Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Apple proposes adding "model" element to HTML that displays 3D content

github.com

Post reply on HN
Print view View original

181–190 of 241 posts

Re: Apple proposes adding "model" element to HTML that displays 3D content

#181

Post by dmitriid »

gfxgirl wrote: post #138
JusticeJuice wrote: post #50

Based on the comments so far, I think I'm in the minority who think this would be a good feature for the web. This proposal might not be perfect, but I think the idea is great. There's lots of good reasons to semantically want to embed a 3d model in a page, like any other type of media. Say you're writing a course explaining how differential gears work - a 3d model of that could be really handy to look at to learn. A…

The problem is no feature set will meet even 5% of the needs of most devs. Maybe you'd like to display a "model" of a city. Oh, too big, needs to be streamed, LODed, etc. Maybe you'd like to display a "model" of a forest. Oh, well, you need a terrain renderer that dynamically generates geometry from height data plus you need a specialized renderer to render the grass and bushes else the frame rate will be 1fps. You'd…

> Oh, you say, we'll just add features to cover those use cases. Now Chrome will have features A,B,C, Firefox, A,B,D, Safari B,E.

Yay! OpenGL with extensions all over again

Re: Apple proposes adding "model" element to HTML that displays 3D content

#182

Post by pphysch »

brundolf wrote: post #148
JusticeJuice wrote: post #50

Based on the comments so far, I think I'm in the minority who think this would be a good feature for the web. This proposal might not be perfect, but I think the idea is great. There's lots of good reasons to semantically want to embed a 3d model in a page, like any other type of media. Say you're writing a course explaining how differential gears work - a 3d model of that could be really handy to look at to learn. A…

My first thought was "this probably wouldn't work for games, and other uses of embedded 3D content are pretty niche" But then I thought, maybe they're niche because something like this doesn't already exist

3D is a poor choice for many visualizations because when it is rendered in 2D, information is hidden behind the model being transformed over time (either manually by the user or via automatic rotation, etc).

IMO it should only be considered if there are challenges with flattening the data in question. For example, if the data represents an actual scanned 3D object.

On the other hand, rendering arbitrary data as "3D" models is usually a cute waste of time. There are much more effective ways to explore and analyze it.

Re: Apple proposes adding "model" element to HTML that displays 3D content

#183

Post by shawnz »

dragonwriter wrote: post #168
bastawhiz wrote: post #155

Earlier quoted context omitted.

Apple isn't doing anything that's incompatible with X3D. In fact, they didn't even specify which kinds of models they would accept (there's a `type` on the elements). You could easily point one of those at an X3D file. Sure, you could put an tag into a HTML document, but there's no ` ` equivalent (like you can use with SVG) to load that content from another URI. When dealing with extremely large (byte-wise) or comple…

> Sure, you could put an tag into a HTML document, but there's no ` ` equivalent (like you can use with SVG) to load that content from another URI. Yes, there is, the HTML tag. https://www.web3d.org/x3d/content/HtmlObjectTagForX3d.html

is not a satisfactory equivalent for this tag any more than it's a satisfactory equivalent for the tag itself. is semantically a superset of both.

Re: Apple proposes adding "model" element to HTML that displays 3D content

#184

Post by achairapart »

evrflx wrote: post #40

I would love to see that apple negotiates their wishes but have in turn to implement web APIs they currently keep back by choice. And let users use a real different browser on their device.

Google: "Hey Apple, here is your pet tag for your next AR/VR secret project. Now, here there are all the PWA APIs that are still missing in Safari. Thank you."

Re: Apple proposes adding "model" element to HTML that displays 3D content

#185

Post by comex »

theunspoken wrote: post #163

I am very fed up with standards lacking a free and open reference implementation. It's very easy to produce a specification for embedding 3D models into a browser, but who is going to follow up with and actual complying browser? The big guys, that's who. Google, Microsoft, Apple can just throw piles and piles of money at the problem; but what about everyone else? What about Firefox? What about Palemoon? What about Fa…

Google and Apple's engines are open source (and Microsoft now contributes to Google's). What would be the benefit of a separate reference implementation of this standard? I suppose it would be easier to port to other browsers if it were designed as a standalone library rather than tightly integrated into an engine. But the idea of going to pains to design a standalone library seems to contradict your later statement that you want an entire reference browser.

A web browser is orders of magnitude more complex than a video encoder or a Wayland server. Not aiming for high performance would make a browser a little less complex, but not by much, so there doesn't seem to be much point. Minus that, you're just asking for someone (who?) to write a new web browser from scratch. It would certainly be nice if there were more independent web browser codebases in general, but I don't see what special benefit a new one would bring to the table just due to being marked as a "reference implementation".

Re: Apple proposes adding "model" element to HTML that displays 3D content

#187

Post by Jasper_ »

southerntofu wrote: post #178
bhhaskin wrote: post #124

Earlier quoted context omitted.

Seems already possible using a custom web component.

Sure, but that doesn't make standardization useless so people stop reinventing this same wheel over and over again for very basic needs.

I wouldn't call "display a chair in a web page in an inexact and implementation-defined manner, with air quotes 'interactivity'" a "very basic need".

This seems like a very specific use case, which I think is best left to custom code.

Re: Apple proposes adding "model" element to HTML that displays 3D content

#188

Post by onion2k »

soylentgraham wrote: post #107
dangerface wrote: post #68

I think apple should just fix their webgl implementation. They seem to be missing a fundamental feature of the web it's interactive. They have an interactive attribute for the tag but no description of what this means? What does it mean to interact with a 3d model? Can 3d model interaction be accurately defined in a single attribute? Why not just use javascript for that logic? The whole point of having multi media li…

Its a few lines of code... for a coder, and even then, its 100 lines of setup, shaders, model loading. Vs a single tag to embed a model. For a non-coder. You dont have to know how to encode/decode videos to add video to a page. Im surprised it's taken this long. (Whether it'll turn out good, not sure, but it will definitely be adopted, and fast)

Vs a single tag to embed a model.

3D models are complicated. If the element is complex enough to support anything the user decides to embed then it'll put a huge burden on browser developers. That's fine for Apple and Google, but it'd be another nail in Firefox's coffin. On the other hand, if the element only supports (say) the gltf spec then it won't be a simple case of adding a model for a non-technical user. They'd need to export or convert their model, textures, shaders, animations, etc.

3D support is probably an order of magnitude harder than video. And video is hard enough.

Re: Apple proposes adding "model" element to HTML that displays 3D content

#189

Post by pjmlp »

dmitriid wrote: post #181
gfxgirl wrote: post #138

Earlier quoted context omitted.

The problem is no feature set will meet even 5% of the needs of most devs. Maybe you'd like to display a "model" of a city. Oh, too big, needs to be streamed, LODed, etc. Maybe you'd like to display a "model" of a forest. Oh, well, you need a terrain renderer that dynamically generates geometry from height data plus you need a specialized renderer to render the grass and bushes else the frame rate will be 1fps. You'd…

> Oh, you say, we'll just add features to cover those use cases. Now Chrome will have features A,B,C, Firefox, A,B,D, Safari B,E. Yay! OpenGL with extensions all over again

It never stopped being an extension spaghetti, Khronos loves extensions.

https://webglreport.com/?v=2 reports 10 extensions.

And you can get the full spectrum over GL, GL ES and Vulkan, here https://gpuinfo.org.

Post reply on HN