Quote:

Hi Fubeda, sorry if I gave the feeling I do not appreciate your work. My comments suffered of my prevention against ZX ports, but after having tried your game I have to admit that it is much better than many other msx "native"games and by far better than the average msx ports I have seen, (but this was easy ) Well done!

Hey man, no problem at all. I have no hard (bad?) feelings for anyone here.
I really feel the game can be a lot better as a native MSX version... in this way
I completely agree with you (I would like a native version better!), but there is
no native version at this time (only a beta, but hey, we have MKid beta for several
years now and it never saw the day light) and I did my best to make the game "as
playable as possible", even with some speed limitations.
I hope dvik do not quit his native version, the tricks you were discussing on
this forum are really clever to evercome the MSX1 pattern limitations and the
result can be a really nice MSX game.

Quote:

I think no one could say : 'Daniel is time wasted': i and other msx people greatly apreciate the effort. I think also that the only drawbacks are that whatever efforts could be, a msx1 3.5 mhz convertion is often undoable due to the differences between the machines, and due to the code that 'emulate' the speccy vram management. Different is the situation in 7mhz or more performant machines.....

Well, this is something important to say: I will be focusing "nice games" (considering my personal opinion, of course), and that means I will be not thinking only "this game has to run on MSX1 systems at 3.57MHz". In fact, most games only will be full and nice on MSX2 with 7MHz and memory mapper. Why:
- MSX2: color palette adjusted to ZX Spectrum one (sometimes improved). MSX1 color palette is sometimes awful for ZX games.
- 7MHz: usually there is the need to *add* code to the game (the games are very cleverly designed, almost no room for optimization, besides the extra time to plot the screen). That and the f****** WAIT at the M1 cycle at MSX always means the game will be slower (ok, Speccy has problems with contended memory (VRAM), but they also do clever tricks to avoid accessing contended memory during ULA work!). So, 7MHz is almost always a must for faster games, with higher frame rates. For slower machines, slower games (or those not full-screen) would run nice... and the faster probably will run in a similar fashion as GnG (the b&w and frameskip features are implemented in the core of MSX screen updater and is game independent.

Quote:

However, Daniel keep up converting!, there is a lot greater probability to have a zx converted game that a msx1 game from scratch. Re-writing a game is extremely time consuming.

Yup! This is exactly the point. Only as a matter of information, there have been in my stack several MSX native games (an Alien Adventure like Snatcher, a remake of NES Gauntlet, a graphical/text adventure mixing Screen 12 and Screen 7, a remake of an already ZX ported game to use Screen 8, and so on)... but they have been there for years and most probably will never see the day light. I simply have not the continuous time (and focus, to do it in a larger timespan) to finish them. In most of them, the basic engine is ready (usually the graphical one), but the lack of time to create the full script interpretor (usually implemented with only a few commands), script generator and graphical data/mappings... made me halt the development.
One day I decided I would take smaller tasks. Translate games, improve games... was the first step. I liked it. Some will say: "but these are huge projects!" ... yes, but they are composed of hundreds of small tasks that give me a result "in real time". That is: I upgrade a game with a palette, and then I am able to see it with better colors. I add a progress map to a game, and soon I can see it working. I add new music to a game (PSG or CD or whatever) and soon I can see the results. If one of these "enhancements" fail, I just skip it and continue adding others. When I feel tired of mess with that game's code, I just release it and start another project. In other words, I can stop the development when I feel it's time, because the software was ready from the beginning!
Well, this is not the case with tape-to-disk conversions nor with porting games... but this porting was a try to see if I could cope with this somewhat larger project (e.g. I must complete at least a minimum number of small tasks to produce something I can release) and I was happy with the result: I didn't got bored in the process and even had the wish to improve the ported game.

I believe this is a good approach for those that want to do something but have no time (or focus) to create a full game. Of course, it requires a lot of knowledge ... but once this knowledge is acquired, you are ready to go.

[a bit offtopic]
Fubeda, do you like the idea of a screen5/screen2 double game ?
IMHO some msx1 games could be updated to msx2 in screen 5 in a way similar to what konami did with kings valley.
Take Goonies, it could become msx2 with limited effort... I do not remember a lot of "tile based" animations and
there is no scroll. It is just a matter of replacing tiles with copies of 8x8 blocks...

I'm not proposing to improve Goonies porting it to msx2 (almost ), just asking what you think of the subject...

I like this GnG port. I mean it remains within what you'd expect of a Spectrum port but it's a nice addition to the MSX Game library and has been well ported. Pitty it never was ported before.

To go to what Artrag say, I do agree that some games like The Goonies (and Knightmare but it's been done already), deserve MSX2 improvements and it wouldn't be too hard. Even a simple port from Screen 2 to Screen 4 would sometime be a nice improvement to give better sprites/colors to games with too much flickering but that are otherwise pretty decent (Dunk Shot comes to my mind).

And if converting Spectrum games is your thing, how about converting some of those 3d racing games we never got on the MSX (Buggy Boy, Lotus Challenge,...). Some of them were not bad at all, even for a Spectrum.

IIRC upgrading Knightmare, due to the scroll, was a true pain in the ass.
I agree that screen 4 helps for sprites, but the main background graphic even with a new palette, stay the same.

More static games could go in screen 5...this is the essence of my proposal.
I have in mind kings valley and I'm trying to recollect which games could be suitable for such a treatment...

Quote:

[a bit offtopic] Fubeda, do you like the idea of a screen5/screen2 double game ? IMHO some msx1 games could be updated to msx2 in screen 5 in a way similar to what konami did with kings valley.

I believe is a good project, but a difficult one (MSX speed-wise).

Quote:

Take Goonies, it could become msx2 with limited effort... I do not remember a lot of "tile based" animations and there is no scroll. It is just a matter of replacing tiles with copies of 8x8 blocks... I'm not proposing to improve Goonies porting it to msx2 (almost ), just asking what you think of the subject...

Hehehe... Several years ago, when developing the Alien adventure, I made a "fast hmmm" routine to do
that sort of thing (in that case, it was used to plot text - 6x10 pixels patterns - data in screen 7, but it can be changed
to do anything) and it was pretty fast. This is the engine (I would like to improve it with things I know now and
I didn't at that time)I was aiming to use in the port of the game to Screen 8 (or 5, if screen 8 proved to be "too
slow". So, I believe it is possible, but goonies has a few challenges: it messes a lot with name table (which
means screen 5 plots): the waterfalls, the giant rocks, falling blocks and waterdrops, blinking objects... I believe
it is doable, but not as easy as King's Valley II (which has almost no background moves besides palette color
cycling). The same goes to Vampire Killer (which some say was developed as a dual MSX1/2 game and then
released only as MSX2 game).
It would not be just a "plot map" thing, but would need a "dirty map" to plot only modified blocks. It is feasible,
since I heard it is the "trick" behind Akin (and the lost CoreDump), but will need some effort to make it run at
the same speed as the original.

Notice that I believe Goonies *has* Z80 available: the game runs only at even interrupts. On odd interrupts it only
runs the audio routine. So, there is (almost?) an entire interrupt to run. Also, sprite cycling can be removed, at the
cost of more attributes to be plot on VRAM... which means the need of a color cache (similar to what I had done on
KMG).
Also, it would need at least 64KB, but probably 128KB of VRAM, since there will be the need to store all patterns in
different colors, since in Screen 5 one doesn't have a color table and "real-time" decompression is out of question
(I do not believe there is that much idle processor power... hehehe).

Well, this is my analysis of the problem.
I believe I have 10% or somewhat like that of Goonies disassembled and commented (it was the the ROM I used to
develop MSXDASM) and was the first place where I saw the forward DJNZ technique.

With PingPong permission the idea is to replace the code that does the pattern name update with something like this

static uchar memory[768];
bufferptr = RAM_address_of_the_window;
for (i=0;yy= sourcey;yy<sourceyy+leny;yy+=8)
{
    for (xx = sourcex;xx<sourcex+lenx;xx+=8)
    {
       patternno = *bufferptr++;
       if (patternno!=memory[i])
       {
                 copy (8*(patternno % 32),8*(patternno/32) ), page1 to (xx,yy),page0 (8x8 pixel fast copy)
                 memory[i] = patternno;
       }
       i++;
    }
}     

maybe adding also double buffering (we have full of room in VRAM) so why not use it

Quote:

I like this GnG port. I mean it remains within what you'd expect of a Spectrum port but it's a nice addition to the MSX Game library and has been well ported.

Thx!

Quote:

Pitty it never was ported before.

I believe they thought a "b&w" and blocky scroll ... and even this way slower port would not be interesting business.
AFAIR, GnG is the first ZX full port in a looooooooooooooong time... was there any non-commercial full ZX game port before?
(I really was not able to find this information).

Quote:

To go to what Artrag say, I do agree that some games like The Goonies (and Knightmare but it's been done already), deserve MSX2 improvements and it wouldn't be too hard. Even a simple port from Screen 2 to Screen 4 would sometime be a nice improvement to give better sprites/colors to games with too much flickering but that are otherwise pretty decent (Dunk Shot comes to my mind).

Screen 2 => 4 is a lot more simple on Goonies.

Quote:

And if converting Spectrum games is your thing, how about converting some of those 3d racing games we never got on the MSX (Buggy Boy, Lotus Challenge,...). Some of them were not bad at all, even for a Spectrum.

I will keep that in mind. When I finish two more, I can try something like a racing game. Could you guys select *one*?
My vote would be for Lotus (Espirit Turbo) Challenge... I don't know it on Speccy, but I was very fond of Lotus 3 on PC (similar to LETC, but fullscreen when playing alone).

Yup, Lotus Esprit Turbo Challenge was ported to the Spectrum and was decent enough (think Wec Le Mans level). Buggy Boy was also nice on the speccy but not nearly as good as the Arcade. You may be able to see videos of those on youtube

Non racing speccy games worth porting: Rainbow Island and Turrican. The latest being a real challenge :-)

What about Rainbow Island and Turrican in vscreen ...

Quote:

With PingPong permission the idea is to replace the code that does the pattern name update with something like this ... maybe adding also double buffering (we have full of room in VRAM) so why not use it

This is nice, but one should test what is faster: dirty blocks list OR screen plot with difference checking.
Also one has to verify that sometimes is needed to change a pattern with the same number because its
color has changed.
One of the problems to add this feature is the amount of work involved with discovering everywhere
when the game modifies the VRAM nametable, pattern table and color table. Konami usually hardcode
VRAM addresses inside shape's data, which is a pain in the ass. In KMG I had to disassemble a lot
of the code just to discover where were every shape so I could modify the hardcoded addresses... to
point to the new nametable... and do the correct pattern number correction when needed. Also, if the
game changes pattern colors on the fly (it is not common for Konami, but in King's Valley they had done it),
there is the need of a 16bit-per-pattern name table (which means a bigger-than-768 bytes nametable).
And when the color of a pattern is changed... ALL places where this pattern appears should be marked
as "dirty".
Using double-buffering may be something nice, but can slow down things. The amount of data that
needs to be plot will increase. Also, a possible delay in displayed action and the real action may
cause some trouble, but this has to be tested (since it's not a faaaaaaaaast game).
Also, remember that there will be the need for two nametables when using FB. That means that
hardcoded addresses must be updated every time you change the active screen, to point to the
new nametable (at main RAM!). Sometimes this means a lot of Z80 cycles spent.

Most of these things I had learn the "painful way": trying to use these things on KMG. Some worked,
some don't. All in all, that means that sometimes a "simple thing" became a "very complex thing".
I think it is a nice project indeed, but the one who accept this challenge must know that is not as
easy as it seems. It's a pretty good challenge, though.
If anyone is interested in doing this, maybe I can help. But I am not currently on the mood to take
such a big challenge... If no one accept this challenge, maybe in the future I can do it. It just will
not be now.

BTW, it would be nice to have variant CD and FM music on Goonies.
Have you ever thought of playing the game with the original Cindy tune?

Quote:

What about Rainbow Island and Turrican in vscreen ...

Are you volunteering ?

I think its a very well done port. I was surprised that it is so smooth. Imo the ZX version itself is a bit too simplified though and its unfortunate that its only three levels. But its nice to finally have a GnG version for MSX

What about Sim City ?

( Sorry, double post )