Incorrect Physics calculation

[Benski Trenkins]

Hello all,

I am experiencing the weirdest behavior lately.

When I upload a set of walls for one of my buildings, I always make a simplified Physics shape. However lately, it seems that the calculation on those physics seem off.

Example: uploaded a set of walls, placed a physics shape under the physics tab in the upload floater and analized it. Simplified it a little and closed holes.

I hit "calculate" and it tells me my model is 0.6 prim on phyisics (roughly 2.5 on download and 0.5 on server)

Rezzed this model inworld, and as expected it tells me it cost me 3 prims.

I set the physics shape from Convex Hull to Prim and voila: 17!!!!!!!!

Simple but straight forward question: Why did it tell me 0.6 on physics and give me an actual 17 on physics???

I know it's the physics, when I check in the build menu - more info (on land impact)

What I do know this is not a viewer issue, I got the exact same behavior on 2 different viewers (Firestorm and LL viewer)

[Kwakkelde Kwak]

The uploader won't tell you the physics weight of your "physics prim".  As far as I know it will tell you the convex hull weight.

[Drongle McMahon]

It's as Kwak says, that's the weight if you leave it as type "Convex hull". There has been at least one feature request to add the "Prim" physics weights here for a very long time, but so far no luck on that. You can't tell until you upload it.

[Benski Trenkins]

That's weird I could have sworn it did tell me till about a few days ago. 

Also after analizing, it tells me n/a on triangles, which never happened before either.

The physics weight seems to be way more than I am used to on similar sets of walls. I am thinking maybe some librairies are missing or such. 

Really weird, this is new to me. 

[Kwakkelde Kwak]

If you hit analyze, the physics engine uses hulls instead of triangles, that's exactly what it does. So n/a makes perfect sense. Only if you do not hit that button you will get a number of triangles. As far as I know it has always been like that.

[Benski Trenkins]

Yeah I looked into it and that is normal behavior. I do see a difference though on what the outcome is. And am at this point reinstalling the official viewer. I think somehow I must be missing files. 

I still am confused on the physics settings. I used that daily, and cannot understand why the physics are now so insanely high. Even with older models I uploaded before when I got exactly the primcount it told me I would get (with physics properties on "prim") are now insanely high. 

On both viewers, on both grids (main and aditi) I am confused.

[Kwakkelde Kwak]

There is a big physics bug it seems, Drongle is looking into it (or at least I hope he still is). No idea if that is recent or not though. Short version is that depending on the order of the verts/tris in the file the physics cost can vary from next to nothing to an insane amount for the exact same shape.

SL will never tell you the cost of a triangular physics shape before/while uploading, Drongle already mentioned that.

[Benski Trenkins]

Then the variable numbers I noticed before were not set to the right physics. 

I see indeed that it only tells convex hull. In my humble opinion, adding the prim physics in the uploader ( or instead of convex hull) should not be somethign we need to ask for, but should be in there. 

Pure logic: we upload a physics model together with our mesh models. In most cases a physics model that is working as we need it. In case of houses specially. I am not one bit interested in the convex hull cost, since that renders a house useless anyways. I am uploading a physics model to people can walk in, climb the stairs etc etc. 

After doing that physics part in the floater, to then see the "convex hull" physics weight is beyond pointless. It is right out idiotic that they did that like this in the first place. 

Just my 2 cents on the subject. This is frustrating to work with. Now I know this I need to preload every single model (which at times can be a couple of dozen a week) on the betagrid first. 

Never did that up untill today after I got the hang of it

 

It also still does not explain the difference in cost.

Like I said, a model (that I uploaded months ago) is set to prim physics, it is 5 prim land impact. (4 on convex)

Now I used that model to test a bit to see what is wrong... now the VERY same model (not altered in any way) after several tests is 19 prim on the same physics settings, And no matter what I try or play with, it stays 19.

Up untill now I thought I figured it out roughly. Couple of dozen houses, couple of hundred uploads up till now, and all of a sudden it seems the rules have been changed? 

After the replies I got here, and thinking back a little bit... I am starting to suspect the last restart?

[Kwakkelde Kwak]

---

Benski Trenkins wrote:

 

After the replies I got here, and thinking back a little bit... I am starting to suspect the last restart?

---

I have a hard time believing restarting simulators or upgrading simulator software has any affect on upload calculations. To me it makes little sense that each simulator has software for calculating uploads, this particular piece of software is probably on another server.

To be honest I have no idea what does cause the difference.

You could file a jira, with a dae file attached and links to the object inworld both before and after you noticed the change. They should be tagged with a date the moment you created them (or at least when they first became part of your inventory), which might give LL a clue.

Btw, I totally agree it is very annoying to say the least you can't see the physics cost before upload.

[Benski Trenkins]

Weird issue, weird indeed.

Yesterday I got so frustrated that I desided to reinstall the offiial viewer from scratch. I had little hope this would help, since I don't think the calculation is done viewerside.

But what you know... it did work. My older models are again what they used to be, also on Firestorm (and did not reinstall that one)

Maybe it was luck, maybe was missing some files, but it seems that the high Li on physics part is resolved now. 

The model I spoke of earlier is again what it should be. 

It confuses me, but it works, so for now, I am happy.