Digital cameras are a ubiquitous consumer and professional product here in 2023, and because of the wide availability of parts it’s relatively straightforward to construct one for yourself. Four decades ago though, film was king, but that hasn’t stopped [Georg Lukas] from building a digital camera for the 1984 market. The hardware is definitely from recent years, the extremely affordable ESP32-cam board that many of us will have worked with already. Meanwhile the 1984 part lies in the recording format, it makes EGA 16-colour low-res pictures and stores them in the archaic TGA file format.
A low-res camera is fun, but there are two other angles on this which are definitely worth some time. The first is that his description and code are worth a read for anyone with an interest in programming an ESP32 camera, while the second invites us to consider whether such a camera could have been made using parts available in 1984. We remember camera peripherals for 8-bit microcomputers which were a C-mount lens positioned over a decapped RAM chip, and thus we can’t help wondering whether an RGB split to three of those sensors could have been constructed. Whether a 6502 or a Z80 with 64k of memory could have processed the three images into one is another matter, but at least if any of you want to try there’s a handy 1984 computer still popping up on eBay.
CCDs were available and ADCs too. No need to decap a poor RAM chip.
Complete camera’s were available too…
The point is not the technology, but the availability and cost of it. A thing easily forgotten when looking back into the past. The same goes for “why did people used audio tape instead of disks or harddrives, the technology existed, sure it did, but if you can’t buy it when you can’t afford it, or if you local shop doesn’t offer it, then it isn’t there to be used, is it?
And let’s be honest, using RAM for a camerasensor is just good-old-fun.
“We remember camera peripherals for 8-bit microcomputers which were a C-mount lens positioned over a decapped RAM chip, and thus we can’t help wondering whether an RGB split to three of those sensors could have been constructed. ”
Why go to all the trouble of 3 sensors when you can use one with a colour wheel and take three pictures?
Minolta did in 1995 with the RD-175. For the price of a small car you got two green CCDs, one red/blue CCD and 1.75 Mpx (interpolated).
“Why go to all the trouble of 3 sensors when you can use one with a colour wheel and take three pictures?”
That’s what space probes did. They had/have an HQ monochrome camera and selectable filters, including infrared and other non-visible wavelenghts.
In the hobby corner, the color wheel was in use in the 70s in amateur radio.
The “Frame Sequential” approach used three different transmissions for Red, Green, Blue. Each had a transmission time of 8 seconds, like the original SSTV.
Example:
https://www.youtube.com/watch?v=1q66zazyJLU
Last but not least, the Amiga platform used the color wheel in as early as 1986.
The software/hardware kit “DigiView” supported digitizing color images using a black/white “composite” (VBS) video camera.
The color wheel could be turned manually, or by an electro motor that’s driven by the scanning software.
Because a color wheel in front of a mono sensor is terrible at taking pictures of anything moving.
True. To be fair, real film had the same limitation. I remember that from my youth.. All these ruined photos! 😅.
To its defense, a mono sensor has the highest efficiency/sensitivity. It catches most of the light, while RGB sensors must share that same amount of light among each others.
It’s not the film’s fault you left the shutter open too long. The very best modern digital camera will have the same issue if your settings aren’t correct for the subject and light.
The problem with a color wheel is that 3 distinct images need to be made then combined. Any movement will result in a trippy rainbow effect. I made a rainbow filter/gravity shutter aparatus back in the film days to purposefully get this effect. Moving water was really interesting. Totally gimmicky, but it was fun to make and experiment with.
There are things like exposure times and so on.
Image quality might be lower if the shutter is closing too fast.
Some cameras, like compact cameras, don’t have a provision to change shutter times, also.
There was a dude pre WW1 who used this approach to make color photos of Russia. There’s some amazing photos of people in wonderfully vivid traditional dress, but they all look like they’re trying not to poop because they’re being so careful to not move.
TARGA was not a low-end format, though.
It was used as a 24-Bit Hi-res format on many Shareware CD-ROMs of the 90s.
Those CD-ROMs had multiple folders containing same pictures in multiple resolutions/color-depths.
GIF, as an interchangable format, was used for graphics at 320×200 and 640×480 in 256c, while 800×600, 1024×768 and up were stored in TARGA/TGA (and/or BMP and PCX, alternatively).
In the Windows 98SE days, TARGA was still being recommend for flatbed scanner usage.
I remember my first video card capable of 24 and 32 bit video. It was an Orchid (from when that was a Western Digital product line) VESA Local Bus card. I got the manual and drivers from Western Digital’s dialup BBS – even though WD had sold off their video card division.
WD was the only PC parts company I’ve ever known to provide support for product lines they no longer owned.
What really made that Orchid card go was UniVBE and SciTech Display Doctor. It worked quite nicely with Windows for Workgroups 3.11 on a 486.
Would the 1984 market include file transfer to Big Brother? Much like we do today?
You could actually buy pre-decapped RAM chips in the form of the IS32 sensor. I used to work for a UK company that made consumer level cameras using said chip. It was possible to make a reasonably portable digital camera as they also made an SBC that could interface to the camera module and had 32k of battery-backed SRAM so you could write code to store a bunch of images. It was just a matter of also finding a suitably portable power source to run the thing :)
That was a clever way by Micron to sell off-spec DRAMs. With early 64k parts, yields were rather low due to the large chip which forced manufacturers to do strange things like selling half-dead chips as 32k DRAMs.
Mr Burns: get me George Lucas!
Smithers: George Lucas isn’t available, sir
Burns: then get me his Eastern European counterpart!
I lol’d