xterm and such does indeed handle TTF versions perfectly well, when the fonts are configured to have antialiasing turned off and set to the point that is lined out on your page. Just I notice there's a totally different bitmap format to use.

[ continuing the discussion from here regarding Jo22's Epson MGA: ]

Jo22 wrote:

The chip is an old 2764 by the way (ceramic type, non-CMOS, 70s style),
so there might be a compatibility issue with the reader..

That's a possibility- as far as I can tell, the highest bits of each byte (in the font data) appear to be missing... I don't know much about this sort of thing though. Perhaps someone else has an idea of what's going on?

Jo22 wrote:

Maybe that piece of code is also accessible from DOS (debug, nssi) ?
If you provide me with some instructions, I would be happy to dump the chip once more. ^^

Well, CGA/MDA cards don't expose the font in memory - the character generator ROM is only accessible to the card itself. There's no BIOS extension like there would be with EGA/VGA, so nothing is ever mapped to the system's address range unfortunately.

Jo22 wrote:

Edit: I think I can make out some characters! I've highlighted to what I think are numbers and letters.
But it's hard to see, so maybe there's some corruption going on. Maybe another dump will clarify the matter.

Yeah, I can tell that the entire ASCII range is there, and that there are three fonts: a 14-line one (for MDA mode) and two 8-line CGA fonts (thin and thick strokes, as in the IBM CGA and a few other clones).
It's just that the first few pixels of each character line (a random number of them, from 0 to 3) are missing... and those would be the highest bits of each byte, IIRC. Perhaps it's a matter of reading speeds being too high, or so?

Another approach would be to reconstruct the font based on a decent photo of the screen. 😁 Silly as that sounds, I do have a program that generates a text-mode screen which should be optimal for that purpose... however, it'd only show us a single font (the 'thick' CGA one) out of the three.

Anyway, thanks for the photos and info. Searching some more, it seems like this series from Epson includes at least three cards...
- Epson MGA (Q205-AA - your card)
- Epson CGA (Q505-AA)
- Epson EGA (Q206-AA)

So, a couple more to look out for!

VileRancour wrote:

Well, CGA/MDA cards don't expose the font in memory - the character generator ROM is only accessible to the card itself. There's no BIOS extension like there would be with EGA/VGA, so nothing is ever mapped to the system's address range unfortunately.

I was afraid of that. This reminds me of my old Sharp MZ computer, which also had a dedicated character generator..

VileRancour wrote:

Another approach would be to reconstruct the font based on a decent photo of the screen. 😁 Silly as that sounds, I do have a program that generates a text-mode screen which should be optimal for that purpose... however, it'd only show us a single font (the 'thick' CGA one) out of the three.

Hey, that's sound like a good idea! I've got an old frame grabber somewhere in my cav.. -ahem- in my flat. 😁
So you don't have to rely on a shaky picture taken by a cheap camera.

And I'll also keep trying to rescue that EPROM.
Maybe I can build myself an old school EPROM reader in the near future.

Would be a fascinating project!

VileRancour wrote:

Anyway, thanks for the photos and info. Searching some more, it seems like this series from Epson includes at least three cards […]
Show full quote

Anyway, thanks for the photos and info. Searching some more, it seems like this series from Epson includes at least three cards...
- Epson MGA (Q205-AA - your card)
- Epson CGA (Q505-AA)
- Epson EGA (Q206-AA)

So, a couple more to look out for!

I'll keep looking for them (anyone likes to join the search?)! 😀

ooh snap. this font pack is awesome. I am making a retro disk magazine, and this will make ansi in opengl look normal!

Jo22 wrote:

VileRancour wrote:

Another approach would be to reconstruct the font based on a decent photo of the screen. 😁 Silly as that sounds, I do have a program that generates a text-mode screen which should be optimal for that purpose... however, it'd only show us a single font (the 'thick' CGA one) out of the three.

Hey, that's sound like a good idea! I've got an old frame grabber somewhere in my cav.. -ahem- in my flat. 😁
So you don't have to rely on a shaky picture taken by a cheap camera.

Well, if the ROM really won't play nice with your reader... here's the program if you feel like trying it out:

Hmm, perhaps there's a way to toggle the 8x8 font used in CGA mode between the 'thick' and 'thin' versions (hopefully an easy one, unlike the IBM cards)?

BloodyCactus wrote:

ooh snap. this font pack is awesome. I am making a retro disk magazine, and this will make ansi in opengl look normal!

Cool- which disk mag is that? (and who's your ANSI artist?) 😉

I've been using the IBM VGA font with the windows 7 command prompt for a while now and it feels much more natural. Thanks!

(using VGA8 ttf scaled up, though maybe a 88.8% width VGA9 could also work to match its aspect and keeping the extra line...?)

VileRancour wrote:

BloodyCactus wrote:

ooh snap. this font pack is awesome. I am making a retro disk magazine, and this will make ansi in opengl look normal!

Cool- which disk mag is that? (and who's your ANSI artist?) 😉

I am a huge disk mag junkie, wrote a few in my time too 😜 writing a new engine, basically it will look a bit like imphobia-ish. linux+windows (if I can be bothered). sdl2/opengl.

i dont have an artist or musician. currently the mag only has a codename until I can come up with something better. just working on all the programming and doing resolution scaling for fonts and such. Id rather get it complete before contemplating content, tho I am writing some content as I go..

BloodyCactus: heh, godspeed - make it good. Imphobia-style would demand pretty stringent oldskool aesthetics. 😁

leileilol wrote:

I've been using the IBM VGA font with the windows 7 command prompt for a while now and it feels much more natural. Thanks!

(using VGA8 ttf scaled up, though maybe a 88.8% width VGA9 could also work to match its aspect and keeping the extra line...?)

Was thinking about the whole aspect correction mess. VGA8 itself would actually have to be 83.333% width to maintain original aspect at square pixel resolutions, and 8/9ths of that (for VGA9) would be 74.074...% -- so I think 75% is a good enough approximation that shouldn't rasterize too badly (especially if scaled up by an integer).
Any suggestions on how to work aspect-corrected versions into the collection? names like "Px437 AT&T PC6300-2x TrueAspect" would be pushing it a bit. 😉

I'd still like to take a crack at truly scalable versions of the VGA (/MDA/CGA/...) fonts, without the angularity of Nouveau IBM - think something like this but with sharp 90-degree corners, or so. Haven't found a vectorizer that'd get me half way though...

Knocked out an ANSI viewer, looks pretty sweet. I am loving these fonts... but only at very set font sizes, as noted scaling these using TTF methods if your not on that sweet spot eg 16pt or something they are a mess for ANSI (since ansi use all the filled blocks and such). text might be ok on most font points but the scaled blocks and such...

anyway Ive picked 3 font points for 3 display resolutions (>1600, >1200 <=1600, <=1200) that looks good/works! need to figure out if its worth implementing blink attribute or not...

VileRancour wrote:

Well, if the ROM really won't play nice with your reader... here's the program if you feel like trying it out:

ascii40c.zip

Yes, I'll do this. But sorry if it takes some time. Maybe in a few days or a week..

BloodyCactus: boo on the blink attribute, only iCE colors are real 😉

Jo22: sure man, no rush at all. Thanks for looking into it!

yeah, got iCEColour from day one. would be nice to have blink to be complete, but since I render to opengl textures, I'd neen a complete alternate one for the blink'd screen on/off. pain. just going to skip blink I think.

For anybody that uses Linux and needs to disable antialiasing on the fonts themselves you need to provide fontconfig .conf's for the fonts themselves

I'll provide a example one so that you know what I mean:

90-bigblue-terminal.conf

1<?xml version="1.0"?>
2<!DOCTYPE fontconfig SYSTEM "fonts.dtd">
3<fontconfig>
4	<match target="font">
5		<test name="family">
6			<string>BigBlue TerminalPlus</string>
7		</test>
8		<edit name="antialias" mode="assign">
9			<bool>false</bool>
10		</edit>
11	</match>
12</fontconfig>

Hello everyone,
I've managed to do a composite capture of the Q205-AA card.
Picture quality isn't well, though. The grabber doesn't completely support NTSC, I think.

Well, the vertical inversion can be easily corrected... but the odd resolution, not so much 😉
I might still be able to use that to visually fix the missing bits in the ROM dump, at least for this particular font. Thanks for taking the time to do it!

Ah,ok. At least it can be useful somehow. The odd resolution was caused by the software, I think.
There were several other issues with the software, aswell. For example, the screen shot button didn't work with DXVA enabled.
So I had to disable the whole directdraw acceleration in dxdiag. 😵

And there's still hope for a proper dump! 😀 I''m in the process of making an reader for this old EPROM..
But I've got to be careful. Several of them don't support the 2764 type, but the 27256/27512. So there are some modifications necessary..
If I'm to hasty, I could accidently trigger the /PROG line or cause a short circuit. So I need some time for testing (I'll try other 2764s first).
At least, it's not a 2716 tri-voltage monster! 🤣

Good news! I've got a schematic for a classic parallel port reader..
And so far, I make good progress with the assembling. Hopefully, this one is slow enough for that EPROM.
But before I can try it, I need to get hold of some EPROMs for testing first.
Anyway, I'll do my best to finish this one soon.

PS: I've attached a pic of the reader in its early stage.

That looks like a pretty hardcore project- I salute your skillz. 😁 Curiosity still bugs me regarding why the ROM doesn't read properly with normal methods... hope this works, although there's no need to hurry anything (certainly not on my account!)

Thank you very much for these kind words. 😊
I tried to arrange the wires as tidy as possible, thus making the whole thing to look way too complicated I guess.. ^^
But I really have to thank you. I always hoped I could build something like this myself someday and
I'm really glad I finally had a reason to try to do so.. Overall, it was a "ruminant" experience I'd say.
And it astonishes me that it already works in its current state!

Speaking of this, I've already dumped a few ROMs of several video cards (esp. my Trio32/64's and Virge's ROMs).
And thankfully, most of them match the results of my USB programmer so far (verify ok). 😀

But I have got the feeling there are still things to fix and improve (adding some more caps maybe).
So please don't be disappointed if it doesn't work with the 2764 right away.
- I've created for you another dump of the CGA ROM, but its content differs significantly from the inital file.
And the size is also different, probably doubled. Hope that's okay.

And I'm also curious to what causes this strange behavior of the 2764!
It's not the kind of chip I'm used to. It is higher than a normal one (see picture).
Maybe it was an early high-capacity model ? This would at least explain the height. If it was high-end at the time,
the manufacturing was surely more difficult and they needed more space (heat problems, larger core).
But that's just guessing of mine. The date on the label says 1985, so it's at least possible it was a leftover from an old stock.
Or maybe its appearance was just normal for a chip from the east (sov. chips often also looked different than their western counterparts).

However, this still doesn't explain the issues. Maybe it's just faulty (the floating gates, maybe ) or suffers bit rot ?
And if the chip was indeed somehow faulty, then why does it work so fine on the graphics card ?

Maybe the original datasheet can help us to find out.. The EPROM is a "M5L2764K -2" and was made by Mitsubishi.
It took me some time to realize this. When I compared the logo on the chip with some companies' logos I initally
ended up with Matsushita Eletronics Corp. (they used the same logo at some point, funny).
Luckily, I've found the right one later on, because both companies were present in my USB programmer's chip database
when I searched for a "M5L2764" entry.

I really hope we get this mystery solved. If not, I'll try to get hold on another EPROM programmer next month.
Anyway, I don't feel comfortable to post too much in this nice thread without providing any usable results.

Jo22 wrote:

- I've created for you another dump of the CGA ROM, but its content differs significantly from the inital file.
And the size is also different, probably doubled. Hope that's okay.

Thanks, that's interesting actually-- this one seems to have groups of characters mixed from between the different fonts... but not in any consistent way that I can make sense of, heh:

It's also double the size of your previous dump, so I was hoping that at least all the 'pieces' would be there, but in fact the same data seems to be repeated four times (just by eyeballing it... may be even eight with an odd offset).
The strange order/interleave here is mystifying, but having these extra pieces could at least be helpful. I wonder how much of the total font data could be manually reconstructed from this, plus your first dump, plus your composite capture - I might give that a try in my next bout of spare time. 😁

Once again, I truly appreciate all the effort you're putting into this. If no-one here can tell what's going on with that chip, perhaps the folks at the Vintage Computer Forums may have some insight about why it resists dumping so stubbornly...