2.0.0-dev.1.7 #59

lexaknyazev · 2017-12-15T12:49:34Z

Added NODE_SKINNED_MESH_WITHOUT_SKIN, MESH_PRIMITIVES_UNEQUAL_JOINTS_COUNT issues.
Removed Catmull-Rom as valid spline interpolation mode.
Fixed issue pointer corruption on empty accessors.
Improved drag-n-drop validator UX when assets contain many issues (Drag-and-drop validator hangs on some models #60, Validation report more than 1,000,000+ issues and my browser cannot open it. #65).
Fixed crash on malformed JSON (spectacularly screwed-up glTF JSON yields validator stack trace #66).
Fixed cubic splines unit length check (Incorrect accessor error for cubic spline rotation tangents #70).
Added support for KHR_materials_unlit extension.
Added MULTIPLE_EXTENSIONS warning.
Updated the list of reserved extension prefixes.
Added GLB_UNEXPECTED_BIN_CHUNK, BUFFER_NON_FIRST_GLB, and BUFFER_MISSING_GLB_DATA (GLB chunks order #67).
Added ANIMATION_SAMPLER_INPUT_ACCESSOR_TOO_FEW_ELEMENTS and ANIMATION_SAMPLER_OUTPUT_INTERPOLATION issues.

Spec says: ***When the node contains `skin`, all `mesh.primitives` must contain `JOINTS_0` and `WEIGHTS_0` attributes.*** That makes sense to me, but maybe it's this code that's right, in which case we should update the spec.

emackey · 2017-12-15T18:05:01Z

Is this ready? I see #58 is un-checked above, but it looks like related code is on this branch already.

ISSUES.md needs a refresh, other than that this all looks good.

lexaknyazev · 2017-12-15T18:07:04Z

@emackey #58 hasn't yet been implemented. I'm going to finish it over the weekend.

disables syntax highlighting on big reports.

lexaknyazev · 2018-02-28T16:59:50Z

Remaining skinning stuff (joint indices, normalized weights, and hierarchy limits) will go to the next release. This one will be published this week.

lexaknyazev · 2018-03-06T17:18:23Z

@emackey @donmccurdy
This is ready for review. Please check if language of CHANGELOG.md and ISSUES.md is clear enough.

donmccurdy · 2018-03-06T18:15:08Z

lib/src/base/node.dart

+            if (mesh.primitives
+                .any((primitive) => primitive.jointsCount != 0)) {
+              context.addIssue(LinkError.nodeSkinnedMeshWithoutSkin);
+            }


Is it valid to specify skin on a node with no mesh at all?

It's forbidden by JSON-Schema:
https://github.com/KhronosGroup/glTF/blob/7b2215ced6fe20a6155ca4d238ff313a7348b7e8/specification/2.0/schema/node.schema.json#L87

donmccurdy · 2018-03-06T18:19:44Z

lib/src/errors.dart

+      new SemanticError._(
+          'MESH_PRIMITIVES_UNEQUAL_JOINTS_COUNT',
+          (args) => "All primitives must contain the same number of 'JOINTS' "
+              "and 'WEIGHTS' attribute sets.");


I think I like the idea of error or warning here, but is this mentioned in the spec?

https://github.com/KhronosGroup/glTF/tree/master/specification/2.0#node

The spec requires each primitive have JOINTS_0 and WEIGHTS_0, but I see nothing to say that one primitive couldn't have JOINTS_1 in addition to JOINTS_0 when other primitives do not. The spec would lead me to think that is valid, while this error would lead me to think it is not.

I don't have a data-driven opinion on this. We haven't yet seen models with more than 4 influences per vertex.

What do you think?

About whether it should be allowed, or whether the validator should say something about it?

I haven't seen any examples either. I'd prefer that primitives be more consistent than not, but elsewhere we've asserted primitives could have different attribute sets without discussing skinning specifically...

The message is sort of saying one thing and testing another, but tentatively, I'd lean toward making the error code something more generic (MESH_PRIMITIVES_INVALID_JOINTS_COUNT) and then we can change either the behavior of the test or the message text depending on what we decide in the future.

I'd lean toward making the error code something more generic

Agree.

@zellski do you have preference on this?

Thanks for checking -- no, no strong preference. I feel mildly irked by the thought of completely heterogenous primitives in a mesh, but I have no rational basis for that irritation. When we implement animation at Facebook, I expect we'll treat primitives in a mesh as almost completely unrelated to one another, so no strong "day job" preference either.

Considering that skinned mesh transform should be ignored, all primitives of the skinned mesh must have joints. Whether they have to have the same number of influences per vertex is a different question.

I'll update the code and tests.

Turns out we have a separate issue for skinned primitives without joints:

glTF-Validator/lib/src/base/node.dart

Lines 198 to 201 in d376c03

if (mesh.primitives

.any((primitive) => primitive.jointsCount == 0)) {

context.addIssue(LinkError.nodeSkinWithNonSkinnedMesh);

}

So I propose to replace "must" with "should" in the description of MESH_PRIMITIVES_UNEQUAL_JOINTS_COUNT (while keeping error code) and to reduce its severity to Warning or Information.

Sounds good 👍

donmccurdy · 2018-03-06T18:21:10Z

lib/src/utils.dart

+  if (warnOnMultipleExtensions && extensionMaps.length > 1) {
+    context.addIssue(SemanticError.multipleExtensions,
+        args: [null, extensionMaps.keys]);
+  }


Not sure I understand this one ... why is having multiple extensions reported as an issue? Your test case makes sense — multiple material extensions may interact in undefined ways — but reporting an issue when KHR_materials_unlit is used with KHR_draco_mesh_compression makes less sense.

This warning is effective only when several extensions are defined for the same object.
For now, it's enabled only for material objects:

glTF-Validator/lib/src/base/material.dart

Lines 89 to 90 in d376c03

final extensions =

getExtensions(map, Material, context, warnOnMultipleExtensions: true);

Sounds good.

emackey · 2018-03-06T18:57:15Z

ISSUES.md and CHANGELOG.md look good to me.

Currently having trouble with pub get. Running dartsdk-2.0.0-dev.32.0-windows-x64 and I cleared my old pub cache. Running pub get prints only one message:

Resolving dependencies...

But this returns immediately, without any delay that would normally be associated with attempting to download a number of packages from a server. No further messages are printed, errors or otherwise.

I'm pretty sure this is a problem with my local machine or local network. I tried switching back to master and the old dartsdk, which used to work, and it's not getting any packages or errors now. I may need to try this from a different machine on another network.

lexaknyazev · 2018-03-06T19:27:17Z

@emackey
You could try pub get -v.

emackey · 2018-03-06T19:51:46Z

pub get -v

Thanks. Corporate cert went bad, fixed now.

emackey

Looks good to me. Also I tested this in VSCode with a mismatched Joint/Weight sample, seems to work fine.

emackey · 2018-03-07T19:04:11Z

This is published to npm now.

https://www.npmjs.com/package/gltf-validator

lexaknyazev and others added 3 commits December 15, 2017 14:22

Bump version

8a3a6c3

Tighten up skinning requirement on primitives

721299c

Spec says: ***When the node contains `skin`, all `mesh.primitives` must contain `JOINTS_0` and `WEIGHTS_0` attributes.*** That makes sense to me, but maybe it's this code that's right, in which case we should update the spec.

Add NODE_SKINNED_MESH_WITHOUT_SKIN

3023804

lexaknyazev added this to the 2.0 milestone Dec 15, 2017

lexaknyazev added 2 commits December 15, 2017 17:24

Style tweaks

9479ef1

Add MESH_PRIMITIVES_UNEQUAL_JOINTS_COUNT

895fcaa

Removed Catmull-Rom spline interpolation

cfcbf23

lexaknyazev force-pushed the 2.0.0-dev.1.7 branch from 99dae0f to cfcbf23 Compare January 2, 2018 21:31

lexaknyazev added 8 commits January 5, 2018 12:32

Fix pointer corruption on empty accessors

accd29c

Drag-n-drop frontend prints elapsed time to console,

1ec34af

disables syntax highlighting on big reports.

Update deps

569d5cc

Update Prism.js to 1.9.0

63e0384

Fix #66

e7312cd

Bump SDK version & regen ISSUES.md

185a444

Update versions & clean new lints

97d3e6d

Use remainder instead of modulo, fix #70

9fd834a

lexaknyazev force-pushed the 2.0.0-dev.1.7 branch from 7545802 to 9fd834a Compare February 24, 2018 00:07

lexaknyazev added 4 commits February 24, 2018 04:18

Fix cubic splines for normalized accessors

13296b4

Improve drag-n-drop validator UX

7fab59c

Add support for KHR_materials_unlit, add MULTIPLE_EXTENSIONS issue

c476b54

Tighten GLB validation with new issues

b29493f

lexaknyazev mentioned this pull request Feb 24, 2018

GLB chunks order #67

Closed

More cubic splines validation

a187e66

lexaknyazev force-pushed the 2.0.0-dev.1.7 branch from 52ca8e9 to a50b553 Compare February 27, 2018 21:56

Use proper types

f357392

lexaknyazev force-pushed the 2.0.0-dev.1.7 branch from a50b553 to f357392 Compare February 28, 2018 17:02

Tighten types

a8a5698

lexaknyazev added 2 commits March 6, 2018 19:11

Refactor cubic splines validation, add tests

217cf7c

Cleanup HTML, .gitignore, remove deprecated deps, update SDK

d376c03

lexaknyazev requested review from emackey and donmccurdy March 6, 2018 17:07

donmccurdy reviewed Mar 6, 2018

View reviewed changes

emackey approved these changes Mar 6, 2018

View reviewed changes

lexaknyazev added 4 commits March 7, 2018 19:27

Adjust MESH_PRIMITIVES_UNEQUAL_JOINTS_COUNT.

1c43a04

Minor tests & utils cleanup

3f2db63

Add Travis CI config

eda1bbe

Update README with Travis status, update usage example

09a3074

lexaknyazev changed the title ~~[WIP] 2.0.0-dev.1.7~~ 2.0.0-dev.1.7 Mar 7, 2018

lexaknyazev merged commit 42d1a52 into master Mar 7, 2018

lexaknyazev deleted the 2.0.0-dev.1.7 branch March 7, 2018 18:14

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

2.0.0-dev.1.7 #59

2.0.0-dev.1.7 #59

lexaknyazev commented Dec 15, 2017 •

edited

Loading

emackey commented Dec 15, 2017

lexaknyazev commented Dec 15, 2017

lexaknyazev commented Feb 28, 2018

lexaknyazev commented Mar 6, 2018

donmccurdy Mar 6, 2018

lexaknyazev Mar 6, 2018

donmccurdy Mar 6, 2018

lexaknyazev Mar 6, 2018

donmccurdy Mar 6, 2018 •

edited

Loading

lexaknyazev Mar 6, 2018

donmccurdy Mar 6, 2018 •

edited

Loading

lexaknyazev Mar 6, 2018

zellski Mar 6, 2018

lexaknyazev Mar 6, 2018

lexaknyazev Mar 7, 2018

donmccurdy Mar 7, 2018

donmccurdy Mar 6, 2018 •

edited

Loading

lexaknyazev Mar 6, 2018

donmccurdy Mar 6, 2018

emackey commented Mar 6, 2018 •

edited

Loading

lexaknyazev commented Mar 6, 2018

emackey commented Mar 6, 2018

emackey left a comment

emackey commented Mar 7, 2018

	if (mesh.primitives
	.any((primitive) => primitive.jointsCount == 0)) {
	context.addIssue(LinkError.nodeSkinWithNonSkinnedMesh);
	}

	final extensions =
	getExtensions(map, Material, context, warnOnMultipleExtensions: true);

2.0.0-dev.1.7 #59

2.0.0-dev.1.7 #59

Conversation

lexaknyazev commented Dec 15, 2017 • edited Loading

emackey commented Dec 15, 2017

lexaknyazev commented Dec 15, 2017

lexaknyazev commented Feb 28, 2018

lexaknyazev commented Mar 6, 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

donmccurdy Mar 6, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

donmccurdy Mar 6, 2018 • edited Loading

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

donmccurdy Mar 6, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

emackey commented Mar 6, 2018 • edited Loading

lexaknyazev commented Mar 6, 2018

emackey commented Mar 6, 2018

emackey left a comment

Choose a reason for hiding this comment

emackey commented Mar 7, 2018

lexaknyazev commented Dec 15, 2017 •

edited

Loading

donmccurdy Mar 6, 2018 •

edited

Loading

donmccurdy Mar 6, 2018 •

edited

Loading

donmccurdy Mar 6, 2018 •

edited

Loading

emackey commented Mar 6, 2018 •

edited

Loading