Open skonig-google opened 6 years ago
Ha, that's very interesting.
Can you help me get some clarity at where to draw the line between things that should go into GLTF-like formats and 3DModel
? How should we think about the relationship between these two things from a layering perspective?
Would something along the lines of "GLTF is to 3DModel
as JPEG is to <img>
?" That is, you use JPEG to store/serve an image, but you use <img>
to set its scale (e.g. <img width="123" height="456">
) and position (e.g. <img style="position: absolute; x: 123px; y: 456px;">
). Or is a better analogy here something akin to "GTLF is to 3DModel as JPEG is to SVG"?
Is that the way you are thinking of 3DModel
?
Taking the <img>
tag analogy further, can you have (and position and scale) more than one GLTF model in a 3DModel
(e.g. composition)?
Also, since this just came out today too, wondering what would be the relationship between 3DModel
and web-components based models. Example:
https://magicleaphelio.com/devsamples/animated-model
<ml-model
id="birdModel"
src="res-2/hummingbird.fbx"
materials="
kmat: res-2/hummingbird_Material.kmat;
textures:
res-2/hummingbird_Material_Albedo.png
res-2/hummingbird_Material_Reflection.png
res-2/hummingbird_Material_AO.png
res-2/hummingbird_Material_Normal.png;"
model-scale="0.3 0.3 0.3"
model-animation="flightLoop, false, -1"
model-animation-speed="1"
z-offset="300">
</ml-model>
Yeah, the relationship between glTF and 3DModel is like JPG to . For one thing there are different 3D model formats (Apple has proposed USDZ for example), but conceptually they both represent 3D models and they both would have constraints that are independent of the format (like where they can be placed).
In the short term a web component could be used to provide declarative
markup around 3D models, since the only way to render them at present is
via WebGL. In the medium term a proper
On Wed, Aug 8, 2018 at 7:13 PM sam goto notifications@github.com wrote:
Also, since this just came out today too, wondering what would be the relationship between 3DModel and web-components based models. Example:
https://magicleaphelio.com/devsamples/animated-model
<ml-model id="birdModel" src="res-2/hummingbird.fbx" materials=" kmat: res-2/hummingbird_Material.kmat; textures: res-2/hummingbird_Material_Albedo.png res-2/hummingbird_Material_Reflection.png res-2/hummingbird_Material_AO.png res-2/hummingbird_Material_Normal.png;" model-scale="0.3 0.3 0.3" model-animation="flightLoop, false, -1" model-animation-speed="1" z-offset="300">
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/samuelgoto/arpub/issues/2#issuecomment-411583326, or mute the thread https://github.com/notifications/unsubscribe-auth/APftR8jDCDWXo1YSBoAaT9sIQKjNasdzks5uO3CmgaJpZM4VssfY .
Many AR experiences involve placing 3D assets (e.g., glTF models) into the real world. Most of our examples to date have been about bringing 2D assets into the real world though.
I wonder if it's worth having explicit support for 3D models as an asset type. Doing so could also provide properties or constraints about how the 3D object should be rendered, for example:
An example of such an annotation might be the following: