iPad iHitori

Discussion in 'iPhone and iPad Games' started by djflippy, Jul 2, 2009.

  1. djflippy

    djflippy Well-Known Member

    Mar 9, 2009
    4,442
    0
    0
    [app]iHitori[/app] - $1.99

    "Elegant, challenging and fun!

    iHitori for the iPhone is a polished, addictive game for anyone who loves logic games like Sudoku or Hitori. Whether you’re just learning to play or are already a fan, you’ll appreciate the challenges and sophistication of iHitori.

    Game Features:

    A sophisticated puzzle generator that creates a unique puzzle each time you play. No patterns so each game is challenging and fun.

    Control: choose your puzzle size and number of tiles to control the level of difficulty.

    Elegant, attractive Asian-themed graphics: soothing and easy on the eyes.


    Game support: online forum to connect with other users, get your questions answered and submit feedback to the developers.

    User-friendly interface.

    RULES:
    The objective of iHitori is to remove the tiles containing duplicate digits, by marking them in red, in all rows and columns. The remaining digits must appear once and only once for a row or column to be deemed as solved.

    The game is won when all rows and all columns are solved.

    A tile can have three colors:
    White: An unmarked tile.

    Green: A tile that has been touched once. Green tiles are marked temporarily while the user is deciding which of the tiles should be ‘removed’ from the row or column. Green tiles will turn red when touched twice.

    Red: A tile that has been touched twice. Once marked as a red tile, it will be left out when evaluating the puzzle solution.

    All the tiles can be turned from white to green, from green to red, or from red to white.

    Tiles must obey the following rules:
    The digits in the white and green tiles only appear once in any row or column.
    The tiles marked in red cannot touch one another.
    The white and green tiles must be connected in a single continuous path. In other words, the red tiles cannot be positioned in such a way that pockets of non-red tiles are isolated."

    [​IMG] [​IMG]

    [​IMG] [​IMG]
     
  2. Hopefully better than that piece of crap Chillingo one I wasted bucks on.
     
  3. shawn_10

    shawn_10 Well-Known Member

    Jun 11, 2009
    815
    0
    0
  4. djflippy

    djflippy Well-Known Member

    Mar 9, 2009
    4,442
    0
    0
    #4 djflippy, Jul 3, 2009
    Last edited: Jul 13, 2009
    Okay, I've put a little time into this one. First and foremost, I want to mention that it is a beautifully realized version of the original game, presented as it should be, in the Asian thematic style. The sumi-e inspired logo is amazing. The whole game is just pleasant to look at; simple, elegant, understated style.

    The gameplay has elements of Sudoku and Nurikabe, but it is a solid game, in its own right. The goal is to ensure that there are no duplicate numbers in any row or column. There are three states for each number, the default white, the green to indicate a potential candidate, and red to indicate the negation of the tile.

    Everything works as intended, but unfortunately, it does have a user interface issue that I find frustrating. At any time, a tap will cycle a tile through each of the states/colors...unless it would create an illegal move. The problem this creates is when you have a green tile adjacent to red/negated tile, it will not change to red, when you tap it. It does nothing and you can't switch it back to the default white, without flipping the red tile back, temporarily. So you have to do one of two things, either get in the habit of flipping your reds first, which is not ideal, or using the green state, right up until the puzzle is complete, which almost eliminates the usefulness of having a candidate state, in the first place. Other than that one usability issue, I think the game is quite good.

    BTW, aficionados should note that the player can opt for puzzles from 5x5 (with "Tiles to Find" from 5 to 8) up to 9x9 (with "Tiles to Find" from 17 to 23).

    UPDATE: The usability issue described above has been addressed in the latest update patch and it feels great now. Just wanted to make sure I updated my post accordingly.
     
  5. Selina

    Selina Well-Known Member

    Sep 28, 2008
    202
    0
    0
    Kent, England
    I agree this is a beautifully presented game, and works very smoothly.

    I have also found this to be a problem.

    What I wondered though, was whether Hitori is supposed to have 'unique' solutions, because this one doesn't?

    I also feel that this app would benefit from some sort of tracking mechanism so that you can see how you are progressing - at the moment you just play a game, get told you have solved it and that's it?
     
  6. Yes of course it SHOULD. All Japanese logic games are ONE UNIQUE solution. No idea why people clone these games and screw up the fun of them by not making them unique solutions as they were created. Hudson would be rolling in its grave if they knew.
     
  7. elevataincorporated

    elevataincorporated New Member

    Jul 4, 2009
    4
    0
    0
    thanks, we're listening!

    @djflippy: Thanks very much for your review, djflippy. Very glad to hear that you like the game and its graphics. We are taking note of all feedback (here and elsewhere) and will be releasing an update soon.

    @everyone else: Thanks for giving the game a try! We appreciate your feedback as well. I'll have to let our developers answer the technical issues, but we are listening to everything and will be updating/improving the game as much as we can.
     
  8. djflippy

    djflippy Well-Known Member

    Mar 9, 2009
    4,442
    0
    0
    Welcome to the forums! Beautiful game. Glad to see that it's being supported by the devs.

    When you do the update, it would be nice if the tile flip animation was sped up just a bit. It feels a little too long, particularly when you make two flips, in succession.

    I agree with Selina that there is very little feeling of progression. The inclusion of a "Campaign Mode" where the player moved through a series of progressively harder levels, might be the answer. You could still have a "Custom Mode", like you currently have, where the player simply adjusts the settings to their liking. Global leader boards might help too, but now we're talking about a lot of work! :) Addressing the usability issues seems like the first step.

    Good luck on your game.
     
  9. elevatatech

    elevatatech Member

    Jul 5, 2009
    12
    0
    0
    iHitori Update

    First of all, than you all for providing useful feedback on iHitori.

    We just updated the behavior of the tiles so that you now can turn a "green tile" back to being a "white tile" when turning it into a "red tile" would trigger an invalid move. We also changed the duration of the flip animation from .5 second down to .3 second.

    We are going to build the game and submit to Apple today as an update. It may take a few days for Apple to approve it and push it down to iTunes.

    The next update will be adding a switch on the settings page where you can tell the program to pick a random number of tiles without telling you how many it selected. The switch will allow you to also tell the program that you want to control the number of tiles to find manually.

    Concerning the uniqueness of the solution of the generated puzzle, it is a non-trivial problem to solve on an iPhone in a reasonable amount of time. There are two families of algorithms to generate a puzzle within the constraints of the rules of the game:

    1. Start from a pattern and create "holes" in it to determine what the tiles to find are going to be. Once these tiles have been determined, assign digits to them and to the other "non duplicate" tiles.This is the easiest option since the algorithm is pretty simple. The major drawback of this approach is that it generates semi-obvious puzzles where tiles are disseminated in such a way that the player can "guess" where the tiles are since they are all positioned from a pattern with a few tiles displaced. This is not the approach we took with iHitori.

    2. Build a puzzle from a blank grid by distributing duplicate tiles within the constraints of the rules of the game by picking them at random. Once the dupes have been determined, assign the digits. This is the approach that we use inside iHitori. This is a much more complex algorithm since we now need to check that the rules of the game are followed every time a duplicate tile is added. This algorithm generates puzzles that are *not* obvious.

    Approach #2 guarantees that there is *at least* one valid solution to the puzzle. In order to guarantee that there is one and only one solution to a generated puzzle, we would need to implement an extension to the algorithm by doing either of the following:

    2.a. Once a puzzle is generated, solve it and detect if there are at least two solutions. In this case, discard the puzzle and start over.
    2.b. Solve the generated puzzle and if there are at least two solutions, implement another algorithm that will "fix it".

    Option 2.a. has the major drawback that we have no idea how long it is going to take before a puzzle with a unique solution is generated. This means that now and then we will hit a series of puzzles without a unique solution. The game would then get into a long loop of "generate puzzle", "there is more than one solution", "discard the puzzle and try again". This is not, in our opinion, an acceptable solution.

    Option 2.b. is the one we are working on right now. The iPhone has the computing power equivalent to a Pentium 100MHz (we are talking about the CPU, not the GPU) and it is *significantly* slower than a dual core Macintosh. This kind of algorithm that "fixes" a puzzle can be resource intensive. We want to keep the time needed to generate a puzzle under 3 seconds otherwise the user experience degrades rapidly. At this point, we have a first version of the algorithm that "fixes" puzzles but the iPhone is not fast enough to run it under 3 seconds. We are still working on modifying and optimizing the algorithm to reach this goal.
     
  10. Selina

    Selina Well-Known Member

    Sep 28, 2008
    202
    0
    0
    Kent, England
    Wouldn't this also circumnavigate the unique solution issue? By assembling a collection of puzzles 'up front' they would have already been screened for uniqueness? Though you would still need to re-program for the 'custom' mode, so maybe it wouldn't help in the long run... :(
     
  11. elevatatech

    elevatatech Member

    Jul 5, 2009
    12
    0
    0
    Pre-computed puzzles

    The idea of pre-computing puzzles on a machine that has the horse power to do so is an interesting one. We were thinking about doing something like this. But then, at some point, we run out of puzzles. Then, we would have to push an update to Apple with more puzzles regularly, but I don't know if Apple would agree with this approach since they ask specifically about the changes made in an updated version so they can test.

    Another idea we had was to move away from an embedded puzzle generator running on the iPhone to go to a subscription based model, using the new features added in 3.0. This way, we could generate puzzles on our machines, upload to our servers and push them to the iPhone. An obvious drawback is that this means that only users running the new OS 3.0 could take advantage of this.

    Let us know if you think that the subscription model would be interesting.
     
  12. djflippy

    djflippy Well-Known Member

    Mar 9, 2009
    4,442
    0
    0
    Some thoughts

    The primary concern for users would be whether it is a paid or free subscription. Personally, I think the apps that will be successful with the micro-transaction model are the ones that are content heavy; by this I mean, brand new levels, characters, new powers, etc. I think the problem you may face is that the levels might not seem different enough to warrant future purchase, as players may think "it's just numbers...", not levels. If you mean free updates, I don't think the need for OS 3.0 is a big barrier to entry, but you would need to further explain your plan.

    A few puzzle game examples, currently in the app store:

    [app]Nurikabe Full[/app] has a ton of predefined puzzles that ship with the initial purchase of the app. There are so many levels that scale well in difficulty, I think most users will get their money's worth before they run out of content. Another lesson learned here is that they present the puzzles in an excellent framework, which shows progression and allows users to see which puzzles they've completed, but skip around in each difficulty tier. Not to mention the gameplay is brilliant. I think this is a good example of puzzle game done right.

    [app]Masyu Bug[/app] is another compelling puzzle game. Their model is to to deliver three levels of difficulty, split between three different apps. Personally, I would have rather spent $3 on "one with everything". No one wants three icons cluttering up their valuable iPhone space. Now that 3.0 is out, it would make sense for them to deliver the sequels via purchasable updates, in my opinion.

    [app]Jabeh[/app] is one my favorite puzzle games and it uses the method you currently use, but with three levels of difficulty, which is nice. New levels are supposedly downloaded with some frequency, but it is transparent to the user. It's kind of mysterious and I would rather know when I get new levels.
     
  13. I could wait 10 seconds for a new puzzle no problem as long as it's unique. I've spent 10 to 15 minutes wasted on a puzzle because I didn't know there were multiple solutions so I was stuck thinking I did something wrong somewhere in the puzzle which in fact I didn't. For example, on PuzzleManiak ones that are also generated on the fly probably take 10 seconds or so on LightUp IIRC. Easier levels or so are less than 3 secs or so.
     
  14. elevatatech

    elevatatech Member

    Jul 5, 2009
    12
    0
    0
    Update

    We have reviewed the way we generate a puzzle and we have found another algorithm to generate puzzles with a unique solution guaranteed (i.e. there is one and only one solution, guaranteed by the algorithm.)

    For now, we can successfully run this algorithm "by hand" and we are going to code it to see how fast it can run on the iPhone. It is an evolution of the algorithm we have been using so far.

    This looks very promising because, on paper, it doesn't take a lot of operations to make a decision when assigning a digit to a tile.
     
  15. PhagoCychotic

    PhagoCychotic Well-Known Member

    Mar 22, 2009
    1,089
    0
    0
    Thanks, I look forward to this improvement in a future update. When this happens, I will definitely buy right away.
     
  16. #17 mrbass, Jul 13, 2009
    Last edited: Jul 13, 2009
    Same here as soon as I hear back from those who have it that it has one unique solution I'll probably get it too. Also the UI updates sound good too.
     
  17. badmanj

    badmanj Well-Known Member

    Count me in also. I'll happily pay a few dollars more for this if it's unique - I can see you guys are putting the work in and I'm more than content to pay for that. I might be the exception around here though ;-)

    Jamie.
     
  18. djflippy

    djflippy Well-Known Member

    Mar 9, 2009
    4,442
    0
    0
    I gave this a try, when the update dropped, and I think the timing and tweaked flip behavior both feel excellent. Great job, guys!

    I updated my initial post accordingly.
     
  19. elevatatech

    elevatatech Member

    Jul 5, 2009
    12
    0
    0
    Update

    Thanks to all of you giving us valuable feedback.

    We now have a new coded version of the algorithm that generates puzzles. There are still a few tweaks to make to it. It is working fine on an iPod touch 1G (the slowest iDevice released by Apple.) A 9x9 puzzle gets generated and then "fixed" in about 2 seconds which is very reasonable to keep the user experience fun. As I wrote earlier, building an algorithm that generates Hitori puzzles with a unique solution is a tricky problem to solve, especially when it has to run on an iDevice that has limited computing power.

    We should have a new build ready this week, then Apple will probably take 5-10 days to review it before it gets posted to iTunes.

    An interesting side effect of implementing a "fix" to the generated puzzle is that it is now capable of producing grids that are much more difficult to solve. We are going to modify the way the difficulty level is managed and instead of selecting how many tiles are "hidden", we'll simply offer three levels of difficulty: "easy", "average" and "hard". The size (between 5x5 and 9x9) will still be configurable.

    I'll let everybody know when we have submitted an update to Apple for review.

    Thanks.
     

Share This Page