dupuqub's Forum Posts

  • As a fix, I would suggest trying other browsers and see how they behave in the meantime, maybe the issue is not present elsewhere.

    As a hail mary, maybe try clearing your web cache and restarting the browser and/or updating your GPU driver.

  • github.com/Scirra/Construct-3-bugs

    You should post your findings here, follow the guidelines, and optimally send it with a minimal example file where the bug occurs and steps to reproduce the crash if possible.

  • I can confirm, it is happening in some of my older projects too.

    I just posted an issue in the github page, but I was unable to find out what prompts the crash.

    If anyone has any information on it, PLEASE share whatever you got.

    If I cannot find the issue I lost some seriously important stuff.

  • Does anyone else get this issue with letters in the tabs not displaying the correct color?

    The texts in the Sheets and Layouts tabs should be white, they even show as white when I try to edit them.

    As you can see below the comments show the correct color, which is a white text. I only have this issue with the tabs.

    Tagged:

  • sine behavior should help you with that.

  • oosyrag I thought you might be interested in this.

    After some testing, I observed that if you only check X OR Y axis when creating the prob table, you get discrepant values when applying to both axis, as in a map for example, so the only thing I changed was testing in X AND Y when creating the table.

    For future reference, this is the final product if someone needs it for 2 axis.

  • ok. Im gonna have to take some time and study what you did, but you definitely solved the issue, thank you so much!

    EDIT: For future reference, since the file might be lost in the future. Oosyrag's solution is the following:

  • can I look at the file you did that. That's exactly what I'm unable to do.

  • I didn't really understand your answer tbh. What should I get from this table you provided?

    My point on flattening the Perlin noise is to get terrain generation (meaning I still need numbers to be close to its neighbours), but without the distinction between common and rare occurences.

    The answer I'm seeking is most likely a formula that gets values from a range (which is definitely needed since Perlin noise diverges further from 0 and 100 the more octaves you use) and transforms it into something linear.

    EDIT: the distribution of your values seems to fix my issue now that I payed more attention, but how did you get these numbers in this table?

  • After normalizing the values to a 0-100 range, I get this curve on the perlin/classic2D.

    On the left are the ranges, as in 0_9 is from 0 to 9, and on the right the percentage.

    But I suck at math, so what do I have to learn to make this curve more linear, meaning I want to make values equally proportional across the board, "flatten the curve" I guess, e.g. like 10% per range.

    EDIT: I did some research and I guess this is a bell curve, or a gaussian, but when researching "how to flatten a bell curve formula" I had no success.

    Tagged:

  • Try Construct 3

    Develop games in your browser. Powerful, performant & highly capable.

    Try Now Construct 3 users don't see these ads

  • Yeah, I think I’m stuck with normalization after all because I’m working with infinite scrolling, so it’s impossible to add all values to a table.

    Again, I really appreciate the help, y’all gave some nice ideas.

  • Yeah… You are talking about normalization. The general point I was making was trying to avoid normalization because it is inherently imprecise.

  • I don’t know if I understand the idea completely, but wouldn’t weighted tables still be affected by the range since they still work under coordinates generated by the noise?

    EDIT: and you mentioned values above or below the peaks would be the same thing, that would be if I already ran normalization calculations and know what the peaks are, right?

  • If you use for instance from [1-30] is water and for [31-100] is dirt, you can choose some sections to be iron ores like from [2-4].

    The less precision I get with normalization, the more I lose ranges to work with.

    Like from [1-30] out of 100 is never really 30%.

    It’s a small detail, and maybe you are not bothered by the very same thing, but It’s there.

  • I clearly want a number from 0 to 1, not 0 or 1.

    My problem is that normalization works differently for each setup. Like if I ran 100 million times to normalize integers for instance, I’ll have to run that same amount for each noise configuration I’m using, and for terrain generation I’d have to normalize billions of calculations or reduce precision.

    My point is that if I had a noise, resembling perlin, but without the normalization, I could reduce initial loading times and also get more accurate settings.

    But from what I gather it doesn’t exist.

    I appreciate the help though. Thx.