. And usually they are written in a way that they can be ported easily. But this could also be an AI-assisted task.

6. AI use is not on/off or black/white. There are nuances. Experienced coders doing AI-assisted coding vs. coding noobs just judging the result by how the UI looks or what the feedback from users (alpha-testers) is. And everything inbetween.

7. Emulator devs might achive more and do more demos instead. ;)

Stupid me didn't make it a URL. ;) Here it is:
https://archive.org/details/demoscene_SecondReality-FutureCrew

I'm quite torn on this, a refined recompile is much better than DOSBox in the browser which has always been horrible, and YouTube / video in general struggles with demoscene content.

I understand why people are angry but I hope cooler heads prevail over time and we can enjoy more properly synchronised, properly frame paced, refresh rate accurate and non-stuttering demo presentations in the browser in the future.

I just wish the ethics of AI hadn't boiled over the way it has. The technology is fascinating, but capitalism has poisoned the well.

Do you really have to ask?

Fwiw, it would be a fun thing to see an interesting recompilation of the original source, ie. get to see how the Stars tunnel really works for example. I think this tech would be useful for that. But it's still a waste of carbon dioxide in service of the capitalist overlords, so no.

Quoting without comment.

https://github.com/ConspiracyHu/SecondRealitySDL/blob/main/Source/Parts/TUNNELI/TUN10.CPP

Ah i misunderstood, you meant the nooon stars tunnel, nevermind :D

I'm going to guess here, but what I would expect is that they render a regular voxel landscape to an off-screen buffer (such as 128x128) and then just use a polar table lookup to map this to screen space (i.e 320x200)

yes, and the fov of the voxel pass needs to be 360 degrees.

Voxel ball is the same thing, just with the voxel pass turned upside down.

It can be done in a single pass too by having a table that describes how to plot a 1 angle sector on the destination screen.

That's pretty much the goal of https://github.com/sasq64/demarc and its already pretty far along, running everything on pouets top-100 for instance.

But as have been said, it's AI assisted coding so don't bother with it if you're not OK with that.

Oh, except for the "Browser" part. Its for dekstops, preferably Linux.

NoooN Stars Teardown
https://claude.ai/artifact/HFgcoP4vY6ULruAVzEc9fg

+1000

i'd attribute this phenomenon to "fucking idiots with no original ideas in their head suddenly being able to make stuff, in spite of lacking the intellectual or practical skills to do so previously"

what Tom said

What Tom said … and that’s why they are going to defend this tech to the bitter end … because they can now LARP as actual creative people having actual original ideas to communicate … previously, they never got around the “how” to even arrive at the “what” and “why”. add to them those that just procrastinated their skills into oblivion, now getting this poisoned crutch. It’s not going to stop.