-
Notifications
You must be signed in to change notification settings - Fork 3.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
tileset.modelMatrix
not working for .i3dm
with ModelExperimental
enabled
#10255
Comments
Not sure if it's the same problem, but I noticed in this sandcastle that the i3dm models seem to be oriented incorrectly: I need to check if the bridge is an i3dm, but it also seems to be oriented incorrectly: |
Some more details:
|
Not quite... it doesn't seem to make a difference for the cubes example. Sounds like this will require some more digging to better understand the matrices in the instancing code. |
These are the expected orders of transformations
(Computed tile transform includes the tileset's model matrix and the tile transforms) |
I'm starting to understand this in bits and pieces. I've been taking a closer look at the above equations and trying to translate them to the current code's variables. Let me try to organize my thoughts so far:
Let me re-write the desired transforms a little bit:
Introducing That said, there's some gotchas here:
Some other potential problems in the current code:
I have to think a bit about the cleanest way to fix these problems. |
Hm, looking at the old code for 3D Tiles 1.0 formats, it was a bit different:
Furthermore, I found at least one case where |
Update: I tried an experiment:
and that seems to fix things. What was happening is the glTF root node had a z-up-to-y-up transform that was getting ignored, so the axis correction was not being canceled out properly. This still doesn't address the |
Regarding the |
Found this while testing for #10250. With
ExperimentalFeatures.enableModelExperimental = true
, the instanced model doesn't show up at the correct position, instead remaining on the ground. The PR itself did not introduce it; I tested onmain
and the same behavior occurs.Here's a sandcastle to demonstrate.
The text was updated successfully, but these errors were encountered: