Why did the physics cost change just with a toggling

[Chic Aeon]

This is a corner display. I wanted the physics to be correct.  I uploaded with Linden viewer. When it first appeared the land impact was 8 because of the physics cost. I could live with that but it did seem high. I wondered what it would be like if I made it convex hull and turned it phantom. I didn't WANT that but was just curious.  It was 3. Then when I changed it BACK to prim it was still 3.   This is good news BUT ----

 

WHY ?????

[Wulfie Reanimator]

Does the LI stay at 3 even if you pick it up and re-rez it?

[ChinRey]

39 minutes ago, Chic Aeon said:

WHY ?????

Why is one of those questions we simply don't have an answer to. Science has tried in vain for ages and even religions, with their far less stringent QC and lower need for credibility, are struggling to come up with a plausible answer.

However, sometimes we can at least explain what happens or, such as in this case, what doesn't happen. When you switch physics type in an object the viewer doesn't always update the physics weight away so you get an incorrect value. Usually you can fix it simply by welecting some other object and then return to the one with the false weight reading. Sometimes you have to tp away from the trouble or even relog to force the viewer to recheck the weight. Wulfie's solution also works since the viewer considers new copy as a completely different object.

Edited October 27, 2022 by ChinRey

[Chic Aeon]

4 hours ago, Wulfie Reanimator said:

Does the LI stay at 3 even if you pick it up and re-rez it?

No AND they revert to 8 on relog using EITHER the Linden Viewer of FS so this is just and "oops" it seems.  That's a lot of physics cost.  Oh well.  

ODD.     

Thanks for the thought.

[Aquila Kytori]

21 hours ago, Chic Aeon said:

That's a lot of physics cost.  Oh well. 

On 10/27/2022 at 1:43 PM, Chic Aeon said:

When it first appeared the land impact was 8 because of the physics cost. I could live with that but it did seem high.

 

 

Physics costs can be puzzling at times. For non analysed physics 1 or two small and or thin triangles (small, thin relative to the overall size of the model), or a face that is not quite flat can increase the physics cost a lot.

I would have guessed at a cost of between 1 and 2 or even less but to check I did a couple of test uploads.

It is hard to see exactly what the back corner of you model looks like so I made two Physics models, 1 with a right angled back bottom corner and one with a back bottom corner at approx 45°. Both models had the same number of triangles.

And size matters, (for planes type, non analysed physics) the bigger the model the lower the physics cost. I guessed at  8 x 8 x 4m. 

I didn't bother making a high Lod model , instead I used the physics model for both visual and Physics.

 

First with the 45° back bottom corner :

 

When rezzed the physics cost was 2.1

 

Next. to see if in made any difference I rotated this physics model 90° along the X axis in Blender, Applied rotation and exported :

 

Rezzed now the Physics cost was reduced to 1.3   :

 

Same thing with the second physics model, 90° bottom back corner  :

 

Rezzed Physics cost is 1.2   :

 

And rotated before exporting :

 

Rezzed  gives the lowest Physics cost of 0.7   :

 

 

Almost identical physics models, both consisting of 80 triangles can result in a factor of 3 difference in physics cost .  ( 0.7 to 2.1 )

 

What to do with your Physics model ?  Hmmmm , can't see anything wrong with it.

I'm not sure what going on with those 2 (?) angled corner faces, I would try a test upload with those faces removed.

Also to increase the average triangle size of your physics model I would drag the bottom edges of your columns and walls so that they are aligned with the bottom edge of the model.

Larger average tri size  usually means lower Physics cost.

If your columns and walls have top faces in the physics model delete them.

 

If all else fails you can always share your physics model here for use to look at ? 

 

Edited October 28, 2022 by Aquila Kytori

[Chic Aeon]

Thanks for all of that :D.  Here is a sceenshot from above. Honestly I am OK with the 8 land impact as the physics are perfect and I am still using analyzed physics LOL.  I have tons of land impact, it is temporary and I am not selling it. It becomes 3 when turned phantom but I really like physics as perfect as I can get them.  The point of having correct physics on this was so that items don't get stuck in the bounding box. I can deal with that of course but it is just plain messy. 

 

My OMG moments that night were the physics cost CHANGING to 3 even when prim when they had been 8. But we know now that it was temporary and a relog puts it back to 8. So it really wasn't "changing" at all.  The build is basically 7 meters by 7 meters so far as size goes. 

I am SURE your way is better LOL.  I am just old and set in my ways and have never been good at Planes physics *wink*.   I think it is likely the angled corner upping the cost but I want that corner angled and I want the physics to work LOL.  So there is that. Stubborn gal. 

 

If I change the build to convex hull (and then would make it phantom of course) this is what I get. So our models appear to be different somehow too.

I do very much appreciate that time you took with this.  Always good to have your wisdom reminding us :D. 

EDIT: I thought that the 3 prim land impact shifting MIGHT be because of the OS-SL viewer so I did a test in the Linden viewer and indeed it switched to 3 even when changed back to PRIM from convex hull. Hard to see in the screenshot but that is a 3 showing as cost. It reverts back in the Linden viewer also --- so NOT a viewer issue it seems 

Edited October 28, 2022 by Chic Aeon
spelling and adding info

[Aquila Kytori]

4 hours ago, Chic Aeon said:

Honestly I am OK with the 8 land impact as the physics are perfect and I am still using analyzed physics LOL.

The reason I thought that you were using Planes type, not Analysed physics is because in your image we can see the triangles that make up the collision surfaces. I maybe wrong but before, for Analysed physics you couldn't see those when in Physics view?

Now, using Firestorm, there is very little difference to distinguish between the two types of Physics.

In the screenshot below I think the difference in blue colour is only due to the difference in Physics costs :

 

For analysed Physics the Physics cost is the number of Hulls multiplied by 0.36. So the physics cost should be known in advance. If it is different from what you expected then you have a reason to complain lol .

Recap for anyone new to SL mesh that may be reading this,    13 Hulls x 0.36 = 4.68  :

Edited October 28, 2022 by Aquila Kytori

[Chic Aeon]

22 minutes ago, Aquila Kytori said:

The reason I thought that you were using Planes type, not Analysed physics is because in your image we can see the triangles that make up the collision surfaces. I maybe wrong but before, for Analysed physics you couldn't see those when in Physics view?

Now, using Firestorm, there is very little difference to distinguish between the two types of Physics.

In the screenshot below I think the difference in blue colour is only due to the difference in Physics costs :

 

For analysed Physics the Physics cost is the number of Hulls multiplied by 0.36. So the physics cost should be known in advance. If it is different from what you expected then you have a reason to complain lol .

Recap for anyone new to SL mesh that may be reading this,    13 Hulls x 0.36 = 4.68  :

I am indeed using analyzed "cube" physics.  Here is a screenshot (remember I am in the dark ages with 2.78). 

While it doesn't look like it in the picture the columns are all "separate" before joining and are not touching.  So I have eleven "cubes" with the ones making the corner being parallelograms rather than cubes -- one vertices moved to make an angle.

 

I am ALSO using the OS version of FS although I UPLOADED WITH THE LINDEN VIEWER as FS doesn't "do" physics in the OS version.  I am going to  log in with Linden viewer again and take a screenshot.  I "may" care about this in the future LOL and SOMEONE may care about it now.

It seems to be doing exactly what it says it will do -- in the LINDEN viewer.

When I look at your very cool animated graphic I have LOTS more hulls than you. I have tried it the more streamlined way but I have never been happy with how the physics works. Too "mushy" for my tastes. Hence I am sticking with Droogle's (? think that is close) method from long, long ago LOL.   But I think this has been explained well now for folks that encounter the issue.  

As far as I know there will not be a way to get this to upload with correct physics in OS.  I'll give it a try. THERE the planes method may be the only way to go.  It's iffy. I have managed getting houses up OK but ya never know. 

 

Thanks again!!!!   Hope you got to drop by ARTSCAPE and see your 3D work in the big gallery. Always loved that.  

 

Edited October 28, 2022 by Chic Aeon
spelling and adding info

[Aquila Kytori]

Go back to the Firstorm for SL viewer and in the Physics panel of the uploader check the number of hulls after hitting the Analyze button.

In the Firstorm view you have that little slider which can be used to "explode" (separate) each physics hull as seen in the anim. gif I posted earlier.

Remember each colour in the physics preview indicates a separate hull. In your preview panel there are way to many different colours (Hulls) !

There is a reason for this and its the reason why your Physics cost is so high. 

You see in my anim.gif of a very similar model to yours that there are only 13 Hulls. Yours is showing 35 !

 

Also the LL viewer does not indicate the real Physics cost of the the model when it is set to Physics Shape type Prim. The Firstorm viewer does.

Calculation of Physics cost before upload was 13 x 0.36 = 4.68 and the Firstorm viewer indicates ............. 4.68

 

Note also I didn't take a chance on the SL Physics analyzer getting it right. Take a look in the preview window of the gif and see how each section of wall between the pillars are separate hulls. Separate hulls in Blender before exporting. (with a little space between each of them which helps the Analyzer to see that I wanted them to be counted as separate hulls).

The rule is don't give the Analyzer the option to create hulls where you don't want there to be any. Your mesh YOU decide.  lol

 

 

Edited October 28, 2022 by Aquila Kytori

[Aquila Kytori]

39 minutes ago, Chic Aeon said:

I have LOTS more hulls than you. I have tried it the more streamlined way but I have never been happy with how the physics works. Too "mushy" for my tastes.

if you are still refering to Analyzed physics ...........

take a look at the three pillars on the right compared to the three pillars on the left. Each pillar on the left is a single colour indicating each is a single hull. Now the three on the right, each pillar is shaded with at least three different colours ! Why would that be? They need to be checked in Blender.

So to start with, in Blender check your pillars and also make each section of wall between pillars clearly separate boxes.

Your physics model in the Physics panel is indicating 166 vertices mine 104. Why such a big difference ? Again needs to be looked at in blender.

 

Edited October 28, 2022 by Aquila Kytori

[Chic Aeon]

14 minutes ago, Aquila Kytori said:

Also the LL viewer does not indicate the real Physics cost of the the model when it is set to Physics Shape type Prim. The Firstorm viewer does.

Actually it IS there at the bottom.  And I don't have any of the features that are new as I have the OS version.  I have tried running them side by side and it hasn't worked well. And with the new TOS I may be doing a fair amount more over in OS LOL. Hence that is where I am at.  Since I rarely make anything with physics needed these days it isn't really an issue for me. 

 

So I understand what you are saying and maybe you were posting when I was posting - who knows. But at the moment it is a moot point for me.  IF I start making houses in SL again (doubtful) then I will definitely use the FS viewer. But again it doesn't seem likely. I am more into art these days and customer support which happily is pretty much non-existent as I always test test test before putting something out  :D.  And I see another comment from you so ......

[Aquila Kytori]

16 hours ago, Chic Aeon said:

Actually it IS there at the bottom.

ooops yes. I was looking only where I was expecting to see it, saw an empty space and decided it wasn't there .

And we can only give suggestions, sometimes even suggestions that were not asked for,   you are  not obliged to follow any of them lol.

 .............. And ty, I will check out Artscape this weekend. Its been a long time since I wandered around SL just for the fun of it. 

 

 

 

Edited October 29, 2022 by Aquila Kytori

[Chic Aeon]

OK. The model itself is prestine  but the pillars in the physics model were too close it seems. Moved them farther out and the physics weight went way down.  This is good. And you can breathe out now.  I am still very puzzled about the reverting part of the issue but guessing it had something to do with the physics model "touching".    

 

Quick edit:  I have three NEW copies of the display out now and two are 4 LI and one remains at 3 after being set to prim. They are identical. I am going to guess that the one that says 3 will be 4 when I log on. So this apparently had nothing to do with ME but is an odd bug :D.  

 

Edited October 28, 2022 by Chic Aeon
adding info

[ChinRey]

1 hour ago, Chic Aeon said:

I am going to guess that the one that says 3 will be 4 when I log on

That's right. 11 hulls adds up to 3.96 physics weight so 3 LI is that pesky LI display bug again.

Btw, it's too late for this build but it may be worth remembering for the future:

1 hour ago, Aquila Kytori said:

take a look at the three pillars on the right compared to the three pillars on the left. Each pillar on the left is a single colour indicating each is a single hull. Now the three on the right, each pillar is shaded with at least three different colours ! Why would that be? They need to be checked in Blender.

This problem can usually be fixed in the uploader. Choose Retain % in the simply method menu and try values between ... say 80-90%. You have to know what to aim for to find the right percentage of course but if you do, it's quicker and easier than sorting it out in Blender.

[Chic Aeon]

Just an addendum to anyone who might care.  I uploaded this over on "the main" OS grid and it worked amazingly well. The physics weren't perfect but they were much better than I expected. This using the FS SL/OS viewer. It still got a bit confusing over on the right even with the cubes moved away some from the back wall on the physics.  

[Jenny Ashland]

A simple solution is to make a regular invisible prim and attach it to the base of the item, make sure it is the root prim. Then with everything linked set it to "none" then set it to prim. This makes anything that is not physical more or less phantom while keeping the part you have physics on solid. It should reduce your land impact count.