shinkan's Forum Posts

  • rozpustelnik When I exporting my frames I always export them at same size - no matter what's the animation or pose, always export them for example at 512x512. This will make sure all frames are correctly align so image point is always on same place. Only when everything is in place and works like it should I will crop them all to reduce size for each frame.

  • newt it's actually true. While making sprite sheets C2 adds now a fringe for each image to get better seamless results (it was added few releases back). Which is great but a bit strange approach, because you can't make perfect p^2 textures anymore. You need to make them a bit smaller to fit p^2 on export... Which again in some cases can be problematic.

  • Seeing as how the export is made to use pow^2 for sprite sheets, wouldn't it be nice to have something that says you have nth much space left for the next nearest p^2?

    To be honest, after a lot of tests I still can't figure out how C2 exports to sprite sheet. Sometimes it will put all frame into a sprite sheet and sometimes it will leave all frames as separate objects... But if Image editor could show me how my sprites will look like after export then I'm all in.

    https://www.scirra.com/forum/viewtopic.php?f=146&t=118152&p=849467#p849467

  • Yes, there is.

  • It should always be only an "Animation editor" not Image editor.

    This is more important

  • besides this Ads issues... any news about music not changing volumes etc?

  • Ah, maybe I was wrong - and it actually is probably more useful that way anyway. If you create an object at runtime then you might create it again later even after destroying it, so keeping its textures loaded helps prevent jank on the next creation.

    Weeell not necessarily xD

    Action to unload object would be really nice.

    Scenario of how this could be useful. I was struggling with this for mobile games

    You have a background made from 20 different objects and a lot of instances of them (tilemaps, tileddbg and sprites), and then on start of the layout you could simply paste all this objects into a Paster plugin and destroy/unload those objects because you don't need them anymore for this layout.

    So instead of ending up with a bunch of different objects you have only one object that uses less memory and is easier to manipulate.

    From my tests 1 sprite with few animations, each one have four frames with big images takes about 50mb of image memory.

    After adding a Paster and pasting all of this sprites memory goes to about 60-65mb - Paster+current sprite loaded into a memory.

    15mb or even 30mb instead of 50mb it's quite a lot for mobiles

    Please consider adding an option to unload objects from memory, pretty please ^^

  • Try Construct 3

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

    Try Now Construct 3 users don't see these ads
  • No: the runtime treats separately objects which are placed in the layout, and objects that are not placed in the layout but created at runtime. If you destroy the last of an object that was placed in the layout, it is still not unloaded until the end of the layout. If you destroy the last of an object that was not placed in the layout, it will unload it.

    hmm, did something change in releases since this post?

    In r187 I have two layouts. layout1 is completely empty, layout2 have a sprite with few frames.

    In runtime on layout1 when I click mouse button it creates Sprite from layout2, at this point image memory goes up to ~28mb. But if I then destroy this object image memory stays at ~28mb.

    It looks like this "If you destroy the last of an object that was not placed in the layout, it will unload it" is not working, because after destroying last instance of the object that was initially not placed on layout1 it will keep it in memory instead of unloading it.

    Edit: Switching between layout1 and layout2 shows image memory correctly. ~3mb for layout1 and ~28mb for layout 2. So i would assume that C2 is not unloading objects from memory even if you destroy the last instance of object from other layout.

  • Yep, build fail, same as others with XDK 1536

  • eli0s Yes. It works quite nice. Also I made a few modifications for my personal use so I get more params to set (like: glow radius, falloff radius, light intensity, light color and most important effect is not visible on transparent pixels of the object)

    But like I said, until effects of that kind can't accept multiple sources you can't really make fun stuff with them. For example you can't have 5 lasers illuminating background wall...

  • I just found a program that parts from the same principle! Maybe it's worth a look. http://crazybump.com/

    Crazybump was around for a very long time, years at least... It can produce nice results but it's not free.

  • There's lot of normalmap creators out there already. Free, paid and even online ones, and never had any problem making them for 2d images. So I really don't get what's the point of making yet another soft that works exactly the same as others currently available.

    And also, don't forget that current normalmap shaders available in C2, official one, and one made by some other guy (which actually makes normalmaps on the fly without a need of additional objects) are still limited to only one light source. Which makes things complicated if you want to do some fancy effects.

  • DOWNLOAD, two flavors!

    (SHADER) original mode7 with alpha channel fixed:

    https://www.dropbox.com/s/l3os31etolq3i ... d.zip?dl=0

    (SHADER) same, but optimized for fullscreen usage (faster!):

    https://www.dropbox.com/s/bclis2lvbige6 ... d.zip?dl=0

    What's the difference between them, except that one have fix alpha channel and other is faster? <img src="{SMILIES_PATH}/icon_e_wink.gif" alt=";)" title="Wink">

    Faster one does'n not work with alpha channel?

  • Don't know if this is a best approach but i always set window size to 1280x720, layout "Unbounded scrolling" to Yes and "Fullscreen" to scale outer. Then I manipulate my images to fit current device resolution. It's been working great for me on desktops and mobiles with same/lower/higher resolutions.

  • Since you have the map coordinates of the FOV tiles, you could test the tile distance with a Pythagorean theorem and than round the result to get an integer. Then you could set a tile frame as desired...

    shinkan

    Can't you just use the tilemap's "set tile range"?

    hehe yeah... I was looking at the "Roguelike" object in the events, and completely forgot that it's just a simple plain tilemap in the end