We are writing with an unusual problem which we have never experienced previously in 8 years in RLC. This problem has existed from the beginning of the build a few weeks ago. In a build with a beach angled at a 7.30 degree incline, we repeatedly fall through the collidable walking surface for no apparent reason. There are no other collidable props in the vicinity, we have inspected carefully for any remnants from previously deleted items, and cannot understand his phenomenon. For walking surfaces we have used court floors, surface water and flat planes, many removed and replaced repeatedly, all experiencing the same behavior. The walking surfaces have lengths changed from as long as 20,000 clicks to as short as 6,000 clicks, and the problem persists.
At this point I have exhausted all potential solutions of which I can conceive, and am totally at a loss on how to proceed.
We would welcome and appreciate any input.

Thank you for your input.

I have also had problems with overstretched collidable surfaces in the past. In a test case i did a surface water with a scale factor of 1.0, and even a series of court floors next to each other scaled at 1.0, but the problem persisted. Still fell through not the edges but the middle of the surfaces. I also tried adjusting the angle a bit to no avail.

Thank you for your response.

The center of the build is at x 30,000, y 50,000 and z 70,000. The ramp runs parallel to the z axis, and is centered at x 30,000, y 49548, and z 78543.

The surface water was scaled at x 1, y, 1 z 0.36.

There are no other collidable objects in the area (other that the two surface waters at either end.

The articles to which you referred are excellent: I devoured them when I first started hear some years ago.

You seem to have the correct items.

There are two other questions that follow from reading your reply carefully. You state that there are no other collidable objects. I assume you did check for red lines while in deco mode. It could be good to iterate through all objects by looking at the performance counters. It is not that those itself are important, but they can be a quick way to check where invisible objects lurk around. You might have some very large boxes deep outside this bubble. I designed the matrix build system as performance method. Yet it is possible you might have some big object way outside. If the performance counters do not reveal any object, maybe you need to click at each object in your list.

Alternative, if you still have space in your zaby, just change the z axis or x axis by putting a minus in front of all objects you can find in that bubble. (You can also drop the y by 80000 but that requires a fast mind)
Of course you then need to redo a landing point, but it might be the quickest way to get out of trouble.

It be good if you try to see what the issue is. If you drop through an angled collider then it in general is scaling (y axis mostly).

If you are stuck at the bottom of the stair then it could be that your collider is not deep enough. That happens rarely but still. Almost always it is the horizontal floor scaling that is the problem. If the scale in z or x is more than 2 you have your issue.

If you are stuck at the top of the stair step then it might require some fine tuning in scale and dept. (sorry for those that never made a stair... the following is kinda difficult to follow). Since we are 570y tall for a scale of 1y it follows that 0.01y scale = 5.7y position. Hence if you drop the y position of a collider by 6 and increase the scale by 0.01 you are now 0.3 misaligned with another collider. (6 - 5.7). This minute change is what sometimes stops us when two dissimilar objects are made collidable. Like a court floor next to a surface water might work in one way but stop you in the reverse. So by repeating the above steps you can cycle through 0.3, 0.6, 0.9, 0.2 (drop of 5), 0.5, 0,8, 0.1 (drop of 5), 0.4, 0,7, 1. Any of those might work.

Enjoy.

PS. I seem to have a few days off. Internet is censored where I live and it requires a fair bit of routing and VPN to have a quick look here. Working takes most of my daily life.

Hmmm, I forgot how crappy the forum post reply works... Had to write it twice.

I always have tried to avoid spoiling the forum with nonsense, so have not put some partial explored experience in "rule". But since you did mention that the angled collider fails, I might have to risk being found wrong.

There in my experience at least three more unpublished issues beyond the y scale.

Sometimes rotation over z does unexpected things. In my test zaby (which is a decent idea to have. Just another place where you can quickly pull up a few things to test with same or similar numbers) I would sometimes not have any rotation despite a non-zero number in z. Or it would move wild and uncontrolled. But while the effect was persistent between visits, it did change by location. Or by duplicating the object and deleting the first. I did have more exciting things to do so gave up on this possible rotation z bug. You might have noticed that my free stair tool does keep the rotation of z at zero.

Another issue is rounding. Try jumping into the abyss. After half an hour falling your body starts to disintegrate with eyes, mouth and body parts moving in a disturbing macabre way.

The matrix build method was partial based on the above observation. It tries to use the view horizon of 42300 for speed, aka, you can create a bubble with that max size without getting a sky bleed and without having to process objects in other bubbles. This is best done by centering around the 70k xyz distance. However the rounding showed up sometimes on the 70k y. Stairs would sometimes fail. Hence the correction to keep the max distance at 50k in y direction.

I spend a fair bit of time in the granny engine. Many of my recommendations on optimizing were based on this. I wrote a none-deco viewer that solved a number of deco theft cases. That does not make me the final answer on everything, simply that I have at least some understanding when stating that the optimization will skip certain tasks to get a better frame rate. Hence if things are busy the body animation lags. So too the collider calculation. If there is an busy emitter, or laggy TV, you might while walking go inside the object before the kick-up pushes your y position to stay on top of the surface. Hence slides and stairs sometimes fail while facing something with a high load. Your performance counter might give an indication.

The least likely issue is that of squeezing. It appears that if scaling is really extreme (like a flat plane with 300x and 0.2z) then sometimes it fails. Again this is a rare phenomena that can be solved by moving a few points. My stair calculator warns if things get too "thin". The solution is obvious to replace the collider with something that has a none square box (like a pole or rail).

So I recommend to take one of the following actions:

Swap xz rotation and scale. This would flip the object without too much work but avoids the possible z bug
or
Move the bubble over x or z by putting a minus in front of the position of all the objects in that bubble.
or
Split your surface water into multiple to smooth the scale extremities.

Enjoy and good luck.

PS. Work demands my return. It was good to see that a world I much enjoyed a decade ago is still alive.

This is most likely not the issue here. But for later groups of people I'd like to reiterate one problem that sometimes gets created. When a lot of long timers started they read the Zaby manual as this stuff was new out and the Devs rarely gave us a clue how to do stuff. This clearly showed the green chassis of any object sat clean on the floor, not poking through it.

Later generations of people came in with no modernised Zaby manual in evidence, in fact the existing manual was hidden from sight. A lot of these people started throwing objects in and often wondered why the floors had weaknesses. I went under a lot of these floors, believe you me some were the tiniest of gaps, and pushed the objects back up level with the floor. No more weakness and no more falling through the floor. In the many properties I visited the weakness was often in a random place nowhere near the object that poked through the floor. But on fixing it that weakness vanished and the floor was sound.

Again, that may not be the problem in this case but please people, take heed. Put your objects level with the floor, not protruding through it. And listen to SifuPeter, he has a great deal of experience and knowledge.

I would also ask Siren from Exquisite Designs to comment on the problem. She has a wide scope of knowledge and has built a few beach properties. She also shows, when not too busy, a keen wish to help people where possible. Marta too has a wealth of information. By speaking to a group of people you are most likely to find the best result. Good luck and I wish you well.

It might be useful to see if you could tilt the horizon. That is, normally water would be horizontal and the beach would be angled. But nothing stops you from tilting the water and everything else, aka the sky etc. That way the illusion would be similar, but with a flat collidable.

The other thing to check, as over the past years of head scratching problems , I have come across, have you tried what you are doing in a new template, to eliminate the problem may be a template that glitches when purchasing, I have in the past had issues with collides and some animations of fish etc, that just would not resolve.

But on acquiring a new template and building the same, or even in some cases just applying a layout of the work already done to it, the problem up and vanished like a fart in the wind