Yup, what codah says: you want the Pin behavior:

https://www.scirra.com/manual/99/pin

stricky

Page thru the manual; you'll get a couple more badges, should push your rep up to where you can post url's.

Anyway, I really don't have any projects I can reference that have the massive number of images yours does; my current project uses 10mb image memory, so it's not that great a comparison.

However, looking at the memory usage of node for a couple sample exports, I did notice something interesting: the 64bit build seems to consume twice as much memory as the 32bit build, at least for my game (~900 events).

It's worth mentioning that I'm using an older 32bit build of node that I find to be more stable, as per this thread:

Also, you say you can't preview your game anymore to debug...that seems pretty bad, frankly. Are you using an SSD or a traditional hard drive?

As long as your game is targeting PC, I really think you'll be okay. Even a cheap, low end notebook has at least 4GB of memory in it.

Your coins shouldn't be solid (I'm guessing they are). Solids stop other solids. I can jump on the platforms fine; but the player runs into the coins.

Oddly, the player moves thru the coin when jumping...I really can't make more specific suggestions without seeing a capx.

Well... mostly I want the uniform motion, which I can do mechanically, but I was hoping I could just add them to containers dynamically.

I think you are skipping some steps on us here. Can you explain what you are trying to do in a little more depth?

linkman2004

Actually, I don't know if this quite answers the question. Is it more efficient to make all objects on a layer the same blend mode as the layer?

For example, if I make a layer additive, but all the objects are blend mode 'normal', the objects on the layer 'add' to the layers below, but not to each other.

However, if I make both the objects AND the layer additive, the objects add to each other, in addition to adding to the background.

Finally, I can make the layer 'normal' blend, and all the objects 'additive'; in this case, the objects add to each other, but the layer itself is blended normally.

So, really, it's three different effects. But...which is the most efficient? I'm tempted to think there isn't really much difference, unless the objects themselves have different blend modes from each other.

I was just playing around with this yesterday. C2 has no built in path tool, so you have to go at it with raw math.

Here's what I have so far. I'll warn you, it's quite crude:

https://www.dropbox.com/s/cv2gume289gl2 ... .capx?dl=0

I'm sure I could smooth the paths out with a little more work; the main problem is that, with lots of paths, this would become a nightmare to make sense of in layout view. Add a few more paths and it starts to look like your level is sprouting warts.

Thankfully, I don't need this in any of my current projects. I think I'll wait for a proper path plugin.

1. Open two instances of C2: your old project and your new one.

2. Go to the event sheet you want to copy. Click on the very first event so it is highlighted.

3. Now, scroll to the bottom of the sheet, hold shift, click on the very last event.

4. All you events should be highlighted now. Press copy.

5. Go to your new project. Go to an event sheet (empty, other events, no difference). Press paste.

6. If the new project does not satisfy all the dependencies referenced in your copied code (obj and var names, behaviors, etc), then C2 will tell you the first error it runs across.

7. Add/tweak objects/vars/behaviors until you can paste the event sheet in without issue (at least, without c2 complaining).

8. ???

9. Profit!

You shouldn't be having any trouble. Did you look at the auto-runner example that comes with C2?

Start a new project, and when 'template or example' comes up, search for runner. This uses platforms with the solid behavior + bullet (to move the plats).

You mean at runtime? I don't think you can do this.

If you just want certain objects to be created when another object is created, you can accomplish that with an 'on created' condition.

However, if you want to take advantage of the picking capabilities of containers (their most useful feature IMO), you'll have to figure out another approach.

Wow, these event based engines are murder on the cpu... for R0J0hound's event based example (raycast4.capx), I get 10fps and 90% cpu on an i5 3570k.

I assume your game is broken into multiple layouts for different levels?

If so, save a copy of your project somewhere (so you don't accidental go ctrl+s) and experiment with deleting layouts. The memory usage estimate is based on your most intensive layout (whichever one has the largest variety of objects, usually). You may find that a particular level is chewing a lot more memory than your others.

Another question is this: are you targeting desktops exclusively? Because, if so, this may be less of a problem. Most modern desktops/laptops have tons of memory; 2.7GB is a lot, but there are games that use much, much more. As you said, node seems to be dividing the memory across multiple processes, so you are unlikely to hit the 32bit limit.

However, if you were targeting mobile...well, I don't think you are anymore.

The most memory you'll find in a mobile device is 3-4gb, and that's only for the highest end phones and tablets. With the overhead of other apps and OS processes, your game still wouldn't run. It would probably be more trouble than it's worth to strip down your art to the point where you could sandwich it into a mobile's memory.

TheWyrm

Hmm, I missed that in the beta build notes. I guess if ludei's webview build is cordova compatible, then it really isn't any more work for C2 to support it. Well, hey, I'm all for someone giving Intel some healthy competition.

It still looks like Canvas+ is going away though..."no longer supported by Construct 2".