First Round of 3D Tiles 1.0 updates #301

ggetz · 2018-04-17T23:13:16Z

First round of edits for 3D Tiles 1.0, tried to minimize organizational and formatting changes to be tackled later.

Fixes #298
Fixes #295
Fixes #280
Fixes #272
Fixes #263
Fixes #225
Fixes #211
Fixes #184
Fixes #140
Fixes #94
Fixes #60
Fixes #8

Add link to OneSky blog post

pjcozzi · 2018-04-17T23:30:29Z

Please target a new 1.0 branch.

lilleyse · 2018-04-24T18:43:07Z

For #8, add extras and extensions to boundingVolume.schema.json, batchTable.schema.json/hierarchy/items, and featureTable.schema (even though b3dm, i3dm, and pnts derive from this and include these properties themselves, it would be good to be explicit in the feature table schema).

Can you also open a PR making any spec changes into the vector tiles branch?

lilleyse · 2018-04-24T17:21:50Z

TileFormats/PointCloud/README.md

@@ -121,7 +121,9 @@ Per-point normals are an optional property that can help improve the visual qual
 The normals will be transformed using the inverse transpose of the tileset transform.

 #### Oct-encoded Normal Vectors
-Oct-encoding is described in [*A Survey of Efficient Representations of Independent Unit Vectors* by Cigolle et al.](http://jcgt.org/published/0003/02/01/). An implementation for encoding and decoding these unit vectors can be found in Cesium's
+Oct-encoding is described in [*A Survey of Efficient Representations of Independent Unit Vectors* by Cigolle et al.](http://jcgt.org/published/0003/02/01/). Oct-encoded values are stored in unsigned, unnormalized range ([0, 65535] or [0, 255]) and then converted to a signed normalized range ([-1.0, 1.0]) at runtime.
+


A few tweaks for this line

pnts only uses oct16 encoding so only the [0, 255] range needs to be mentioned

Minor wording

Oct-encoded values are stored in unsigned, unnormalized range [0, 255] and then mapped to a signed normalized range [-1.0, 1.0] at runtime.

A similar explanation needs to be added i3dm in the Oct-encoded Normal Vectors section but using the range [0, 65535].

lilleyse · 2018-04-24T17:27:53Z

TileFormats/Batched3DModel/README.md

@@ -64,18 +64,19 @@ These semantics define global properties for all features.
 | Semantic | Data Type | Description | Required |
 | --- | --- | --- | --- |
 | `BATCH_LENGTH` | `uint32` | The number of distinguishable models, also called features, in the batch. If the Binary glTF does not have a `batchId` attribute, this field _must_ be `0`. | :white_check_mark: Yes. |
+| `RTC_CENTER` | `float32[3]` | A 3-component array of numbers defining the center position when instance positions are defined relative-to-center. | :red_circle: No. |


Drop the word instance since it is specific to i3dm.

lilleyse · 2018-04-24T17:47:23Z

TileFormats/Instanced3DModel/README.md


-`.i3dm`
+Instanced 3D modela tiles use the `.i3dm` extension and `application/octet-stream` MIME type.


Instanced 3D modela -> Instanced 3D Model

lilleyse · 2018-04-24T18:04:54Z

schema/extension.schema.json

+    "type": "object",
+    "description": "Dictionary object with extension-specific objects.",
+    "properties": {
+    },


Change to "properties": {},

lilleyse · 2018-04-24T18:44:44Z

README.md

-`b3dm` and `i3dm` tiles embed glTF. According to the [glTF spec](https://github.com/KhronosGroup/glTF/tree/master/specification/1.0#coordinate-system-and-units) glTF uses a right-handed coordinate system and defines the y axis as up. By default embedded models are considered to be y-up, but in order to support a variety of source data, including models defined directly in WGS84 coordinates, embedded glTF models may be defined as x-up, y-up, or z-up with the `asset.gltfUpAxis` property of `tileset.json`. In general an implementation should transform glTF assets to z-up at runtime to be consistent with the z-up coordinate system of the bounding volume hierarchy.
+#### Tile content
+
+`b3dm` and `i3dm` tiles embed glTF. According to the [glTF spec](https://github.com/KhronosGroup/glTF/tree/master/specification/1.0#coordinate-system-and-units) glTF uses a right-handed coordinate system and defines the y axis as up. By default embedded models are considered to be y-up, but in order to support a variety of source data, including models defined directly in WGS84 coordinates, embedded glTF models may be defined as x-up, y-up, or z-up with the `asset.gltfUpAxis` property of the tileset JSON. 


Use a 2.0 link for https://github.com/KhronosGroup/glTF/tree/master/specification/1.0#coordinate-system-and-units

lilleyse · 2018-04-24T20:07:22Z

Styling/README.md

-## MIME Type
-
-_TBA, [#60](https://github.com/AnalyticalGraphicsInc/3d-tiles/issues/60)_
+> Implementation Note: Point cloud styling engines may often use a shader (GLSL) implementation, however some features of the expression language are not possible in pure a GLSL implementation. Features that must account for these inconsistencies include evaluation of `isNan` and `isFinite` (GLSL 2.0+ supports `isnan` and `isinf` for these functions respectively); the types `null` and `undefined`; strings, including accessing object properties (`color()['r']`) and batch table values; regular expressions; arrays of lengths other than 2, 3, or 4; mismatched type comparisons (`float` to `bool`); and handling of "index out of bounds" errors.


Is it possible to have a bulleted list within a quoted section? That might be easier to read.

lilleyse · 2018-04-24T20:07:49Z

Styling/README.md

-`application/json`
+## File Extensions and MIME Types
+
+* Tileset styles use the `.json` extension and the `application/json` mime type.


Same note as before where the extension is not required.

lilleyse · 2018-04-24T20:14:03Z

TileFormats/Batched3DModel/README.md

@@ -119,16 +120,20 @@ The vertex shader can be modified at runtime to use `a_batchId` to access indivi

 When a Batch Table is present or the `BATCH_LENGTH` property is greater than `0`, the `batchId` attribute (with the parameter semantic `_BATCHID`) is required; otherwise, it is not.


The gltf snipped above is still based on 1.0 techniques/shaders. For 2.0 all that's required is a primitive contains a _BATCHID attribute.

lilleyse · 2018-04-24T20:15:26Z

TileFormats/Batched3DModel/README.md


-The file extension is optional. Valid implementations ignore it and identify a content's format by the `magic` field in its header.
+> Implementation Note: Clients may also use the glTF [CESIUM_RTC](https://github.com/KhronosGroup/glTF/tree/master/extensions/1.0/Vendor/CESIUM_RTC) extension and define points relative to center.


We are phasing away from this extension, so I wouldn't include this note.

lilleyse · 2018-04-24T20:23:16Z

TileFormats/Batched3DModel/README.md


-`.b3dm`
+Vertex positions may be defined relative-to-center for high-precision rendering [1]. If defined, `RTC_CENTER` specifies the center position and all vertex positions are treated as relative to this value.


Here it should be noted that RTC_CENTER is z-up whereas vertex positions are y-up.

In general I think our coordinate system messaging is still not quite there... still need to think about the clearest way possible to describe it.

lilleyse · 2018-04-24T20:28:38Z

~~Is #225 (padding) in here?~~ moved to #301 (comment)

lilleyse · 2018-04-24T20:34:30Z

I made some edits as I went along in 192a632. More to come later.

lilleyse · 2018-04-24T20:38:32Z

#295 and #280 will need a bit of work before they are done, maybe we can discuss offline
#184 - might be good to explicitly mention that the top level json is not required to have a specific name
#225 - doesn't seem to be included here. It would be good if each tile format has a easy to find link to the padding requirements

I think the rest of the issues look good or are covered by comments.

Copyedit

ggetz · 2018-05-09T15:36:37Z

Added glTF CESIUM_z_up extension: KhronosGroup/glTF#1338

ggetz · 2018-05-09T16:50:09Z

Updated figures.

…ely solid yet

Upgrade tilesets to version 1.0

lilleyse · 2018-05-09T20:50:41Z

Awesome work! Ready to merge to 1.0.

lilleyse and others added 4 commits June 20, 2017 17:14

0.0 to 1.0

0e22b00

Add link to OneSky blog post

7cc26a8

Merge pull request #300 from AnalyticalGraphicsInc/pjcozzi-patch-1

5a62d6f

Add link to OneSky blog post

3D Tiles 1.0 updates

59247c4

ggetz requested a review from lilleyse April 17, 2018 23:13

ggetz changed the base branch from master to 1.0 April 18, 2018 13:32

slchow and others added 7 commits April 21, 2018 09:00

copyedit 1

101e8a5

copyedit 2

12d2583

figures

7c9d494

copyedit

90b53dd

RegExp

86d05ae

Response to Gabby's comments

d68fb03

regExp()

77c87f8

lilleyse mentioned this pull request Apr 24, 2018

Add RTC_CENTER support for b3dm CesiumGS/cesium#6488

Closed

lilleyse reviewed Apr 24, 2018

View reviewed changes

Tweaks

192a632

slchow and others added 6 commits April 24, 2018 13:59

Merge branch 'master' into copyedit

9e7dfc7

Sean corrections

5586e25

capitalization

5117393

Other edits

7169305

Merge pull request #302 from slchow/copyedit

ba91831

Copyedit

Merge branch 'master' into tileset.json-edits

899b3a3

ggetz mentioned this pull request Apr 26, 2018

Drop support for loading a tileset based on directory alone CesiumGS/cesium#6502

Closed

ggetz added 2 commits April 26, 2018 12:26

Spec clarification edits

d3fce88

Address CRS, padding, boundingVolumes, geometric error

026d7dd

Updated figures

c954b2b

lilleyse and others added 3 commits May 9, 2018 14:29

Combine crs paragraphs, use some of the original wording, not complet…

36dde46

…ely solid yet

Formatting fix

5ba688c

Tweaked CRS

384e749

ggetz force-pushed the tileset.json-edits branch from 5268eb6 to 384e749 Compare May 9, 2018 19:13

ggetz and others added 4 commits May 9, 2018 15:53

Merge branch 'tileset.json-edits' into version-upgrade

98854ad

Merge pull request #234 from AnalyticalGraphicsInc/version-upgrade

f5bffdd

Upgrade tilesets to version 1.0

Add acknowledgements

4bc6f90

Edits

6dce1a4

lilleyse merged commit 5208a72 into 1.0 May 9, 2018

lilleyse deleted the tileset.json-edits branch May 9, 2018 20:50

lilleyse mentioned this pull request May 11, 2018

Add TOC to all Tile Format specs #229

Closed

ggetz mentioned this pull request Jun 27, 2018

Tileset content.url to content.uri CesiumGS/cesium#6744

Merged

2 tasks

ggetz mentioned this pull request Jul 6, 2018

Move batch table hierarchy to extension CesiumGS/cesium#6780

Merged

2 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

First Round of 3D Tiles 1.0 updates #301

First Round of 3D Tiles 1.0 updates #301

ggetz commented Apr 17, 2018

pjcozzi commented Apr 17, 2018

lilleyse commented Apr 24, 2018

lilleyse Apr 24, 2018

lilleyse Apr 24, 2018

lilleyse Apr 24, 2018

lilleyse Apr 24, 2018

lilleyse Apr 24, 2018

lilleyse Apr 24, 2018

lilleyse Apr 24, 2018

lilleyse Apr 24, 2018

lilleyse Apr 24, 2018

lilleyse Apr 24, 2018

lilleyse Apr 24, 2018

lilleyse commented Apr 24, 2018 •

edited

Loading

lilleyse commented Apr 24, 2018

lilleyse commented Apr 24, 2018

ggetz commented May 9, 2018

ggetz commented May 9, 2018

lilleyse commented May 9, 2018


		`.i3dm`
		Instanced 3D modela tiles use the `.i3dm` extension and `application/octet-stream` MIME type.

		@@ -119,16 +120,20 @@ The vertex shader can be modified at runtime to use `a_batchId` to access indivi

		When a Batch Table is present or the `BATCH_LENGTH` property is greater than `0`, the `batchId` attribute (with the parameter semantic `_BATCHID`) is required; otherwise, it is not.


		The file extension is optional. Valid implementations ignore it and identify a content's format by the `magic` field in its header.
		> Implementation Note: Clients may also use the glTF [CESIUM_RTC](https://github.com/KhronosGroup/glTF/tree/master/extensions/1.0/Vendor/CESIUM_RTC) extension and define points relative to center.


		`.b3dm`
		Vertex positions may be defined relative-to-center for high-precision rendering [1]. If defined, `RTC_CENTER` specifies the center position and all vertex positions are treated as relative to this value.

First Round of 3D Tiles 1.0 updates #301

First Round of 3D Tiles 1.0 updates #301

Conversation

ggetz commented Apr 17, 2018

pjcozzi commented Apr 17, 2018

lilleyse commented Apr 24, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

lilleyse commented Apr 24, 2018 • edited Loading

lilleyse commented Apr 24, 2018

lilleyse commented Apr 24, 2018

ggetz commented May 9, 2018

ggetz commented May 9, 2018

lilleyse commented May 9, 2018

lilleyse commented Apr 24, 2018 •

edited

Loading