Duplicate foreground pixel, instead of background, in ninth column, so line drawing characters connect about waitvid.2048

[konimaru]

Tricky, one row type (odd) would come for free (shr vs asr). The even one is not so simple. I'll have a go regardless :)

[lutherjohnson]

I'll let you know what I did in the meantime, to get something that I could use, so you don't spend too much time on it, unless you want to :)

I found that 640x480 with 30 rows of 8x16 characters was much easier to get good timings for, on my monitor. I modified the Parallax (Chip Gracey) Hi-Res Text driver to use an 8x16 font (I'm using the one W.L. Schneider (zx81vga) prepared for use with your 128 P drivers), and then changed the driver to expect an attribute (reverse video only) in odd bytes in the screen buffer, so I can still have that feature. I don't have full color per character, as in your code, I only can set colors by rows, but I have an 8x16 font with stable timing in a resolution where the line drawing characters connect, and that's sufficient for getting my product ready. However I am very interested in what you come up with, just because I find this Propeller video stuff fascinating. Thanks for your attention and reply, I've learned quite a bit looking at your code over the last six months or so :)

[konimaru]

OK, that gives me some breathing space :) The timing on this one is already tight so I have to come up with better code.

I've learned quite a bit looking at your code over the last six months or so :)

I hope it wasn't too horrifying. But I'm glad this collection is of use to someone out there.

[lutherjohnson]

Yes, please take your time. Your code has been very useful, to many people, thank you.

[konimaru]

OK, I believe a have a way forward, at a cost. The colour table will have to move to hub memory and I'm not 100% sure I can keep the blink mode (bit 0 of colour attribute). But otherwise it's looking good.

[lutherjohnson]

O.k., I will take a look when you're ready, but some info on how I use it, and where I am - 1) I don't care about blink mode. I converted the blink mode to a reverse video mode by just commenting out one of those places where the colors were swapped back, based on time, if the low bit of the attribute was set, I swapped the fg/bg colors, unconditionally. I did this because i wanted the full character set, so I couldn't use the high bit for a "reverse video" set of 128 characters. So I moved that feature to the one attribute bit we had in this code, and ditched blinking. As I mentioned before, I converted a code page 437 font file that zx81vga was using with your 128 P drivers to use with this POC 80x25 code. 2) I'm o.k. with the modified Gracey code I have now, it's doing the job and it's fast. But your code is more feature-ful, and also, by the way, more generally happy with a larger range of monitors, without having to modify parameters and re-build. So if it's got some issues, I'm o.k., I have a back up, but if it is ready soon then I might make it the default in my system, and make the code I'm using now, an optional set-up. Or the other way around, but I'll find a way to use it and exercise it, and maybe that will yield more useful information, too. Thank you for your efforts here. I have comments saying nice things about you in my Propeller source code :) Luther

…

On 11/22/2018 12:10 PM, Marko Lukat wrote: OK, I believe a have a way forward, at a cost. The colour table will have to move to hub memory and I'm not 100% sure I can keep the blink mode (bit 0 of colour attribute). But otherwise it's looking good. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AqWwHwvzkyFVoaT4Y52U7GE_cdQc8g4Yks5uxvaWgaJpZM4YYBYc>.

[konimaru]

There is now a 80x25 RAM CDUP driver available which duplicates column 8 as 9. Palette resides in hub memory and provides 256 user-definable entries. Said entries must include the idle sync pattern, i.e. %%RGB_2_rgb_2. This is something I'm unhappy about and this will be fixed if possible.

But for now you can at least use the driver and maybe provide feedback.

[konimaru]

Can you tell a bit more about your product?

[lutherjohnson]

O.k., I'll have a look. Thank you ! Luther

…

On 11/24/2018 01:28 PM, Marko Lukat wrote: There is now a 80x25 RAM CDUP driver available which duplicates column 8 as 9. Palette resides in hub memory and provides 256 user-definable entries. Said entries must include the idle sync pattern, i.e. |%%RGB_2_rgb_2|. This is something I'm unhappy about and this will be fixed if possible. But for now you can at least use it and provide feedback. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AqWwH12HSgwvI6Yn3joDPf5J9cFKoHsiks5uyav1gaJpZM4YYBYc>.

[lutherjohnson]

Here is my web site: http://cpmaker.com There is a demonstration video, showing the machine I took to the San Diego Maker Faire. Note Marco Maccaferri's RC2014 VGA terminal on the prototyping / experimenting area. That board will soon, in a few days, be replaced by a new one I made, but Marco's board was so valuable to me as I ironed out the software. There are a couple of documents that describe the prototype, and my dialect of Lisp, they are a little out of date. At this point, I no longer rely on the Zilog development board, for CPU, or a Microchip development board for USB, I have my own boards for everything. The USB host controller is a little board that is 1" x 2", and the replacement VGA controller that will be back from fabrication in a few days is 1.25" x 2.75". And of course, ask me any questions that come up. Thanks for your interest ! Luther

…

On 11/24/2018 01:44 PM, Marko Lukat wrote: Can you tell a bit more about your product? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AqWwH84FIGNyx6jFAzZv32YvcT1QCR7Tks5uya-kgaJpZM4YYBYc>.

[lutherjohnson]

One thing I just checked on is that the copy 8th column to 9th is done on some characters, but not others. In other words, some characters do what you did before, just blank/background in column 9, but line drawing characters copied column 8 to column 9. I'm not sure if there was a look-up table or what. Can you point me to the place in your code where you copy the column from 8 to 9, and cause it to be there ? Based on the character being displayed we want to do it this new way or the old way. It's probably possible to make the decision far enough upstream so that we branch to one set of code or another, and that code can run as it does now, we wouldn't want to be checking this character-specific thing in time-critical loops at the lowest level. I'm sorry I got this bit wrong, but it seems like it might be incrementally do-able with the code you just finished. Then no fonts need be adjusted, we just add a table that indicates for which characters we want a copy or a blank in the 9th column. Luther

…

On 11/24/2018 01:28 PM, Marko Lukat wrote: There is now a 80x25 RAM CDUP driver available which duplicates column 8 as 9. Palette resides in hub memory and provides 256 user-definable entries. Said entries must include the idle sync pattern, i.e. |%%RGB_2_rgb_2|. This is something I'm unhappy about and this will be fixed if possible. But for now you can at least use it and provide feedback. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AqWwH12HSgwvI6Yn3joDPf5J9cFKoHsiks5uyav1gaJpZM4YYBYc>.

[konimaru]

Oh dear! I guess I better don't tell you how much downstream this really is :) Anyway, I didn't get much sleep last night having caught a cold and I came up with a better solution than the current state of affairs. Stay tuned!

[lutherjohnson]

O.k., thank you. Whatever you come up with is bonus, I have what I need for now. Take care and get well. :) Luther

…

On 11/26/2018 04:18 AM, Marko Lukat wrote: Oh dear! I guess I better don't tell you how much downstream this really is :) Anyway, I didn't get much sleep last night having caught a cold and I came up with a better solution than the current state of affairs. Stay tuned! — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AqWwH4bp8pIDZe6OBR5zwMg1ADPQLPmZks5uy84RgaJpZM4YYBYc>.

[konimaru]

There is now full 9x16 support available (80x25 RAM NINE). Also, I'm sure you have a more useful font file sitting around. If you don't mind I'd like to extend mine a bit.

Note that attribute and ASCII are now swapped in screen memory (attribute << 8 | ASCII).

[lutherjohnson]

Great ! I will try it. Luther

…

On 11/27/2018 02:47 PM, Marko Lukat wrote: There is now full 9x16 support available (*80x25 RAM NINE*). I'm sure you have a more useful font file sitting around. If you don't mind I'd like to extend mine a bit. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AqWwH6gNlEzGSlXK9xMPZdHApAp5B9twks5uzbMGgaJpZM4YYBYc>.

[lutherjohnson]

Yes, I can grab a VGA 9x16 font, or make one from the 8x16 MCGA / VGA font I have ... and in that process, do the blank or copy thing ... Good choice on the strategy to solve this problem, I like it. On the original adaptors, having one 8x(14, 15, or 16) font that was bent and shaped to adapt to different video modes made sense. Your solution here makes MUCH more sense, certainly better for run time performance, in this scenario. Well done !

…

On 11/27/2018 02:47 PM, Marko Lukat wrote: There is now full 9x16 support available (*80x25 RAM NINE*). I'm sure you have a more useful font file sitting around. If you don't mind I'd like to extend mine a bit. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AqWwH6gNlEzGSlXK9xMPZdHApAp5B9twks5uzbMGgaJpZM4YYBYc>.

[konimaru]

Cool. Thanks for pushing, I wanted this to be resolved for a long time now.

I remember liberating the current font data from some driver file but I'd have to dig around again to find the source. If I'm not mistaken, some characters really used all 9 columns. So once you're done just attach it here and we go from there.

[lutherjohnson]

O.k., it will be a few days before I get to it. I do have the font files I need, from the zx81vga version of your code, I got it from out there somewhere, the author/modifier of that font file was W.L. Schneider.

…

On 11/28/2018 12:13 AM, Marko Lukat wrote: Cool. Thanks for pushing, I wanted this to be resolved for a long time now. I remember /liberating/ the current font data from some driver file but I'd have to dig around again to find the source. If I'm not mistaken, some characters really used all 9 columns. So once you're done just attach it here and we go from there. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AqWwH1lZLKjtqoaEfsNQyIkM4AC3hR-Bks5uzjeHgaJpZM4YYBYc>.

[konimaru]

I had some time at work so I built a full version based on a Windows font file. The graphics symbols connect smoothly so I guess you're good to go with this one.

[lutherjohnson]

Wow ! Cool. Thank you !

…

On 11/28/2018 01:21 PM, Marko Lukat wrote: I had some time at work so I built a full version based on a Windows font file. The graphics symbols connect smoothly so I guess you're good to go with this one. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AqWwH_pa5Ud2RJtlVpk0kioF1-qx4oh6ks5uzvA6gaJpZM4YYBYc>.

[konimaru]

OT: Does the mailing list subscriber on your website work (at all)? Just tried 2nd time but didn't get anything yet.

[lutherjohnson]

It adds you to a list, but then I have to manually send you email, there is nothing automatic.

…

On 11/28/2018 01:25 PM, Marko Lukat wrote: OT: Does the mailing list subscriber on your website work (at all)? Just tried 2nd time but didn't get anything yet. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AqWwH6hN5XsOCU_3F6AwLJRADZgD15Ycks5uzvFCgaJpZM4YYBYc>.

[konimaru]

Ah, I read it as "I get a confirmation email etc".

[lutherjohnson]

Pre-launch page went live on Crowd Supply

I tried your latest demo, very nice, thank you. I will start integrating it into my code tomorrow, it won't be hard, I was using a version of it before I switched over to the modified Parallax code, and I kept the version based on your code live in my tree and operating. I'll start integrating your newest stuff tomorrow, I'll mind the fact that attributes and characters are swapped. The next couple of days are kind of busy with many logistics aside from software, but yes, those 9x16 characters DO connect when they're supposed to, and they are so pretty. :)

Thank you !

[konimaru]

I'll mind the fact that attributes and characters are swapped.

I dare say it will be pretty obvious when you get it wrong :) For reference, the printChar() method covers all the essentials.

That machine of yours sounds interesting. While I had contact with the Z80 in the past (4MHz max) I never played around with lisp. Maybe now is the time :)

[lutherjohnson]

I've integrated your latest code, as of this evening, and everything works great ! Pretty characters. I think this issue is done, feel free to close if you'd like. I will use your 80x25 9x16 720x400 mode in my product, but I will give the customer the option of the 80x30 8x16 640x480 mode (adapted from the Parallax VGA Hi-Res Text Driver) too. They both work well. Your mode has the benefit of the choice of color from a palette, for every individual character, and of course those nice 9x16 characters. The 80x30 mode gives 5 more lines on screen, and allows direct specification of colors, but only row by row. They both work terrific with the new VGA card that we just developed and got back last night. The development of that card benefited greatly from study of the Hobbytronics Serial VGA, Marco Maccaferri's RC2014 VGA terminal, and Parallax's VGA adapter boards. And of course, from Chip Gracey's and your software ! Here's a picture of the latest setup in action:

[konimaru]

Nice setup! I'll definitely keep an eye on your project. If you need further support just let me know. Also, I was thinking about creating a font editor, not sure if you already have something like that available!?

FWIW, the person who triggered all this 80x25 business will most likely use this latest incarnation in his projects as well.

[lutherjohnson]

I don't have a font editor, let me know when you have something, but no urgent need, I'm happy with this font ... Thanks !

…

On 12/02/2018 05:00 AM, Marko Lukat wrote: Nice setup! I'll definitely keep an eye on your project. If you need further support just let me know. Also, I was thinking about creating a font editor, not sure if you already have something like that available!? FWIW, the person who triggered all this 80x25 business will most likely use this latest incarnation in his projects as well. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AqWwH767_eOhtsUP6MpHsBMSM1jkkHBpks5u08DsgaJpZM4YYBYc>.

[lutherjohnson]

When I update the palette, do I need to wait for vertical sync, or otherwise coordinate with the video code for access to this table ? Because it seems that the low bits are used for state of some sort by the video code as it works (sync bits ? 0- didn't really understand your comments about them before).

I am successfully updating the palette in my application without any (apparent) bad effects, but I just thought I should ask.

[konimaru]

Access to the palette is whenever you want. However, prior to version 0.15.nine.4 you had to make sure that the sync bits %%RGBS_rgbS, i.e. bits 8..9 and 0..1 are set to %10. As of 0.15.nine.4 this is no longer necessary, I mask those 4 bits out (and %%3330_3330) while loading the colour value and set them to their correct value (or %%0002_0002).

I don't know how your palette is defined (hex, quaternary or who knows) but for e.g. quaternary it probably improves readability when you keep those bits as 0:

        word    %%022_0_001_0, ...

The demo file uses hex notation and - to be honest - I was just too lazy to flip those bits :)

[lutherjohnson]

Oh, o.k., good to know. I thought it was necssary to have those 2's added in to the palette entries, as they are all set that way in your table. I am loading palette entries without incident, but I am also adding the 2's in to each byte as I do.

…

On 12/03/2018 12:42 AM, Marko Lukat wrote: Access to the palette is whenever you want. However, prior to version |0.15.nine.4| you had to make sure that the sync bits |%%RGBS_rgbS|, i.e. bits 8..9 and 0..1 are set to %10. As of |0.15.nine.4| this is no longer necessary, I mask those 4 bits while loading the colour value and set it to the correct value. I don't know how your palette is defined (hex, quaternary or who knows) but for e.g. quaternary it probably improves readability when you keep those bits as 0: |word %%022_0_001_0, ... | The demo file uses hex notation and - to be honest - I was just too lazy to flip those bits :) — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AqWwH-1VX93oL015uqE3LoY0nLSCA7Hxks5u1NXRgaJpZM4YYBYc>.

[konimaru]

I will use your 80x25 9x16 720x400 mode in my product, but I will give the customer the option of the 80x30 8x16 640x480 mode (adapted from the Parallax VGA Hi-Res Text Driver) too.

Would it be beneficial to modify the Parallax driver to at least make it a drop-in replacement, i.e. same features plus 5 more lines but smaller font?

[lutherjohnson]

In a way, I've done that, because I wrap both drivers in a terminal emulator. I don't know. I'm dithering (!) over which driver to make the default and which to offer as an option. The Parallax driver that I have modified to have an 8x16 code page 437 font, and a reverse video attribute, has some advantages. I didn't have to change the code very much, it was kind of amazing how these changes fit in without really changing the code hardly at all. Actually it wanted to stay exactly the same size in bytes so I didn't have to mess with the init/loading stuff, and it did. It supports three resolutions and their characteristics by just commenting and un-commenting "tables" of constants, and I can and have adjusted timing constants to give good performance but match a wider range of monitors. However it doesn't support a color choice for each character. The 8x16 font also has the advantage that I get 30 lines by 80, in 640x480, instead of 25 by 80 in 720x400, Last, the Parallax code has been used underneath a text GUI system that's available on the Parallax object exchange, and that code will work with this code, unchanged. The original Parallax code had only a 128 character 8x12 font, and I needed something better, so after trying Marco's RC2014 VGA terminal code and seeing how it was derived from your code, with a 9x16 cell and an 8x16 font, I went to your latest release and made a couple of mods to it (reverse video for the attribute instead of blinking, and add the second 128 characters from zx81vga's font file), and I was going to stay with that. And then one morning about a month ago I realized that the line drawing characters didn't connect. Marco's changes to your code to keep an 8x16 cell didn't work for me on my monitor at the time - I now know exactly what went wrong with that, and could probably adjust those timings properly - I have done so in the table of values in the Parallax code. But at that time, I thought for sure I needed 9x16 and 720 by 400 to have the best resolution that worked well for the monitors I had tested. So I asked you what could be done with your 720x400, 9x16 cell code to make he characters connect. Hence my suggestion from the tricks that IBM MDA and MCGA used, to turn an 8x16 font into 9x16 on the fly. And then after I asked you about that, I started looking at adapting the Parallax code. I'm accessing all of this code, wrapped in an ANSI terminal emulator, so I'm reconciling features and compatible use through that layer, but it's a fact that the two drivers have a different sense of palette vs. direct color, color by character or color by row, and those differences come through. I see that you have made many versions of this kind of code, it did occur to me that you could do what you have done in just about any resolution. You could probably easily make an 8x16 version of your code that works just like your 9x16 code, at least from the outside, to the user of it. (Marco's mods to your code, with a few adjustments to timing for compatibility with more monitors, would be that). Frankly, I'm happy with 8x16, first I wanted your 720x400 9x16 just because I wanted something better than 8x12, and it was there. Then I needed characters to connect, and Marco's 8x16 mods didn't work on my monitor, so I figured out how to turn Parallax's 8x12 code into 8x16. Then in response to my questions, you delivered a 9x16 cell with a proper 9x16 font. However you got rid of the old attribute encoding, and lost the blink/reverse video bit. That caused other complications. Yes it's working but there are some things that I have to do different, and it's not really the same, I can switch the colors but that's not the same in all cases as inverting bits from the font, independent of changing colors. The difference in model peeks through. So I don't know. I've gone from using the best of what I could find, making small mods for my application, to now considering specifying/making exactly what I want, and that means I have to decide what that is. I don't want to ask you to do too much work. And I can't guarantee that I will use it after that. But if an 8x16 640x480 driver with full font, with timings that are easy for me to get to and change, now that I have more experience changing timings to match monitor quirks, showed up, I would try it. But I might still prefer the Parallax code. I have what I need for now, both drivers work nearly identically under my terminal emulator, and I'm going to provide both drivers as options to the customer. I think I should get the product out there, see what people want, and then go from there. I want to thank you so much for all this work that you have contributed. Right now I need to put the VGA part of my project aside for awhile and finish some other things so I can ready the project for sale. But I expect we will come back to this topic. Thank you ! Luther P.S. if you want to converse outside of this GitHub project, feel free to reach me at [email protected], or my phone number is on the front page of my web site, http://cpmaker.com. Thanks again !

…

On 12/06/2018 01:57 AM, Marko Lukat wrote: I will use your 80x25 9x16 720x400 mode in my product, but I will give the customer the option of the 80x30 8x16 640x480 mode (adapted from the Parallax VGA Hi-Res Text Driver) too. Would it be beneficial to modify the Parallax driver to at least make it a drop-in replacement, i.e. *same features* plus 5 more lines but smaller font? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AqWwH2Bj8xz2QWOcb8Y5EEboYj_CW9FUks5u2NwBgaJpZM4YYBYc>.