Building Replica Amigas To Preserve Digital Artwork

A few years back, the Andy Warhol Museum ran into an unusual problem. They wanted to display digital pieces the pop artist created on his Amiga 1000 back in the 1980s, but putting the vintage computers on the floor and letting the public poke around on them wasn’t really an option. So the team at [Iontank] were tasked with creating an interactive display that looked like a real Amiga, but used all modern technology under the hood.

The technical details on the electronics side are unfortunately a bit light, as the page on the [Iontank] site simply says all of the internals were replaced with “solid-state hardware” and an Amiga emulator. To us that sounds like a Raspberry Pi is now filling in for the Amiga’s original motherboard, but that’s just a guess. The page does note that they went through the trouble of making sure the original mouse and keyboard still worked, so it stands to reason a couple microcontrollers are also along for the ride doing translation duty.

Milling the curved display lens.

While we don’t know much about the computers, [Iontank] do provide some interesting insight into developing the faux CRTs sitting atop the non-Amigas. There were some promising rear-projection experiments conducted early on, but in the end, they decided to use a standard LCD behind a milled acrylic lens. This not only made for a perfect fit inside the original monitor enclosures, but gave the screen that convex depth that’s missing on modern flat panels.

The end result looks like the best of both worlds, combining the sharp bright image of an LCD with just a hint of retro distortion. With a scanline generator in the mix, this technique would be a great way to simulate the look of a CRT display in an arcade cabinet, though admittedly being able to mill down an acrylic lens of the appropriate size would be a tough job for most home gamers.

[Thanks to Derek for the tip.]

41 thoughts on “Building Replica Amigas To Preserve Digital Artwork

    1. For a simple task or pure demonstration purposes, an old laptop hardware running FreeDOS and Fellow may do, even. Along with Kick Start 1.x & Workbench 1.3, for that authentic 80s look in blue/white/orange. :)

      For a more sophisticated or authentic look though, some really good RGB CRT emulation might be to consider. So an old laptop might not be enough. Something that resembles the monitor the artist used really faithfully. A simple scan line simulator, as used by the arcade enthusiasts, is not enough, I guess.

      Last, but not least, what about the sound effects?
      Museums do focus on visuals a lot, but a computer is more of a living thing as a painting.
      I hoe they add some fans, floppy sound effects, hard disk seeking sounds, degaussing buzzes of the monitor, etc.
      Normally, I would not care about these things when I’d built something for personal use.
      But this “thing” is to present a piece of history, so…

    1. Haha, yes, I can picture this. 😂
      Almost as funny as the story about the retrieval of the image files. The guys simply used KryoFlux to dump the images. After figuring out that XCopy on a normie’s PC was not suited to read an Amiga floppy..🙄 In the press, buzz words like “retrocomputer forensics” were used to hype things up. Ah, well. Not sure if I shoud cry or laugh. 🤣😭
      http://eab.abime.net/showthread.php?t=73441
      (Just hope these -um- modest pictures of Warhol don’t ruin the Amiga’s reputation. It was somewhat more capable than this.)

      Now to the Apollo thing, which is way more important .. That’s really sad. Always liked the old footage of mission control rooms. They look so futuristic and professional. And sturdy, qualitative. Now they look like lame office rooms/bureaus. IMHO.

  1. The monitor looks great. Its too bad the hacks that ‘restored’ the apollo mission control in houston couldn’t have done the acrylic lenses to replicate CRT. They just jammed flat LCD panels where the CRTs used to be.

          1. Mmm… no, not sure, because in a CRT both the inner and the outer part are curved with the same radius, so there should not be “lens effect”; but having one side curved and another flat means that the curvature radius are different for each side.

            (Also, sorry for reporting this comment. Once again I pressed the wrong button when trying to reply)

    1. Maybe heat a sheet and let it droop into the screen shape. Then back fill using a resin with the same/similar refractive index. Stick it in a box and agitate it and suck out the air a bit to remove the bubbles.

    1. I totally agree, but I can also understand the idea behind that -um- “project”. The goal was to present the old artworks on the device type it was being made on. A real CRT would be great, but they are nolinger being made. If we consider that this thing has to run 24/7, it makes sense that real hardware was not used. It would break eventually and then no one at the museum would have the skills to repair it. An aging tube could be refreshed with special tools, for example. But these people neither have the skills, nor the muse to do so. A museum wants something that “just works”. The Amiga has about the same relevance as a lamp there. It doesn’t care if a lamp uses a real light bulb, an energy savings lamp or an LED lamp. It’s all about the result. It’s just show. Literally.

      1. Right. I hear you. In the same vein why bother with painting restoration? My kid’s got his crayons… it’s not like anyone will have the skill to restore those Verniers in a few years anyway. No one will notice.

        My point being that, yea, it’ll cost a few bucks to do this accurately. Would someone not spend it on a Warhol painting exhibit? This seems insane to me.

      2. CRTs may not be made, but they’re not that hard to get either. It’s not like they were thrown in a black hole never to be seen again (would solve our plastics problem though).

      1. Yup. Lots of vintage Amiga monitors and motherboards have succumbed to age. The high voltage circuits in monitors, in particular, often have capacitors that need to be replaced. And tgey were trying to make things look like Amiga, which means analog RGB monitors. They weren’t looking for a random collection of “stuff they can find.”

        This was CMU after all… they don’t want to dishonor any of their alumni with a poor show!

    1. Hello random Hack-A-Dayer from a few dozen minutes in the past, here in the near future, we are not yet confused. :-D

      We need one of those bots like reddit with an x year reminder LOL

  2. How do they solve the issue where the art on an LCD looks nothing like the original on a CRT because of how differently they work?

    https://www.reddit.com/r/interestingasfuck/comments/9xg8w3/the_same_pixel_art_on_a_crt_vs_lcd_screen/

    Overlaying fake scanlines on top doesn’t really help. The problem is that you have a gaussian distribution of brightness around the spot where the electron beam projects through the aperture mask, and these blend seamlessly together with adjacent spots so the LCD needs to have a resolution many times greater to emulate the effect.

    UHD (7680 × 4320) mapped over to VGA graphics (640×480) would have 6.75 pixels per pixel, whereas using a standard HD panel would have 1.6875 pixels per pixel, which does not afford you to play with any sort of realistic monitor effect – only things like fake scanlines that actually make the picture worse than the original.

    1. And the problem is even greater for machines of that era, especially Amiga, which were using all the tricks and hacks – basically the special properties of the CRT monitor – to get the output to look better than what it really should have been.

      Things like using special pixel arrangements to abuse the NTSC color carrier bug to get more colors out on the screen (on top of the HAM hackery), or special patterns to make use of the interlaced scan for transparent sprites.

    2. So you’re ignoring the third comment to that Reddit post, where it literally says that the answer is a two step filtering process, the first being to add a blur to emulate the shadow mask, and the second to add the scan lines? It even links to a YouTube video showing the results.

      1. Amiga graphics resolutions in high color mode went up to 640×512 @ 50Hz (PAL).

        Now, trying to emulate that on a regular 1080p monitor means first of all that every other line is a “scan line”, which is not how scanlines work. The line is thicker than the gap, and the thickness varies with the signal intensity, giving some amount of overlap between scan lines. For the emulation you only have two pixels to play with, so it doesn’t look right.

        Then, the sub-pixel order is different on the LCD. On an Amiga monitor, the sub-pixels are ordered in a triangular mesh. On an LCD they’re arranged in vertical strips. This messes up your pixel art to the point that the colors don’t work the same any longer – there were special combinations where dithering certain colors next to each other caused them to blend. Instead, you get fringe color artifacts.

        The solution to just blur the heck out of the picture does not remedy the problems, it just hides the artifacts, and makes the image look, well, blurred – like you had a bad monitor. It doesn’t emulate an Amiga monitor, it creates something entirely different.

        1. With overseas you could get about 720×576 in PAL. But I think Warhol’s stuff was done in the early days, before paint programs supported that sort of thing. And he was certainly working on n NTSC machine.

    3. They make all sorts of curved OLED displays for phones, they could probably make one curved like a CRT. I think at UHD resolution a CRT-curved OLED would be pretty convincing at a normal viewing distance.

      Let me know when someone can do a convincing emulation of B&W vector graphics. I have yet to see anything look like a real Asteroids display in terms of presistence, brightness, and electron bloom on bright spots like the ‘missile’ dots.

      1. As far as I’m aware, current flexible displays only curve along one axis. This is because they are initially made as a flat surface, and then the curve is added later. In order to have a 2-axis curve like a CRT surface, you’d be unable to start with a flat surface. The problem is the same as making a flat map into a globe shape.

  3. “[P]utting the vintage computers on the floor and letting the public poke around on them wasn’t really an option.”

    Were they afraid some museum-goer might gut the vintage box and ram a Rasperry Pi inside it?

    1. No, it’s 30 year old hardware that’s likely to break down with the smallest amount of interaction with it. Better to gut one and let people mess around with that than kill dozens of Amigas because you have a stick up your butt about the wrong kinds of authenticity when preserving artwork (not the computer)

  4. If you find yourself implementing a CRT simulation, please, do not just duplicate 3 lines and then make the 4th one dead #000000 black. I get it, you zorch every 4th line to try to get that CRT raster scan look, the gap between the electron bean traces; but perfect black fails here. Please, try this: for each pixel on the gap lines, set it to 0.25 of the color above plus 0.25 of the color below. Or maybe use 0.125 of each, if 0.25 isn’t dark enough, etc. The idea here is, those electron beam spots were not as sharply focused as your modern1080 line monitor (or even a 768 line one!), you have to fuzz it up a bit, bc once you notice those laser sharp black lines, you can’t un-notice them, and the illusion is lost.

    This rant brought to you by xscreensaver, which got so so close

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.