New dev branch #30

scurest · 2018-07-15T23:07:11Z

This is my new dev branch, which is not perfect, but is a significant improvement on current master, IMO, so I'm PRing it. Main new features are skinning, experimental animation support, near-complete material support, and support for four new glTF extensions (one untested).

Hopefully I can get some feedback and testing. My internet connection is not currently up to the challenge of updating my sample model repo, so I have only tested it with an old revision of the samples ^^;;;

Coordinates

Coordinates are hard-modified at import time, as opposed to soft-modified by eg. putting a transform at the root of the tree.

Meshes

Meshes are skinned.
All texcoord and color attributes are imported. Textures can now use texcoords other than TEXCOORD_0. However, the color layers COLOR_X (X >= 1), though they are imported, don't affect materials. (COLOR_0 does, see below.)
At some point in (2.79.0, 2.79.4], Blender added RGBA vertex colors, so we now use those. Fixes Store vertex color alpha into a second color attribute #26.
>4 vertex weights are supported (untested), see Skinning influenced by more than 4 joints KhronosGroup/glTF#1381
Polygon smoothing can now be disabled as an import option (@zignig).

Materials

Node groups are now created from serialized data in node_groups.json taken from KhronosGroup/glTF-Blender-Exporter/pbr_node/glTF2.blend. Cf. glTF import for glTF export #15 (comment).
Most (all?) glTF properties are now wired up in Blender.
COLOR_0 attributes affect the material. Fixes gltf 2.0 sample model BoxVertexColors doesn't import with vertex colors #23. The work-around for the issue I mentioned there was just to, whenever a material would have applied to both meshes with and without colors, add colors to the meshes without (this means some meshes may have a COLOR_0 that didn't in the glTF file).
Images are cached (so embedded images are not duplicated if used multiple times).
The extension KHR_materials_pbrSpecularGlossiness is supported (this was easy since the group was already in the exporter.)
The extension KHR_materials_unlit is supported (I wrote the group for this; it's just a simple modification of the Metallic Roughness group).
The extension MSFT_texture_dds is supported (Blender basically already did everything for this.)
The extension KHR_texture_transform is roughed in but is NOT TESTED. @sonygod

Scene Graph

Armatures are inserted into the scene at the points where the roots of skins are. All their children are realized as bones. This avoids my unpopular single-armature idea :)
Blender scenes are created for glTF scenes if this is enabled with an import option.
But there are problems with bones still:
- ~~Bone lengths are still crazy~~ Should be okay now.
- Bone direction is still crazy, for the sample models at least. But the bone directions make the bone tree have the same matrix at each bone as the glTF node tree does at each node.
- Bones can't have non-unit scaling (this appears to be a Blender limitation)
- Meshes and cameras that are children of a bone are placed in the wrong spot. I haven't figured out what transform will move them from the tail of the bone to the head yet. (This and the previous point are not a big issue for glTFs created from Blender-like scene graphs.)

Animations

There is experimental animation support gated behind an import option. Since there is no longer a single armature for the whole forest, a single animation may result in >1 actions, one for each object it animates. The actions are named like animation-name@object-name. To play them, select the object, open the Dope Sheet, change to Action Editor, and select the desired action. (Is there a better way to import these?)
Some of the animations still look wrong (eg. Monster).

Cameras

@cinghiale's camera support was merged in.

This removes support for skinning.

…r an unexpected type

of the Blender camera, while the `aspectRatio`, if present, affects the resolution instead of the `angle_x` (since the latter is computed from the `angle_y` and the resolution in Blender)

...using just one cache-lookup function instead of one for every different type of object.

RGBA vertex colors are supported (required) by Blender 2.79 it appears.

Based on creating first a virtual forest that mirrors the scene graph we're about to create in Blender. Details to be documented later... if it works out :)

Skinning is now working (again (again))...

Messy and VERY slow for the moment, but functional so I'm commiting it for my own reference.

Comments explained the main reasoning behind the bone function. Also switching from keyframe_points.insert to the add/co loop gave a **huge** performance boost.

This is for faster slicing. This doesn't make any apparent impact on import times but hopefully it saves some memory ^^;

...because 4-component vertex colors are assumed. Should we use some shims to papers over the Blender version differences and get greater support? I don't have access to !=2.79 vresion atm though so it doesn't matter for now.

They don't point the right way or anything but they're there :)

But have an option to just put everything into the current scene too (I like that better).

This prevents loading the same one multiple times.

Using the groups from the exporter, thus relieving us form writing all our own node groups. The group for Specular Glossiness is also part of the serialized data, so supporting that is getting closer.

We add superfluous COLOR_0 layers to any mesh that uses a material that is ever applied to a primitive that has a COLOR_0 attribute. This let's us add an Attribute node to those materials without the problem of solid red colors coming in. COLOR_1, etc. are still not supported (well, they're imported of course, they just don't influence the materials).

- Avoid creating dummy vnodes for meshes when possible. - Rotate cameras to their (correct?) position - Some slightly tighter docs. - Some more robustness in realizing vnodes; it would probably be possible to use an armature for the whole forest again with minimal code changes

Animating non-bone locations should work now. Curve clean-up should be working too.

The vtree beneath an armature should now be all bones except possibly at the leaves, and we should realize them correctly(?) now. Also fix a typo with mesh names.

Avoids confusion with the similarly named node_groups stuff.

Support for pbrSpecularGlossiness was added here too.

Such as they are, anyway.

- Blender automatically replaced math.inf with HUGE, so I did it for it - Print a message about unknown camera types - Setting a value on a global scene cannot possibly be the right thing to do with the aspect ratio -- it has to go into a property of the camera. Otherwise there's a race condition where only the last camera to get created actually changes the scene. So I backed that out. - Normalized quotes.

...since it's still weird. Make the options prettier and ignore scenes by default too.

...by using multiple attributes sets. We anticipate here, since though there is a PR for this, it hasn't been merged into the spec yet.

If the accessor was used again later in the glTF, we would have seen a modified array since we hard-modified the accessor elements. Avoid doing this.

If the joints were eg. [0,0,0,0] and the weights were [1,0,0,0], previously we would overwrite the weight for 0 three times so it ended up as 0. Now it will correctly be 1.

We hardly have to do anything since Blender will already handle DDS textures so we might as well. Untested though.

Untested.

I initially thought the pose TRS was computed from (final TRS) * (rest TRS)^{-1} (ie. the opposite order of multiplication from what I now believe is correct.) Under this scheme, the pose components could depend on multiple final components, which required bringing the curves under a common domain. But under the correct order of multiplication each pose components depends only on its corresponding final component (and the rest TRS) so that turns out to have been over-complicated and, thank God, we can finally get rid of it! Though, alas! it does not fix Monster.

This ensures that COLOR_0 is always the first (and ensures Blender thinks its the default, for eg. TEXCOORD_0 and UV maps?) and avoids any gaps in the sequence of X.

- Pick better bone lengths (when the bones are rotated the armature actually looks plausible now) - Report a warning when bones had non-unit scalings in the glTF

ksons · 2018-07-17T22:15:41Z

Thanks! Great progress. Did not fully test everything but will look more closely before releasing 0.4.0.

scurest · 2018-07-18T02:15:38Z

@ksons Thanks :)

scurest and others added 30 commits March 19, 2018 12:47

Use empties instead of an armature for glTF nodes

4933ba7

This removes support for skinning.

configure the camera from the gltf data

057e3df

camera/create_camera: intentionally return an uninitialized camera fo…

a314aac

…r an unexpected type

camera/create_camera: handle a missing zfar in the perspective data

9fbe168

camera/create_camera: fallback behavior to handle a missing apesctRatio

d0d955b

create_camera: the glTF yfov must clearly go to the property angle_y

00c50f6

of the Blender camera, while the `aspectRatio`, if present, affects the resolution instead of the `angle_x` (since the latter is computed from the `angle_y` and the resolution in Blender)

removed auto smoothing

a1984ef

Tidy up __init__ and unify the caches...

486e439

...using just one cache-lookup function instead of one for every different type of object.

Tidy up the mesh creator and enable RGBA colors

109dc4b

RGBA vertex colors are supported (required) by Blender 2.79 it appears.

Begin implementing a new node creator

b0db3fd

Based on creating first a virtual forest that mirrors the scene graph we're about to create in Blender. Details to be documented later... if it works out :)

Creates skins and attach them to armatures

283257a

Skinning is now working (again (again))...

WIP animation loading

11cb513

Messy and VERY slow for the moment, but functional so I'm commiting it for my own reference.

Put animation handling in order

50656d5

Comments explained the main reasoning behind the bone function. Also switching from keyframe_points.insert to the add/co loop gave a **huge** performance boost.

Use memoryviews for buffers and buffer views.

e66f0ca

This is for faster slicing. This doesn't make any apparent impact on import times but hopefully it saves some memory ^^;

Cleanup and explain how the vforest works

9f8f073

Note that Blender 2.79 is needed

00c7eb9

...because 4-component vertex colors are assumed. Should we use some shims to papers over the Blender version differences and get greater support? I don't have access to !=2.79 vresion atm though so it doesn't matter for now.

Minor tweaks

654310c

Put cameras in the scene

86531b0

They don't point the right way or anything but they're there :)

Handle creating scenes again

f3bad07

But have an option to just put everything into the current scene too (I like that better).

Add more import options

294600e

Cache images used for texturing

94af3a1

This prevents loading the same one multiple times.

Autogenerate node group data

863001f

Using the groups from the exporter, thus relieving us form writing all our own node groups. The group for Specular Glossiness is also part of the serialized data, so supporting that is getting closer.

Fix some typos in animation.py

128a4b3

Animating non-bone locations should work now. Curve clean-up should be working too.

Avoid bone->nonbone->bone child relations

1ea94f2

The vtree beneath an armature should now be all bones except possibly at the leaves, and we should realize them correctly(?) now. Also fix a typo with mesh names.

Rename node.py -> scene.py

bf112fd

Avoids confusion with the similarly named node_groups stuff.

Large push towards full material implementation

f216c93

Support for pbrSpecularGlossiness was added here too.

Fill in sampler properties

32478f8

Such as they are, anyway.

Support KHR_materials_unlit extension

37a39dd

scurest added 13 commits July 15, 2018 02:19

Merge with upstream master

1ba2263

Merge branch 'zignig/master' into dev

6eef98b

Gate animation importing behind a flag...

891e829

...since it's still weird. Make the options prettier and ignore scenes by default too.

Allow for >4 vertex weights

6147a17

...by using multiple attributes sets. We anticipate here, since though there is a PR for this, it hasn't been merged into the spec yet.

Bugfix: do not modify vertex color accessor arrays

19462a2

If the accessor was used again later in the glTF, we would have seen a modified array since we hard-modified the accessor elements. Avoid doing this.

Bugfix: don't set vertex weights of 0.0

9d3f6c8

If the joints were eg. [0,0,0,0] and the weights were [1,0,0,0], previously we would overwrite the weight for 0 three times so it ended up as 0. Now it will correctly be 1.

Support MSFT_texture_dds extension.

064c3b9

We hardly have to do anything since Blender will already handle DDS textures so we might as well. Untested though.

Implement KHR_texture_transform

26b062d

Untested.

Add COLOR_X, etc. layers in order of X

bbf469c

This ensures that COLOR_0 is always the first (and ensures Blender thinks its the default, for eg. TEXCOORD_0 and UV maps?) and avoids any gaps in the sequence of X.

Slightly better bones

fbd47dc

- Pick better bone lengths (when the bones are rotated the armature actually looks plausible now) - Report a warning when bones had non-unit scalings in the glTF

Work-around re vertex colors for Blender 2.79.0

fe0e1af

ksons merged commit 74f8205 into ksons:master Jul 17, 2018

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

New dev branch #30

New dev branch #30

scurest commented Jul 15, 2018 •

edited

Loading

ksons commented Jul 17, 2018

scurest commented Jul 18, 2018

New dev branch #30

New dev branch #30

Conversation

scurest commented Jul 15, 2018 • edited Loading

Coordinates

Meshes

Materials

Scene Graph

Animations

Cameras

ksons commented Jul 17, 2018

scurest commented Jul 18, 2018

scurest commented Jul 15, 2018 •

edited

Loading