Oooo this issue seems to be gone in 6.5, the entities are getting stuck but I should be able to fix that through event_stuck, thank you Conitec for your update

You provided a relevant informational update, DavidLancaster. Thank you.

I ran some tests using 3DGS version 6.40.5 and it seems I couldn't truly reproduce my issue in an extremely simple, reduced, test level under various conditions.
c_moving player on floor (inside) of large tunnel-like poly-structure (.mdl; polygon = on):
*size increase of poly-structure: no
*slight offset of poly-structure from center axis in MED: no
*increase of polygon count of poly-structure (approx. 1800): no
*poly_collision = 2: no
*increase of polygon count of player: no
*for player, serious 'animation frame' movement of mesh relative to first frame and axis in MED: no
*intentional intersection of small poly-structure and larger poly-structure: no

I think my worthless 'bounding box intersection correction theory' is, for the most part, kaput.

I'm getting issues with the hulls getting stuck inside each other, if I use a very basic shaped model, or place a box in MED with a 100% transparent skin around my player, they don't get stuck. I'm getting a problem with c_trace and use_box, it seems using it on my player and NPC models with lots of animations and verticies really reduces the frame rate even calling it 4-6 times, if I use just a box model with it then it doesn't touch the frame rate. I assumed c_trace uses the bounding box of the model, yet it seems the more complex model I'm using the slower c_trace and use_box is...I hate these sort of issues, and it may be something I'm missing, I set up a basic test level and it is the player model itself causing the problem, I've done things like removing all the animation frames and testing it, it seems to be the model causing the issue.

These sort of things make me want to go back to ent_move and attach invisible bsp blocks around my models for collision, it's tedious to set up and I just wish c_move and c_trace work like I assume they should

Yes, traces cut my fps in half too sometimes ... I've added a sleep in it as far as 'height tracing' goes for enemies ... My terrain are relatively flat, but I can imagine games in which this might be a very big problem indeed ... You can forget about constantly tracing, when having more than 5 enemies visible at once.

What did you do to fix the model dependant problem? I might have such a problem too, sometimes enemies get stuck in one particular model.

Cheers

If after a number of tests, it seemed that I had 'strong' confirmations that there were certain bugs in the engine, especially in the latest public beta version, I might not spend much of my time describing such bugs in threads that are not in thread category 'bug hunt'. I might certainly assume that such potential bugs might never be fixed if all descriptions of such bugs remain in threads that are 'unseen' by the relevant, responsible developers. Of course, I'm not implying that my previous statements in this post actually apply to any individual or observed instance in particular.

Unfortunately, I don't believe I actually have the time to effectively test and properly report all suspected bugs in 'Conitec betas'.

You're right off course, but just because a certain behavior of code has changed since the last update doesn't always mean it's a bug, nevertheless I've experienced some unexpected behavior. Most of the time, those are the result of errors on my side, so I'm not running to the bug hunt forum just yet ..

Cheers

We have the same strong problem here, engine version 6.50.02

Quote:

---

I'm getting issues with the hulls getting stuck inside each other, if I use a very basic shaped model,

---

We are using the same movement code like in the templates. If the entity is
stucking, there is no way out and if it is the player you have to restart the
game. Any idea if it is a bug or do we have to use the USE_AABB in move mode ??