Comments/Ratings for a Single Item

you're efficient at making pieces quickly.

NB : index.js is modified

Well, it still takes about 1-2 hour, depending on how perfectionist you are. Pieces I still would like to have a Wildebeest, Spider and Octopus. The Spider I could probably do, but I haven't a clear idea for how to fit it on a base. The other two would require an enhancement of the Tube tool, as it now only can do structures that ly in the symmetry plane, and then duplicate those by a translation or shear (which both keep it planar).

I have my doubts about the starting arrangements for :

Carrera: ANBKQBNM instead of ANBQKBNM ?

Janus : ANBKQBNA instead of ANBQKBNA ?

In the Wikipedia's article on Capablanca Chess Carrera has the King on the right. I messed up on Janus, though; that has the King left. If will fix it in my branch.

How about this Spider design?

satisfactory in my opinion

This is my stab at the Octopus:

On problem with the Tube tool is that it maps the total length of all tubes in the figurine on the height of the normal & diffuse map. Because of the many long tentacles that only leaves a very small fraction of these maps for the main body, with a low resolution (e.g. of the eye) as a consequence. It really should have mapped all tentacles just once on the height, putting them side by side, as their circumference is very small. So they don't need much width in the maps. (I guess in fact they could all share the same part of the map.) It seems a bit hard to do that automatically. (Although for structures that get duplicated I could assign them each the half with of the map, instead of the full width, and putting them at different height.) So I suppose I must craete some input instruction to tell it a tube has to use the same map area (at least vertically) as the previous one.

This is not the easiest piece to make for a vertical station.

We could imagine the arms very close together or very short , Here is an idea for a chess piece :

https://cults3d.com/fr/mod%C3%A8le-3d/jeu/octopus-rook-chess-pulpo-torre-ajedrez

On Carrera's chess, I always need to check the original Italian book (available on the web). Carrera was clear that the Champion (RN) is on King's side and the Centaur (BN) on Queen's side. He said that K and Q stand on the same square than they are on regular chess. Does that means that white K is on the right and Q on the left? Or does that means that white K is on a black square and white Q on a white square? The only diagram he included is void, but it has a white square at the lower right.

So I don't know if we have QK or KQ on white side, but I'm sure we have a wing with K-M and a wing with Q-A. So, HG's order seems a valid one.

It is worth mentioning that Murray was wrong on this game, as often when he was describing chess variants. He wrote (p.827) that the Champion is BN and Centaur is RN, which the other way around. Murray is not reliable when speaking about chess variants (see also what he said on shogi, on chu shogi, on Grant Acedrex, ...).

If you switch back to scirocco, check the behavior of the commoner, which takes on its line.

The behavior seems an unexpected empty promotion on his line.

To reproduce: Cf1-e4 d8-d7 2. Ce4xd7 Guc8-c7 3. Cd7xe10=Sq+

If you switch back to scirocco, check the behavior of the commoner, which takes on its line.

The behavior seems an unexpected empty promotion on his line.

To reproduce: Cf1-e4 d8-d7 2. Ce4xd7 Guc8-c7 3. Cd7xe10=Sq+

Where did you test this? For the version I build from the latest version in pullreq I just see a Squirrel appear. And I don't know what you mean by 'takes on a line'. In that same version the Commoner for me just moves like a King.

you have to take the squirrel with the commoner on ist first line to reproduce the promotion message.

you have to take the squirrel with the commoner on ist first line to reproduce the promotion message.

OK, I see. This kind of sick behavior can occur with the modification I made to the input routine for handling multi-leg moves: if you click a destination that has multiple moves to it, it requires a third click to disambiguate those. (And for the first click it highlights the 'via' square rather than the final destination.) But if even after that several moves are left that match the input clicks (that are not identical; because duplicats would be removed to begin with), it just assumes that this must be because they differ in promotion choice (as apparently origin, destination and via square are the same), and throws up the promotion popup.

I don't see how in this case duplicat moves would be prodused here; the Commoner has a simple KingGraph.

[Edit] OK, I see. The move is also a move that would be induced by the Dervish on d10. This is what causes teh duplicat. But apparently it is not recognized as an exact duplicat. This comparison is very picky: even if the properties are the same, but appear in adifferent order, it doesn't think its a duplicat. I will check it out. For the efficiency of the AI it would be better if not even exact duplicats were generated. But it would be quite a hassle to figure out which types already have the move you want to induce.

[Edit2] It seems the move as it is normally generated has a extra ep: false property in it, and the induced move that I generated because of induction by the Dervish doesn't have that. So the input routine doesn't see them as duplicats, and hence assumes that they must differ in promotion piece. Not all moves appear to have this redundant ep: false property, though, so I should test carefully for when to add it.

I have another question concerning minjku:

If I play the fire dragon in i6: should the kirin be burned?

If I play the fire dragon in i6: should the kirin be burned?

Yes, it should. I suppose I forgot to do the test for burning  on area moves?

OK, Scirocco should be fixed now. It seems all captures must have an ep property, even if it is set to false, and non-captures should never have it. The moves induced by the Dervish now conform to that, and indeed the problem in the given position went away.

I did not change anything in the moves induced by the Harpy; these are Knight moves, and I already exclude induction on pieces that have Knight moves. (These are easy to recognize by their ID.) So the induced moves can never be duplicats, as pieces either have all Knight moves, or none at all.

I pushed the fix.

The Fire Dragon should now also burn when it makes an area move.

I confirm. minjiku seems to be well done in terms of webgl visuals: I think you could add now : res/visuals/minjiku-shogi-600x600-3d.jpg

I made a little todo list to take stock of what we could add to improve the existing with what I know:

Minjiku :

-difference between doc and sprite for ninja (ship instead of gate)
-add res/visuals/minjiku-shogi-600x600-3d.jpg

Shogis :
rules for shogi, mini shogi, tori
fix sprite for lateral mover, tiger
decide the use of western skin or not

Carrera et janus :
test castling

Team-Mate Chess :
-Use phoenix, cobra, mortar sprites
-add visuals

Scirocco : update rules with new visuals

werewolf : update res/rules/werewolf/werewolf-initial.jpg with wolf

If I can help, please don't hesitate to ask.

For Team-Mate Chess I still want to replace the Eagle by a Spider. (Which could still use a bit of improvement, which again needs some extensions of the Tube tool.)

Likewise, I am thinking of whether I could produce a credible 3d figure for Ninja. I was thinking of the western stereotype of a head wrapped in cloth that only leave a slit for the eyes, with crosed swords on its back. But I have no idea of whether that will work out.

If K and Q for Janus are swapped, the janus castling table also has to be changed.

For Scirocco I want to use the Spider and Octopus for the pieces that are called like that, plus a Wildebeest that I still have to produce.

I am currently traveling, without access to my Linux machine. So it will take some time before I can make any progress on this.

a shuriken could be an option for a ninja (2d or 3d or both)

a shuriken could be an option for a ninja (2d or 3d or both)

It's what I use, in both cases.

I was wondering if we should not split up the fairy-move-model. There currently are two things in that, which are largely independent. One is the graph functions for common fairy pieces, and the cbPiecesFromFEN. These are for use during game definitions. The other consists of a lot of wrapper routines, for move generation and performing them. These are used during playing, and only needed for pieces with various forms of locust capture. They needlessly slow down the AI in variants that do not have such moves. Variants like Capablanca Chess now suffer from that, just because they generate the pieceTypes from a FEN.

So perhaps we should have bot  h a fairy-piece-model and a fairy-move-model.

it's quite intuitive

We can add this file easily. If you want to split, it is possible.