Half bitmap mode

[Stuart]

With the 9918, the colour mixing is done internally in the chip and it has just one video output. With the 9929, it outputs a luminance signal plus two colour difference signals, and the colour mixing is done in the external modulator. There are a couple of links on the motherboard that have to be swapped when switching between the two chips (probably a different crystal as well). So it sounds like you're just using the luminance signal which gives a greyscale output. Might be because someone has put a 9929 in a US console (which had a 9918), or you're not using the modulator (are you connecting directly to the back of the console?), or the modulator is faulty. Or possibly something else!

[RickyDean]

Ok when you say that a console with a 9929A, if it is a non-US console it will display color due to the onboard circuitry? Mine is a US console and the chip was added roughly to it about 1997, as one of the mods advertised by one of the user groups or individuals back then, and yes it uses the standard TI to composite cable, that you would use, not a modulator.

[Stuart]

If you're starting with a US console, then there are two links that you'd have to swap to route the two colour signals to the rear connector. You'd then need to connect a PAL modulator to combine them. If you've got the version of the modulator in the metal case, then you can pick off a colour composite signal inside the modulator, buffer it with a small circuit on a piece of veroboard (a couple of transistors, caps and resistors) and use that to drive a colour monitor. (Probably possible as well with the modulator in the plastic case but I don't know where in the circuit to tap into).

 

If you've got a copy of the console schematic (on the FTP site) you can see the VDP output circuit and the various links.

[RickyDean]

Have you performed this on a US console before? Do you have a schematic to follow? I can source the parts, but would this work on a US NTSC format as I am in the states and do not have access to a PAL modulator. If it is not able to be done here, then no problem as the console goes largely unused anyhow. I have several. Thanks

[Asmusr]

Instructions for modding the modulators are here:

http://www.ti99iuc.it/web/index.php?pageid=119&pagina=mod_proj1&sezione=4&data=Mod_and_Tuneed

 

I tried with one of my metal case modulators but it didn't work well: the picture was falling over every few seconds. Is that because you need the buffer circuit?

 

I wonder if you could get a 9929A to run with 60Hz output of you put it inside a US console? The component output plugs directly into my Sony Bravia TV set.

[RickyDean]

Never mind after doing some research, I realize that it is a 9928A , out of a ColecoVision in my modified TI console not the 9929A, so I will probably be stuck in greyscale unless there is some more information out there than I had 20 years ago. thanks.

[+adamantyr]

Regarding half-bitmap mode and sprites...

 

So using a single pattern table and color table limits you to only using the first eight sprites. This is clearly a technical limitation of how the 9918A sets up bitmap mode.

 

It's not a HUGE shock that it exists... bitmap mode was added to the 9918 design later, and the hybrid modes are completely undocumented by Texas Instruments.

 

What bugs me is, there's no reason WHY sprite planes should be affected... The character pattern table has no correlation with sprites! Has anyone dug into the technical side of it to determine why?

Edited September 16, 2015 by adamantyr

[Asmusr]

It's not completely undocumented, check out section 8.4.2 in the attached. The sprite glitch is even described.

 

 

When in this mode 32 sprites can no longer be used. If you try to put more than eight sprites on the screen at once, they will start to duplicate themselves on the screen.

 

vdp_pg.pdf

[Stuart]

Never mind after doing some research, I realize that it is a 9928A , out of a ColecoVision in my modified TI console not the 9929A, so I will probably be stuck in greyscale unless there is some more information out there than I had 20 years ago. thanks.

The 9928 has the same luminance and 2 x colour difference signals as the 9929. So you'd still need some sort of modulator/encoder to combine them into a colour composite signal. The PAL modulator *might* work as you pick out the signal before the actual modulation stage.

[Stuart]

Instructions for modding the modulators are here:

http://www.ti99iuc.it/web/index.php?pageid=119&pagina=mod_proj1&sezione=4&data=Mod_and_Tuneed

 

I tried with one of my metal case modulators but it didn't work well: the picture was falling over every few seconds. Is that because you need the buffer circuit?

 

I wonder if you could get a 9929A to run with 60Hz output of you put it inside a US console? The component output plugs directly into my Sony Bravia TV set.

 

The modulator instructions I used were from a Danish website, with the original apparently coming from the TI-Revue magazine. The website is no longer active so I've attached a scan of the printout I made years ago.

 

I built my buffer circuit on a small piece of Veroboard, and fed the composite output directly on a length of coax cable rather than fitting another connector to the modulator case. I didn't bother with the audio output as I have a speaker in my console case. I also took the composite feed from the video input to the modulator block, rather than from the capacitor shown in the article.

 

====

 

You want a 992*8* to give you a 60 Hz output, I believe.

[RickyDean]

Okay thanks, I will have to see what I can do. It is not an urgent project, but one that will teach me a little about the video on a TI. I guess I can source a PAL modulator from ebay? I would like to see all my consoles using full color. It would be easier to replace with a 9918, but that was the reason I replaced it with the 9928 many years ago, was to mod my console and make it better.

[+adamantyr]

It's not completely undocumented, check out section 8.4.2 in the attached. The sprite glitch is even described.

 

 

Thank you, I was wondering if there was a TI document describing it. I've never seen this document before, and it wasn't commercially released, obviously.

[+Lee Stewart]

H-m-m-m—My comments in post #6 are probably only relevant to the use of VRAM space and shoehorning a pseudo-text mode into it because all three modes (Graphics2, Split and Split2) in fbForth (and TI Forth before it) use the same standard values for VDP registers VR03 (>7F) and VR04 (>07). In reality, they are all the same, standard bitmap mode. Their apparent differences are managed by the fbForth system software. I am pretty sure I could set up a “Split3” bitmap mode that uses half of the screen for text; but, it would not be the non-standard, half-bitmap mode under discussion. Sorry for any confusion.

 

...lee

[Tursi]

 

What bugs me is, there's no reason WHY sprite planes should be affected... The character pattern table has no correlation with sprites! Has anyone dug into the technical side of it to determine why?

 

Going entirely on faulty memory that I'm not sure is actually true, but, I THOUGHT somewhere I read that sprites 8-31 were added after the initial design, when it was known how many memory slots were available. The only theory I had was it was related to the way those sprites were plugged into the design.

 

Do we still have contact with Karl? Maybe we can ask for insight.

[Asmusr]

Off topic, but I found this in my archive while looking for a starting point for a sprite duplication test.

 

 

This is a test if the "5th sprite on a line" flag. The center line of sprites have the highest priorities. When one of the other sprites tries to cross the center it will be detected by the VDP as the 5th sprite and the program will reverse its direction. During the demo I'm turning the 5th sprite detection on/off by pressing E/D on the keyboard.

 

Edit: YouTube has really messed this up, the file I uploaded does not have a Bouncy demo at the end.

5th sprite crawl.a99

5THSCRAWL.dsk

[Asmusr]

Related to my previous post, I have sometimes wondered how we would make the bridge in Commando so that the main sprite would move under the bridge without being semi-transparent.

 

 

The answer is, of course, to close the gap in the bridge with 2x4 high-priority sprites, which will make the sprites that move beneath it completely invisible, like when I turn off 5th sprite detection in the 5th sprite video above. You could still have characters under the sprites to allow for more detail.

 

[TInaut]

Is it possible to have a bitmap screen whereby the colour data covers 1 8x8 pixel square? I'm looking at shoehorning bitmap mode in. Any suggestions for bitmap table layout would be greatly received. I think there are only a few combinations with bitmap mode.

 

IIRC half-bitmap mode has less colour information, which might make a useful first step for me.

 

Thanks

 

Check out Jeff White's bitmap explanation in Dan Eicher's compendium at:

 

ftp://ftp.whtech.com/datasheets%20and%20manuals/Hardware/ti%20hardware%20compendium.txt

 

It describes shrinking the color table down to 64 bytes for bitmap mode. The first 8 bytes define the colors for patterns 0,8,16,24,etc. to 248. The next 8 bytes define colors for patterns 1,9,17,25,etc. to 249. So each 8 bytes define the colors for 32 characters offset by 8. The last 8 bytes define colors for patterns 7,15,23,31,etc. to 255.

Edited September 19, 2015 by TInaut

[+Lee Stewart]

Any suggestions for bitmap table layout would be greatly received. I think there are only a few combinations with bitmap mode.

 

As Harry implied in post #12, even though the color table can be shrunk to 64 bytes, it still has only two possible starting locations in VRAM: >0000 or >2000, which then allows only one location for the pattern table, the one not used byt the color table: >2000 or >0000.

 

...lee

[TInaut]

As Harry implied in post #12, even though the color table can be shrunk to 64 bytes, it still has only two possible starting locations in VRAM: >0000 or >2000, which then allows only one location for the pattern table, the one not used byt the color table: >2000 or >0000.

 

...lee

No disagreement, they can only start at >0000 or >2000.

 

To clarify, the pattern and color tables can start at the same address. I am sure someone could put this to clever effect. However, if the color table is shrunk to 64 bytes in bitmap mode, that is only patterns 0 to 7 that need be shared if pattern and color tables occupy the same memory.

[Asmusr]

AFAIK the information in the compendium is incorrect, if you shrink the color table to 64 bytes you will also shrink the number of available characters. Thierry Nouspikel writes:

 

• At the extreme, VR3 = >00 results in >40-byte tables. As there are 8 bytes per characters, this means characters are arranged in 8 groups of 32 identical characters: chars 0-7 are identical to chars 8-15, 16-23, etc.

 

See:

http://www.unige.ch/medecine/nouspikel/ti99/tms9918a.htm#hybrid%20bitmap

[OX.]

Related to my previous post, I have sometimes wondered how we would make the bridge in Commando so that the main sprite would move under the bridge without being semi-transparent.

 

 

The answer is, of course, to close the gap in the bridge with 2x4 high-priority sprites, which will make the sprites that move beneath it completely invisible, like when I turn off 5th sprite detection in the 5th sprite video above. You could still have characters under the sprites to allow for more detail.

 

bridge.png

 

Ooh, does this mean we may have a TI99 version of Commando in the coming future Rasmus?

Edited September 19, 2015 by OX.

[Opry99er]

I freakin hope so!!!

 

Bionic Commando would be even better.

[+InsaneMultitasker]

If Commando is ever written for the TI, my hope is that it will be paired with Rob Hubbard's version of the soundtrack

[Asmusr]

If Commando is ever written for the TI, my hope is that it will be paired with Rob Hubbard's version of the soundtrack

 

Perhaps one could find a way to convert Daniel Bienvenu's great ColecoVision version of the tune to the TI?

 

 

http://atariage.com/forums/topic/162196-commando-music-bass-and-drum-loop/

[TheMole]

You can run Daniel Bienvenu's ROM in Meka (the SMS emulator, which also supports a host of closely related systems including the coleco), and capture the output to a VGM file, which can then be converted by Tursi's SPF tools. I don't have a Windows PC, but it should be a fairly easy process.