[ List Earliest Comments Only For Pages | Games | Rated Pages | Rated Games | Subjects of Discussion ]
Comments/Ratings for a Single Item
Shogi 
. Optimized ZRFs for Japanese Chess and some recent variants.[All Comments] [Add Comment or Rating]very nice, but i need to point out that if you go to the 'traditional board' variant, the pieces are setup shifted one square to the left, which means the leftmost column is off the board and just sort of floating in space, and the rightmost column is empty. fortunately the 'modern' and 'large' boards work fine, and the different piece sets are very nice.
I have just uploaded the beta version of a major new update to my Shogi ZRF. It includes more optimizations. It requires Zillions of Games 2.0. And most importantly, it can now use different tunings of piece values. I have discovered a technique for adjusting piece values, and I have used it to create settings with different piece values. This is a beta release, because I have not yet settled on which set of piece values is the best. This requires lots of testing, and I thought some of you may be interested enough to help find out. If you have ever wondered what the most accurate piece values are for Shogi pieces, this ZRF now gives you the opportunity to run experiments on different hypotheses. Please use this comment area as a forum to discuss the results of your experiments and tests with this ZRF.
(modified comment)
<BR>I haven't yet studied the ZRF extensively, although it is Excellent if it works, and final-version-will-be-Excellent if it doesn't, but I wish to mention that in my soon-to-be-released Chess on a Larger Board with not-so-few Pieces Dropped - a 12x8 version with the later drop of supernumerary baroque pieces which do not find their place in the starting setup -, I use such macros as
<BR>(define 1000-points (
<BR> (verify (not-in-zone? full-zone))
<BR> add add add add add add
<BR>))
<BR>where full-zone is the game board and the squares carrying the pieces to be dropped, but <I>not</I> some extra bogus squares, such as
White-Throne, WhiteKingMoved or OFF, which carry no actual piece.
<BR>The (1000-points)s or (10000-points)s are embedded in a low-priority (bogus-moves) move-type which enables Zillions to discard them at once
while adding their value to the piece values.
<BR>Otherwise, Zillions would value my Can(n)on - a flip piece - more than a Queen, or a Halfling-Withdrawer just like a Halfling-Advancer.
<BR>(The drop of a piece also creates a bogus piece Mark, which incites Zillions to drop the pieces when it has nothing better to do.)
<BR>I wrote two hours ago that extra piece types change the value enhancement brought by such macros, because before I introduced my latest piece, (1000-points) required ten adds in lieu of the current six, and Zillions valued the Orthochess Pawn at 6000 points as opposed to the current 8000, but now I believe that the main reason is that I suppressed some extra bogus squares. Anyway, not-on-board? instead of (not-in-zone? full-zone) doesn't work.
(verify (not-in-zone? full-zone)) always discards <I>any</I> bogus move, whether there is a move priority or not.
<BR>I have just asked Zillions to search the same position for one minute, with and without the line
<BR>(move-priority normal-move bogus-move)
<BR>and I got nearly the same depth:
<BR>8/277000 with the move-priority and 8/268000 without the move-priority.
<BR>So the move-priority doesn't seem very useful, after all.
I just updated the Shogi ZRF again. I added a new tuning. This is tuning #7, which currently stands at second place. The first place tuning is currently #1, and #5 is coming in third. My ZRF avoids bogus moves as much as possible. To raise the value of pieces that can be dropped, it merely splits drops between the King and the piece. Giving drops to the piece increases its value, but having the King handle its drops does not. To increase the value of other pieces, I give them a measured amount of drops. These drops could be used if they showed up in-hand, but they never do. For optimization, I use directions instead of zones to check whether a piece is in-hand. I use up and down for in-hand areas, but I avoid using these directions on the main board.
The new Shogi's ZRF does not work under Window XP. The Images can not be loaded.
Since I don't have Windows XP, I cannot investigate this matter myself. But you can help me out. First, download this file: http://www.chessvariants.com/programs.dir/zillions/Shogi-test.zrf This file is just like the official Shogi.zrf except that only the alldrops tuning is available, which eliminates the need to define two sets of pieces, and the piecelist macro is no longer used to list the images for each piece. Instead, each piece has its images listed separately. After you download this file, follow this procedure: Open the ZRF with Zillions. When it complains about a specific graphic file, add the complete name of that file (directory + file name) to a list. Then edit the ZRF by replacing the image that doesn't work with one that does. It doesn't matter what image you use here. It could be the same piece everytime. Any piece you already know works will do. Repeat this procedure until Zillions no longer complains about any graphic files. Then send me the list of images that would not work under Windows XP. This list will help me figure out what is wrong with the files that don't work. I may then send you some test files to see if I'm on the right track.
I have Windows XP and the bmps I cannot open with Zillions are also those I cannot open with Microsoft Paint. They usually date back to 2000 or 2001, such as the hex6x6.bmp file for Glinski, or the boards for alf-grandchess, alf-courier, alf-chinese... as I wrote in a comment for the Omega Chess zrf. Anyway, I cannot open with Microsoft Paint the bmps which belong to the following subdirectories: Abstract-Shogi/Flip (the bmps in Abstract-Shogi are all right) Abstract-Shogi-Big, except the King Abstract-Shogi-Big/Flip, except the King. Hope this helps.
Yes, I think it has helped. XNView revealed to me a significant difference between those you can view and those you can't. You can view the regular bitmaps, but you can't view the RLE bitmaps. RLE is a form of compression used for bitmap images, but if I remember correctly, whether you can view RLE bitmaps depends upon whether your graphics card supports RLE. I will put together a new zip file later using only regular bitmaps.
In fact, it is only the images contained in the repertoire ' Abstract-Shogi ' which put problem. I opened the file ZRF and I abolished all the links corresponding to this repertoire and, then, the ZRF opened normally.
10 comments displayed
Permalink to the exact comments currently displayed.