Mesh bug alert

[ChinRey]

I just noticed that one of the old bugs LL never is going to fix has gotten worse recently:

Look at the two pillars here:

 



The two are absolutely identical except for one thing: the one to the left was rotated ninety degrees to lay down horizontally before uploading and then back afterwards. The one to the right was uplaoded as it is.

What happens is that the dimension along the local z axis is ignored when the LOD switch points are determined so at the distance shown here, the pillar to the left has switched to low LOD model while the one to th left is still in high LOD.

This is a well known old bug for meshes with four or five faces but this one has only three faces and should not have been affected.

So a tip to evberybody: If you uplaod thin and tall meshes, remember to lay them flat first, then rotate them back in position after uploading.

[Drongle McMahon]

Just to check - it's not by any chance because there'a a "secret" face becuase of one having more than 21844 triangles, is it? Certainly looks unlikely from the pictures, but ....

[ChinRey]

No hidden faces and triangle count is 144.

Oh well, I've filed a JIRA about it. It's not going to do any good but at least you can't say I didn't try.

[Whirly Fizzle]

According to Drongle's old bug at VWR-27992, which was moved to an internal project, so we can't see it, this bug only affects mesh with 3 or 4 materials, not 4 or 5 materials."

"Thus it is the models with three or four materials that exhibit the unexpected behaviour while the otherts behave as expected."

VWR-27992 can still be viewed HERE.

Also more info on BUG-6287 - LOD Switch Distance bug? (also 3 materials).

ChinRey's new JIRA issue: BUG-11561 - Mesh uploads have incorrectd switch points

[Pamela Galli]

Thanks for the info, which I will not remember, but at least it is here in the forum and searchable. 

I would be in trouble if I did not have this forum and its generous helpers!

[Drongle McMahon]

Thanks Whirly. Either your memory or your filing system is better than mine - most likely both. If anyone wants the pictures that went along with the jira, they are in message 20 in this thread.

[ChinRey]

---

Whirly Fizzle wrote:

According to Drongle's old bug at VWR-27992, which was moved to an internal project, so we can't see it, this bug only affects mesh with 3 or 4 materials, not 4 or 5 materials."

---

Woops, I was sure it was four or five materials. So nothing new after all then.

Still it's just as well to post a reminder about it here every now and then. Lots of people aren't aware of it. I can't imagine LL will ever do anything about it either - my first JIRA about the problem has its second anniversary soon and Drongle's is more than four years old.

[mikka Luik]

Thanks for the reminder, had sort of forgotten this particular nasty. Wish we could 'sticky' this but will suffice to tape a note to keyboard i guess

[Vulpinus]

---

ChinRey wrote:

---

Still it's just as well to post a reminder about it here every now and then. Lots of people aren't aware of it.

---

Certainly is - thank you!

I keep a folder full of useful bits like this, with helpful file names, when I come across them. Occasionally I even remember to look in the folder :smileylol:

[Nacy Nightfire]

"What happens is that the dimension along the local z axis is ignored when the LOD switch points are determined so at the distance shown here, the pillar to the left has switched to low LOD model while the one to th left is still in high LOD"

Should one of those read "RIGHT"?  This bit is a little confusing even if the image clarifies your meaning. Maybe I'm missing something, though and it's not a typo.

[Vulpinus]

yeah, the pole on the left is the right one, and the pole on the right should be left out because it's broken. :matte-motes-big-grin-wink: