Is the Amiga capable of Teletext?

[Bren McGuire]

i dont see such a button
it's a Toshiba 40vb701B and the manual says it has text mode: https://www.manualslib.com/manual/52...page=40#manual
but i dont see it in the remote!
i am in the uk and this is a uk tv it ought to be supported now i am confused

Originally Posted by indigolemon

Sometimes it looks like the below:

[Jope]

Does your remote look different to this? https://i.ebayimg.com/images/g/z2kAA...I4/s-l1600.jpg

[ross]

Time for technicalities.
Some time ago I had made this diagram for reference, I don't remember if I had already published it, so here it is:



In losso code default start line for VBI TeleText code is line

$8

.
As you can see it corresponds to lines 6 and 319 (respectively for long and short fields) of the (interlace) PAL standad.

Although it are acceptable lines I noticed that in "Final draft ETSI EN 300 706 V1.2.1 (2002-12)" Appendix F.4, 'Allocation of Teletext packets to VBI lines', it's reported:
"Some existing Level 1 and 1.5 decoders may not decode Teletext signals on lines 6, 318 and 319. Thus these lines should be used for Level 2.5 or 3.5 enhancement data only, or non-Teletext signals.."
This means that an attempt to try, for those with problems, could be to use the number

$9

as the first line for the data (and the

$17

as the final one).
It's a sure valid first line in both LOF and SHF.

Another attempt could be to move VBSTRT and VBSTOP to better cover the equalization pulses (it is true that black is used as the 'cover color' but black does not actually correspond to blank.
But in this case there would be some complication regarding the $0 line and the VERTB, which would require changing the code, so for now these are only attempts that can be 'partially' successful.

Nice stuff, really

[losso]

You learn so much... It's not as bad since I leave VSSTRT/VSSTOP as-is?

I got a bit side-tracked with wrapping things up, will try to post a more standard-adhering version with more parameters later.

If I see it correctly, you could avoid outputting data in line 6 (Amiga $8) by pushing the displayed data block down, pressing G and B repeatedly. The number of rows displayed is fixed, though.

[ross]

Originally Posted by losso

It's not as bad since I leave VSSTRT/VSSTOP as-is?

Yes, absolutely do not touch it

If I see it correctly, you could avoid outputting data in line 6 (Amiga $8) by pushing the displayed data block down, pressing G and B repeatedly. The number of rows displayed is fixed, though.

Yep, just a line down should be enough.

[Bren McGuire]

i think i need to go to the eye doctor i found the button! now how do i send some teletext content to it? anybody want to share some tutorial? i got raspberry pis

Originally Posted by Jope

Does your remote look different to this? https://i.ebayimg.com/images/g/z2kAA...I4/s-l1600.jpg

[pandy71]

Originally Posted by Bren McGuire

i think i need to go to the eye doctor i found the button! now how do i send some teletext content to it? anybody want to share some tutorial? i got raspberry pis

perhaps https://github.com/ali1234/raspi-teletext

[losso]

Somewhat enhanced standalone Teletext tester (de-wormed ), with source: https://gru.pe/teletest-01.zip

• Sets VBSTRT dynamically depending on LOF bit, as suggested by ross
• Ability to log VBI interrupts and LOF/VBSTRT settings
• Uses Copper IRQ for its work now, so you can't accidentally lock yourself out with VBSTRT/VBSTOP values
• FPS slowdown
• Option to use different magazine for filler rows/packets (i.e. first page digit)
• Initial white level $AAA
• All (?) keyboard shortcuts documented, space cycles through test pages

For stubborn TVs, many options can still be explored, for example page flags and sub-page numbers.

Built with vasm/vlink, i.e.

vasmm68k_mot -quiet -phxass -x -nosym -align -Fbin -Fhunk -o main.obj main.asm
vlink -bamigahunk -o main.exe -s main.obj

If you want to roll your own test pages, you could use https://zxnet.co.uk/teletext/editor/ to edit any of the .bin files.

[Bren McGuire]

is there any way to get teletext content besides a test screen?

Originally Posted by pandy71

perhaps https://github.com/ali1234/raspi-teletext

[paraj]

Originally Posted by losso

Somewhat enhanced standalone Teletext tester (de-wormed ), with source: https://gru.pe/teletest-01.zip

• Sets VBSTRT dynamically depending on LOF bit, as suggested by ross
• Ability to log VBI interrupts and LOF/VBSTRT settings
• Uses Copper IRQ for its work now, so you can't accidentally lock yourself out with VBSTRT/VBSTOP values
• FPS slowdown
• Option to use different magazine for filler rows/packets (i.e. first page digit)
• Initial white level $AAA
• All (?) keyboard shortcuts documented, space cycles through test pages

For stubborn TVs, many options can still be explored, for example page flags and sub-page numbers.

Built with vasm/vlink, i.e.

vasmm68k_mot -quiet -phxass -x -nosym -align -Fbin -Fhunk -o main.obj main.asm
vlink -bamigahunk -o main.exe -s main.obj

If you want to roll your own test pages, you could use https://zxnet.co.uk/teletext/editor/ to edit any of the .bin files.

My TV seems to be quite forgiving: If I lower whitelevel to $555 it starts to glitch and below that it doesn't work. Lowering horizontal shift to $33 update frequency seems to drop (higher is fine), but it still works. Dynamic VBStop/interlace options don't seem to affect that. Anything else that's interesting to check on a setup that works?

[ross]

Originally Posted by ross

This perhaps means that OCS too.. [cut] ..but we are probably going to open a big can of Worms

Confirmed.
It's a big can of Worms..

But we like these challenges

[losso]

Originally Posted by paraj

My TV seems to be quite forgiving: ... Anything else that's interesting to check on a setup that works?

Good! Thanks for giving it a go. So, the VBSTRT change didn't break anything and maybe even works as suggested.

One question would be if the high latency you observed improves with a lower framerate (pressing S repeatedly in the screen with the mouse cursor). If that works, I'll put that option in the "final" version of the game, too.

Originally Posted by Bren McGuire

is there any way to get teletext content besides a test screen?

An "easy" way would be to throw Teletext data at the

teletext

process. On the Raspberry, you run

nc -l 2000 | ./teletext -

and then you can send Teletext data to it on that port via TCP. Or locally:

your_script.py | ./teletext -

. Pointers to projects/demos that use that approach:

• https://www.pouet.net/prod.php?which=67927
• https://github.com/losso3000/420-years-of-teletext
• https://github.com/losso3000/zxnet-teletext-live-edit

I would have to test, but I think there is no easy way to throw, for example, the output of a Teletext page you made in an online editor at the screen directly, as raspi-teletext handles 3 of the 45 Teletext packet bytes itself and requires properly-encoded headers/characters. See BuffersAndEncoding.kt for some of the gritty details.

[paraj]

Originally Posted by losso

Good! Thanks for giving it a go. So, the VBSTRT change didn't break anything and maybe even works as suggested.

One question would be if the high latency you observed improves with a lower framerate (pressing S repeatedly in the screen with the mouse cursor). If that works, I'll put that option in the "final" version of the game, too.

That does work. At 8.3fps there is basically no lag (after letting buffered frames render). Btw the box isn't blinking (also visible in previous video of same screen) unless I press 'd', but then the screen stops updating.

[modrobert]

Originally Posted by losso

Somewhat enhanced standalone Teletext tester (de-wormed ), with source: https://gru.pe/teletest-01.zip

• Sets VBSTRT dynamically depending on LOF bit, as suggested by ross
• Ability to log VBI interrupts and LOF/VBSTRT settings
• Uses Copper IRQ for its work now, so you can't accidentally lock yourself out with VBSTRT/VBSTOP values
• FPS slowdown
• Option to use different magazine for filler rows/packets (i.e. first page digit)
• Initial white level $AAA
• All (?) keyboard shortcuts documented, space cycles through test pages

For stubborn TVs, many options can still be explored, for example page flags and sub-page numbers.

Built with vasm/vlink, i.e.

vasmm68k_mot -quiet -phxass -x -nosym -align -Fbin -Fhunk -o main.obj main.asm
vlink -bamigahunk -o main.exe -s main.obj

If you want to roll your own test pages, you could use https://zxnet.co.uk/teletext/editor/ to edit any of the .bin files.

I tried 'teletest-01.exe' on my A1200 with Samsung UA23F4003AR LED backlight TV from 2013 connected via composite, it now works to refresh so something has improved, although the updates are at varying slow speed.

On the "Teletext page refresh test" I still have the blinking box, and the "Frame ...." counter gets about 3 - 5 erratic updates every 100 frames at 25.0 fps, sometimes it freezes for a few seconds, then it runs through a few frames, and so on.

Changing the "FPS slowdown" with 's' key doesn't affect much, although it seems to work less erratic with '0' (25 fps) or '1' (12.5 fps) setting on my TV.

PS: At first I thought the "Frame" counter slowed down by a factor of ten or so after a few minutes running the program, but it's probably just the counter display somehow moved the last digit off screen, in other words I only see 4 out of 5 digits (this goes for both Amiga and Teletext screen). Also, the 'teletest-01.exe' binary in zip file doesn't have the executable flag for some reason, had to use 'protect teletest-01.exe +e', perhaps intentional.

[paraj]

Originally Posted by modrobert

I tried 'teletest-01.exe' on my A1200 with Samsung UA23F4003AR LED backlight TV from 2013 connected via composite, it now works to refresh so something has improved, although the updates are at varying slow speed.

On the "Teletext page refresh test" I still have the blinking box, and the "Frame ...." counter gets about 3 - 5 erratic updates every 100 frames at 25.0 fps, sometimes it freezes for a few seconds, then it runs through a few frames, and so on.

Changing the "FPS slowdown" with 's' key doesn't affect much, although it seems to work less erratic with '0' (25 fps) or '1' (12.5 fps) setting on my TV.

PS: At first I thought the "Frame" counter slowed down by a factor of ten or so after a few minutes running the program, but it's probably just the counter display somehow moved the last digit off screen, in other words I only see 4 out of 5 digits (this goes for both Amiga and Teletext screen). Also, the 'teletest-01.exe' binary in zip file doesn't have the executable flag for some reason, had to use 'protect teletest-01.exe +e', perhaps intentional.

Can't comment on the TT technicalities, but once frame counter goes over 9999 it spills into the next line (last digit is visible for me on amiga and TV). Zip files don't support "advanced" file attributes (like +e or +s) so that's completely normal and annoying.

[modrobert]

Originally Posted by paraj

Can't comment on the TT technicalities, but once frame counter goes over 9999 it spills into the next line (last digit is visible for me on amiga and TV). Zip files don't support "advanced" file attributes (like +e or +s) so that's completely normal and annoying.

Aha, OK, I noticed the counter spills over and becomes negative after a while as well.

[ross]

losso relaxing after his work on Amiga Teletext:
https://yle.fi/aihe/tekstitv?P=837#2

[modrobert]

Originally Posted by ross

losso relaxing after his work on Amiga Teletext:
https://yle.fi/aihe/tekstitv?P=837#2

[losso]

It feels appropriate to post a little update here: Teletext on Amiga has been done before, back in 1994, using an ECS-equipped Amiga 500!

I’m not even mad, for from it – figuring out all of this without today’s online resources and off-the-shelf video grabber hardware, studying printed Teletext specifications obtained from then-current TV standards institutions: What an achievement!

So, all hail to Martin, pioneer of the real world’s first native Teletext output on the Amiga (as far as we know right now… )

Technically interesting: Martin, too, ended up with basically the same solution, i.e. stretching 720 hi-res pixels to 736 hi-res pixels with pixel duplication.

[ross]

And, surprise surprise, I also managed to make it work on an A1000, therefore in a much older machine than an Amiga ECS!

It's more complex than on ECS, I assure you
When will I be able to publish it? I do not know...

But 'someone' saw it work