Jump to content

As usual Jira is not perfoming its obligations - Blender Precision difficulties emulating a Prim Slice with a mesh object.


VirtualKitten
 Share

You are about to reply to a thread that has been inactive for 1719 days.

Please take a moment to consider if this thread is worth bumping.

Recommended Posts

Blender Precision difficulties emulating a Prim Slice with a mesh object.

https://jira.secondlife.com/browse/BUG-227312

What just happened?

I contacted the Blender Mesh Development group on SL who could not provide and answer to me currently only stating this was not a blender precision problem. I created a mesh tube 0.09194 long. I was told that if I created another object as far out as the length of the item I created it would move the axis as a splice does in second life with a prim. I had some difficulties in blender getting a box selection of dimensions which I felt was a the most basic requirement in Blender. However I could get this in object mode . I duplicated object so i could get one in object mode without the tiny item at the long end. The measurement .09194 on the z axis. as shown below:

108891_f3c8d0882f97df7f3c6ed54e89e67c5d.png

1) I created another mall object precisely at .018388 giving my bounding box as this length this can be seen in image below 108890_c06ded4365a787b2ed5ade7d47665681.png

2) I imported this mesh via collada dae file into SL and rezzed it in world. I duplicated it so i had two next to each other 108889_e013cf91437eba23cda43c0e4be0d99b.png
**

3,)And then  increased Z on one of them to 50m Both ends of the item grew 108888_6351003e699844aa6cc0fefe79f169b7.png

What were you doing when it happened?

I was enlarging a Mesh item imported into SL expecting it to behave like a sliced prim cut halfway through . I was told by increasing bounding box this would emulate a spliced prim halfway through.

What were you expecting to happen instead?

I was expecting the end as shown in following picture marked by the green arrow to remain in position. instead it extends on both ends . This would appear to be an error of precision but my figures have been checked . Can some one explain why in mesh you cannot replicate a spliced prim? 108888_6351003e699844aa6cc0fefe79f169b7.png  

Other information

I am hopeful for some intelligent replies not conjecture or opinion. It should be noted that it is very easy to slice a prim in half to do this with accurate precision result. How is it then that the mesh produced from Blender cannot match the resolutions of prim in SL?

 *UPDATE THREAD LOCKED

As normal Jira has closed this without a referral to LindenLabs.  I have to wonder what they actually do as none of my issues have ever been resolved by them. Its just typical of this service close the issue and hope you go away. I have already explained that the Blender Development Group was unable to resolve this why then has Kyle Linen done his useful dutiful obligation nothing !

Is it a wonder people are now spamming this group as it appears they are not willing to provide any reasonable required response. Kyle Linden can see from my objects that they are identical in size to those in blender thus this must be something that Second Life is doing different to mesh objects as oppose to prims . I reported this as a bug as it is a bug that you will not mend .Why is that exactly? As you have done nothing in seven days tin law this stands as a legal position without answer.

Thank you for reading this

 

Denise.

Note above: Can anyone help please as Jira seems to do nothing but record information now .

 

Edited by VirtualKitten
To add tags
Link to comment
Share on other sites

37 minutes ago, VirtualKitten said:

I have to wonder what they actually do as none of my issues have ever been resolved by them.  Its just typical of this service close the issue and hope you go away. I have already explained that the Blender Development Group was unable to resolve this why then has Kyle Linen done his useful dutiful obligation nothing !

It seems like you've really misunderstood the purpose of the JIRA.

It doesn't exist for you to post problems you are having with something (especially things not directly related to Second Life).

It exists for you to report problems about Second Life itself, as well as making feature-requests.

This problem involves another program (Blender) so it's off-topic for the JIRA -- and it's going to be closed because it's frankly not Linden Lab's problem.

 

That said -- I just created a mesh tube that is 0.09194 long and so that the bounding box was equal on both sides. It scales exactly as a path-cut prim would, even when resized in-world to a big scale. There's clearly something you're doing wrong and you're acting very immature all things considered.

Edited by Wulfie Reanimator
  • Like 5
Link to comment
Share on other sites

Wulfie , Thank you for you're response. No how can i understand their purpose :). They would rather answer a post why cant I TP why is it i cant teleport than solve and advanced question. Never have resolved any difficult question for me, have a look and see.

They did however give me a steer back here. so there is some points i guess

Well i am a little cat what do you expect. Please view my jira you will find nothing wrong, lots of awkward questions for jira never answered they just close them all! Well it is jira's and Secondlife problems i followed examples on youtube and in Blender Development group with precise measurements as you will see. I am nothing but thorough. I am not alone i think some people are putting spam on it as they are fed up with no proper answers. Not that i agree with that.

Thanks for looking
Denise

Link to comment
Share on other sites

13 minutes ago, VirtualKitten said:

Wulfie , Thank you for you're response. No how can i understand their purpose :). They would rather answer a post why cant I TP why is it i cant teleport than solve and advanced question. Never have resolved any difficult question for me, have a look and see.

They did however give me a steer back here. so there is some points i guess

Well i am a little cat what do you expect. Please view my jira you will find nothing wrong, lots of awkward questions for jira never answered they just close them all! Well it is jira's and Secondlife problems i followed examples on youtube and in Blender Development group with precise measurements as you will see. I am nothing but thorough. I am not alone i think some people are putting spam on it as they are fed up with no proper answers. Not that i agree with that.

Thanks for looking
Denise

Oh... Now I remember you. My previous response still applies:

Please look at my previous post (in this thread) for a link to the .blend file that shows a properly working model. Good luck.

  • Like 2
Link to comment
Share on other sites

You are so wrong wolfie. Let me explain, I have not had one proper answer, just excuses for poor implementation with points. I have had to write around because matters don't get addressed to my satisfaction. It is not an answer for an admin to lock a thread. It should be up to the thread owner . How many times do I have to scream this before its noted. ?

Anyway we are off topic my immediate concern is this non slice in mesh which does not work at 50m

Kind Regards

D

Link to comment
Share on other sites

I remember at one of the old content creator user group meetings somebody asked Nyx Linden why they didn't accept JIRA reports for Firestorm bugs and Lynx' rather dry answer was "Isn't it better to send bug reports to people who can actually do something about them?" ;)

JIRA is a system for reporting bugs in the software and suggesting new features, nothing mroe and ntohing less. This clearly is neither so it didn't belong there. Besides, JIRA is run by the SL developer team and they don't have any epxertise in content creation so they wouldn't have been able to help you.

I think I know what the problem is, though. Look at this:

image.png.6f89a50b346a3b345d1d217fedf4c281.png

If you work in Object mode (which you shouldn't anyway) always apply Location, Rotation and Scale whenever you move or resize something. Location and Rotation values must always be 0.00000 and Scale 1.00000. So hit Ctrl+A and apply location, rotation and scale every time you add or move something in Object mode.

If you need precision measurements like you do here, at least check them in Edit mode and remember it's the outermost vertices that count for the verall isze, not the outermost objects.

(Edit: Strictly speaking , there are at least three different explanations to the problem you have but the method I described, Apply LRS and then check vertice coordinates in Edit mode will fix them all.)

Edited by ChinRey
  • Like 3
Link to comment
Share on other sites

Hmm , Chin Rey

Thank you for your helpful comment at least its something to try:) .

However the objects as you saw from the images on JIRA come in exactly the same size they are in blender . So this transformation you speak of I would of thought should have been visual only otherwise on import the imported mesh would be a completely different size . I am not sure how a overall scale can change the dimensions of a half way intersection as half way of 50%  or 75% is still halfway!

NOTE : It is the resize i of the imported MESH in Second life which is causing the mesh to act like a non spliced prim and grow on each end which is the problem. 6351003e699844aa6cc0fefe79f169b7.png.3e859dd5fa6734df755b541d39c43cfe.pnge013cf91437eba23cda43c0e4be0d99b.png.b2b03eea53b6cd684c739ec374001fdd.png0cad3073487d8b842803d37de0fd8885.png.fbd86e5771c07c4e0a2e9de2be9ec320.pngc06ded4365a787b2ed5ade7d47665681.thumb.png.96e775b29a2495ffa65e22a261c27bf7.png

[I have reattached them ]I would also point out the lack of access to a beta grid is meaning this is costing a lot of L$ to keep importing.

"JIRA is a system for reporting bugs in the software and suggesting new features " from my experience they don't do this either i wanted to be able to read meta data from an mp3 stream which was quite possible with the secondlife command set. They never answered it properly i.e https://jira.secondlife.com/browse/BUG-203003

 

f3c8d0882f97df7f3c6ed54e89e67c5d.png

Edited by VirtualKitten
Link to comment
Share on other sites

1 hour ago, VirtualKitten said:

Thank you for your helpful comment at least its something to try:) .

In case I didn't make myself clear enough, Applying location, rotation and scale isn't likely solve the problem in itself. But it's an essential step to clarify exactly where the problem is. The alternative is to do some seriously heavy math (including some rather complex triangulation since you have unapplied rotation) to calculate the exact offset for the "balancing vertices".

 

1 hour ago, VirtualKitten said:

I am not sure how a overall scale can change the dimensions of a half way intersection as half way of 50%  or 75% is still halfway!

What happens is that the center point of the mesh' bounding box isn't exactly at the end of your beam. You're scaling the object up 5,000 times here. That means a deviation of a tenth of a millimeter becomes half a meter. Regardless of what the people in the Blender Mesh Development Group says, Blender is not that precise. Even at a 100x resizing you have to expect deviations caused by rounding errors. The fact that you have rotation of the object too makes matters worse too - much worse. Of course, it doesn't help that the decimals have to be converted from 64 bit to 32 bit on upload either.

I can't see why you want to upload the mesh at such a small scale btw. Unless you are doing womething really weird, it won't affect neither the uplaod cost or the inworld performance/LI in any way.

 

1 hour ago, VirtualKitten said:

[I have reattached them ]I would also point out the lack of access to a beta grid is meaning this is costing a lot of L$ to keep importing.

Beta grid access is a case for a support ticket. An alternative to the beta grid is to check the meshes on the OS Grid or TAG or some other independent grid. You can't test land impact or physics models there but everything else should be the same as in SL.

 

2 hours ago, VirtualKitten said:

"JIRA is a system for reporting bugs in the software and suggesting new features " from my experience they don't do this either i wanted to be able to read meta data from an mp3 stream which was quite possible with the secondlife command set. They never answered it properly

My experience is that it takes somewhere between two and three years from when they reject one of my JIRAs to when they implement it anyway because they have to. ;)

Link to comment
Share on other sites

Hi Chin Rey,

I supplied images with dimensions showing that my items in mesh was taken at exact size into SL

In regard to this point of complex math I was told by the Blender  Mesh Development it was the same as a door privet and as I was only lengthening a bounding box was the way to go in Z direction . I questioned them on this as I could not find a way of getting dimensions of an object in edit mode there seemed to be a difference of opinion , however it was clear that the only way do do this was to copy the object and delete the little box from the end of it giving the length of the rod in object mode: 09194  . Now I was able to measure the length of my bounding box in object mode at  .18388   and with a little tweaking in edit mode  along the z axis using a constraint this little box could be moved out until it was to double the length of the rod measured in object mode . This was proven by the import as seen on the later screen which showed the exact same figure highlighted on size Z.  .18388  .

image.png.5146c12e57b7e3e3c9f0fdfdcf0d7ad4.png

I cannot see why this then was not accurate.

With regard to scale and rotation . Surely if this was a factor on the export to collada and import into SL would the size and rotations be different in SL than those in SL achieved by this import. I have show the direction with arrows  of this object in SL it is the same and as for the scale it is the same size as that in Blender?

I am really unsure of your points here could you explain by a demonstration what you mean please . I am especially unsure how this as my mesh is ending up in SL the size to start with i expect it to ?

The error creeps in when i resize to 50m on the z axis in SL . The center set by this process is at 5 significant digits therefore any calculation should have a certainty of +- 0.00001. How then is this 50m translated to an error of .01 off axis over 50m?

789448698_testtoseeerror.thumb.jpg.fe4ba4146cfc695d8d9d216993914472.jpg
 

I understand

measurement = (best estimate ± uncertainty) units  how is this then resolving an error of .01 over 50m. It should follow that the error on 1m should be .01/50 = 0.0002 . I am confused that this makes no sense as my object is so small here should be no error  unless SL is creating it.

Kind Regards

Denise

 

 

 

 

Edited by VirtualKitten
Adding images
Link to comment
Share on other sites

I think you are over-complicating things to be honest.

The practical solution is scale the object to the size you actually want it to be, apply scale and rotation, check in edit mode that the signifcant vertices are in the right position. That will fix your problem and there's no real need to spend more time and effort on it than that.

There are some interesting theoretical aspects of this though. I'll try to explain it as simple as possible.

First you have to keep in mind that computers never ever work with decimals the way we do. It's all in base 2 and the decimals will never match up exactly with the base 10 us humans prefer. Normally the difference is insignificant but they can accumulate and the way you are doing this, you are stacking a lot of them on top of each other:

  • Mesh "scale" and "size" are two completely different values. The actual size of the object is determined by multiplying the two. That's two minor deviations multiplied. (Incidentally, all objects in SL are nominally 1x1x1 m in "size". The actual sizing is all done with the "scale" values. It still has to be calculated at upload from both the size and scale values in the dae file though.)
  • Rotation adds another complication. Actual rotation values the way a computer handle them are also binary and the deviation between them and human-friendly degrees is even larger than for sizes.
  • Rotation also means the software has to do triangulation to determine the exact location of the vertices. That's another can of worms both for humans and computers - and a big one too.
  • The dae format we use to transfer meshes to SL is not a binary file format but a text document with the coordinates represented by base 10 numbers with seven decimals. So the binary coordinates in the Beldner document has to be converted to "human" decimals with fsirly limited precision when exported and then back to binary when imported.
  • SL does not actually support the kind of precision scaling Firestorm's abundance of decimals can lead you to believe. I'm not sure but I think it uses 8 bit values for scale so overall size of an object is in steps of 1/256 m.

This isn't a complete list but I think you get the idea.

If I understand you right, the deviation you end up with at 50 m size is 0.01 m. That means the deviation at the uploaded size is aproximately 0.000036776 m and with all those factors in mind, it's well within the margin of error you have to expect. I'm actually surprised it isn't much more.

Edited by ChinRey
  • Like 1
Link to comment
Share on other sites

4 hours ago, ChinRey said:

I think you are over-complicating things to be honest.

If I understand you right, the deviation you end up with at 50 m size is 0.01 m. That means the deviation at the uploaded size is approximately 0.000036776 m and with all those factors in mind, it's well within the margin of error you have to expect. I'm actually surprised it isn't much more.

I think you are now beginning to understand the nature here. However I am really interested in your scaling information and would like to hear more  , However now we are focused on this deviation occurring in my mesh at approx   0.000036776 m  can you please tell me if it is possible to set a mesh within this precision of nine significant digits in Blender?

In an attempt to see if this is at all possible I have a prim version of this mesh i am creating intend to try to use Prim2Mesh to see if it can convert a sliced prim as part of the project. I will also try to make a SL video of what happens to this in a given project so you can see the outcome.

For purposes of transparency i am sharing this Mesh2Prim Mesh Manager dae : http://moe.naonao.biz/mesh/MG_9R2A0LP3IJPV5TH3.html

based on a 50% sliced prim 1m in length. with the cursor at one end  . It should be noted that if you import this item it will be evident its has not reproduced the prim as it was the slice is evident but the actual cursor position is now centered and the bounding box was not created. clearly this is a problem encountered by this product in SL too.

How is it then on the face of it it seems that neither blender  2.10.12 or Mesh Generators ,Prim2Mesh a leading developer ShopNN

can create an object any user can create with a PRIM due to the required mesh setting at 0.000036776 m to accommodate this function working the same as a PRIM sliced in Secondlife?

Does this not seem something that Second Life Developers should be looking into? So then why did Jira just close it?

 

Kind Regards Denise

 

Edited by VirtualKitten
Added link
Link to comment
Share on other sites

3 hours ago, VirtualKitten said:

However now we are focused on this deviation occurring in my mesh at approx   0.000036776 m  can you please tell me if it is possible to set a mesh within this precision of nine significant digits in Blender?

I'm not sure. You definitely can't do it with rotation - that can often frustrating when you wotk in Blender. Once you have applied the rotation and scale you can fine tune the positions for each and every but I don't know if you can get precision up to nine significant digits. What I know, is that vertices that are patently at 0.00000000000000000000 often show up as 0.000001 etc.

 

3 hours ago, VirtualKitten said:

...

but the actual cursor position is now centered and the bounding box was not created.

Yes. There is no way to preserve the bounding box through a prim to mesh conversion.

  • Like 1
Link to comment
Share on other sites

22 hours ago, ChinRey said:

I'm not sure. You definitely can't do it with rotation - that can often frustrating when you wotk in Blender. Once you have applied the rotation and scale you can fine tune the positions for each and every but I don't know if you can get precision up to nine significant digits. What I know, is that vertices that are patently at 0.00000000000000000000 often show up as 0.000001 etc.

 

Yes. There is no way to preserve the bounding box through a prim to mesh conversion.

This all said. I am pretty sure that no mesh can be created to accurately work with the precision required at 50m why then did Second Life say:

Kyle Linden added a comment - 10/Jul/19 11:28 AM

Hi VirtualKitten,

Jira is not for 1:1 personal technical support and is therefore not able to provide one to one support in a timely manner. What you are describing and have depicted appears to be an issue with where you created the origin of the object in Blender. While you are welcome to use Blender to create content for Second Life, we cannot directly assist you with its use.

There are many different ways to seek help from other content creators and there is even a public user group you may want to attend for additional help or questions.

http://wiki.secondlife.com/wiki/Content_Creation_User_Group

For general assistance from other Residents see SL Answers at https://community.secondlife.com/forums/forum/109-answers/ and click Ask a Question.

Thanks!

This reply Kyle Linden seems pretty procedural ambiguity and a total disregard for the problem. Why will there development team not examine things properly or do they just not care !  Jira has not been attending to anything other than why can i not teleport here for ages can we not get back to the real issues and start solving some of them . If blender or no other program can operate at this precision except  Secondlife . Then there claim that MESH is capable of replacing PRIMS is complete rubbish ! Is it not about time they start paying more attention and resolve some of these issues please?

Thakyou you for joining us Kyra Abbatoir and Love
Zhaoying unfortunately no its not a Meme! If you have something helpful to say please join us.

  • Confused 2
Link to comment
Share on other sites

Well, LL’s message to you is pretty clear. They are not going to help you with “content creation” issues.  

Thanks for you reply . But if you read it properly you would see it is just a steer unfortunately no one seems to have the answer here either either! .

Perhaps Kyle Linden should not go off so fast and read the cases before passing them on? Its very disrespectful to the people that work hard on bringing matters to Jira's attention. This is not a content creation issue its a fault with mesh in Second life which is not being addressed and precision  capabilities at 50m objects  the precision 0.000036776  cannot be matched by mesh in Second life as compared to a sliced Prim in world . Why is Second-life not addressing this type of fundamental problem in the way that it operates mesh boundary boxes ?
 

Second life Jira seems seems to be acting akin to a an ostrich i.e burring its head in the sand!

Kind Regards

Denise

 

Edited by VirtualKitten
Link to comment
Share on other sites

1 hour ago, VirtualKitten said:

This is not a content creation issue its a fault with mesh in Second life which is not being addressed and precision  capabilities at 50m objects  the precision 0.000036776  cannot be matched by mesh in Second life as compared to a sliced Prim in world .

It's not the same thing. A polylist mesh is rendered by the viewer from a fixed list of data. If you scale it up, any deviations will be multiplied and you have be prepared to adjust manually to compensate because the software doesn't have the info it needs to be able to do it for you. This is not a Second Life problem, it's a general and inevitable problem with polylist meshes.

A prim is a procedural object - it is generated by the software from a simple mathematic. That means the software doesn't only know where those vertices are supposed to be but also why. That means it can do the necessary compensation automatically when the object is scaled so you don't have to worry about it.

  • Like 1
Link to comment
Share on other sites

1 hour ago, ChinRey said:

It's not the same thing. A polylist mesh is rendered by the viewer from a fixed list of data. If you scale it up, any deviations will be multiplied and you have be prepared to adjust manually to compensate because the software doesn't have the info it needs to be able to do it for you. This is not a Second Life problem, it's a general and inevitable problem with polylist meshes.

A prim is a procedural object - it is generated by the software from a simple mathematic. That means the software doesn't only know where those vertices are supposed to be but also why. That means it can do the necessary compensation automatically when the object is scaled so you don't have to worry about it.

The sliced prim works fine Chin and does not show this error in transformation it is just mesh with this procedure to increase a bounding box which is causing this.

 There is no way round this and i am loosing revenue as cannot create content and being restricted in creating this content by these Second life failures .

The precision is not available in Blender by your own addition to cater for this high precision 0.000036776 required to be altered on the mesh to allow it to ever work. It is clear to me that this precision 0.000036776 is not with blender which without Second life providing a work around that there is no way of creating the same  as a single sliced PRIM in Secondlife from MESH this is really peculiar and not well thought out !

Item is now on a ticket [ #1388564 ] with Second life

 

Link to comment
Share on other sites

No offense Whirly I am fed up with Jira what the point they don't answer, so no loss if they do what you say >>

They have never answered one point properly since I started asking them to look at things what is the point of them being there  this is on a ticket with Secondlife now not the first time i have complained that Jira was doing this. They locked me out back then too as i was insistent that they answer.  Nothing changed. second life cleared the block on my account and reinstated it.

Don't you think its about time the service was reformed. I am quite sure that if you look at the reports they are providing of stating they have resolved maters it is complete fabrication and really all they did was to just go through closing them .

Quite a lot of us have had enough of it we work hard building in Secondlife only to hit brick walls with the same common event happening nothing is resolved . If you review the cases on Jira you will see it is full of silly matters now and spam. what do you expect when its not working Its just my opinion and frustration  for what it is, but i think its shared with others. There is no sly dig as you put it just the truth . I did not close this Jira bug report  Kyle Linden did. I have reported this to Second life in the past that it is not correct procedure to allow anyone in Jira.Secondlife  to close a case . Any case reported to Jira.Secondlife  should be closed by the owner when they have an answer! Otherwise this same non answering of points is going to keep happening  and everyone will eventually avoid it anyway.

Kind Regards

Denise

 

 

 

Matter is now on a ticket [ #1388564 ] with Second life

Edited by VirtualKitten
Link to comment
Share on other sites

1 hour ago, Whirly Fizzle said:

No offense VirtualKitten, but you need to stop editing your JIRA issue with sly digs at the Lindens who triaged it. This is not the first time you have done this either.
If you keep doing this, a Linden will just end up banning you from the JIRA.

I don't take kindly to implied harassment Whirly Fizzle

When i am trying to obtain a solution I have reported this separately to secondlife as abuse on same ticket

Link to comment
Share on other sites

You are about to reply to a thread that has been inactive for 1719 days.

Please take a moment to consider if this thread is worth bumping.

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
 Share

×
×
  • Create New...