A call for video on the web - Opera <video> release on Labs

To really make a splash on the Web, video needs an open solution that can easily be integrated into web pages without the need for proprietary plugins. The HTML 5 <code><video></code> element and Ogg Theora can provide this, and Opera is proud to announce an experimental build that suppports it. So read this article, and download and play with it today.


( Read the article )

1) I am interested in the reasoning behind the push for a new HTML element at a time when browser developers are still putting the final pieces in place for cross-browser support of an element which *already* embeds video, namely <object>. What clear and explicitly technical advantages does <video> have which would be impossible to implement with <object type="video/theora"> ?


2) With the introduction of the <object> element in HTML4, developers were excited about a single element with the potential to replace a host of others used for separate types of external content, especially the disparate and badly cross-browser supported <applet> and <embed> elements. At the time, <object> was seen as the successor, not only to those elements, but to <img> and <iframe> as well.
Indeed, all of Opera, FF and IE7 already support <object type="text/html"> in an <iframe> fashion (Opera and FF even add it to the window.frames DOM object)...
Now, with the promotion of the media-specific <video> element, the original push for a Great Unification of all external media under one generic element seems to have been reversed. My question is this: The W3C originally created the <object> element to replace two very different means for embedding video; how do you plan to convince them to re-authorize co-existing methods for embedding video?


3) Building on my last question: How is the creation of a <video> element different in principle than the creation of, for example, <tif>, <tga> or <raw> elements to include specific image media types outside of the currently recommended <img> and <object>? I say "in principle", because the difference in scope is obvious.

Thankfully it works in Wine.

GreyWyvern, I hope this post from Anne clears things up: http://annevankesteren.nl/2007/03/video

This build of Opera also supports video in SVG, and the 3d <canvas>

Any chance you'll put up p01's 3D snake then

Post edited Wednesday, 7. November 2007, 19:50:23

Originally posted by hzr:

3d <canvas>

There's a working 3D canvas demo in Erik Dahlström's SVG Open 2007 presentation titled: "Quick recipes for SVG Wow!" Unfortunately, I don't know where I've downloaded the slides from... Anyone?

It's not exactly fluent, but that might be my PC.

JeroenH, thanks, I found the presentation[1], but the links to the demos seems to be broken.

[1] http://www.svgopen.org/papers/abstract3/

I thought the W3C decided it didn't like truncated attributes (instead of using <input type="radio" checked />, they wanted people to use <input type="radio" checked="checked" /> This seems like a step backwards, and a bit counter-intuative (plus more difficult to manipulate with DOM, I'd think). Furthermore, it isn't as evident that it is a boolean value, or what the potential options for the controls attribute are. It seems like it'd be better to use controls="display" and control="hide", or controls="true" and controls="false".

I'm assuming this build is similar to Hakon's build, released on March 29, 2007, with quite a bit of work having been done to this build since the last <video> build.

Great demos!! Very exciting. Only, the video didn't stop when I changed the tab's address. (Also an indication of buffering status would be nice.)

GreyWyvern: Look at some of the demos and you'll see why <object> is inadequate and that there are characteristics of video that make it distinct from both <img> and <object>.

iceArdor: I'm sure controls="on" would work. And I doubt it's fixed.

Originally posted by IceArdor:

I thought the W3C decided it didn't like truncated attributes (instead of using <input type="radio" checked />, they wanted people to use <input type="radio" checked="checked" />

It's just XHTML vs HTML. Nothing wrong in using HTML syntax

Originally posted by IceArdor:

This seems like a step backwards, and a bit counter-intuative (plus more difficult to manipulate with DOM, I'd think).

You're wrong. There are no problem with DOM manipulations in HTML concerning shortened boolean attributes.

Originally posted by IceArdor:

Furthermore, it isn't as evident that it is a boolean value, or what the potential options for the controls attribute are. It seems like it'd be better to use controls="display" and control="hide", or controls="true" and controls="false".

For anyone who ever use HTML syntax it is quite evident Actually the only way to use boolean attributes in XHTML is in form controls="controls" which I'm pretty sure also works in this build. Anyway I also prefer XHTML syntax but there are nothing wrong in HTML one. It's only matter of taste.
To be specific HTML syntax is preferred for documents send as text/html. XHTML should not be send with that mime type.


BTW: These demos are awesome. Especially "video-reflect" and "video-filter".

wow, that's amazing! i didn't know something like that was in the pipeline. it sure seems like a good idea to me..

Originally posted by hzr:

thanks, I found the presentation[1], but the links to the demos seems to be broken.

I found the demos.

It's quite amazing what Opera can put in just over 5mb by the way.

Cool. I would like to read more about 3d canvas though...

Post edited Thursday, 8. November 2007, 15:02:33

I like this, I like it alot, but valueless attributes are generally icky. This is probably the wrong place to complain, but I don't feel like signing up for the mailing list just to point it out. What's wrong with controls="none" and controls="all" ?

My god, to be able to do away with crappy flash flv players though. It boggles.

Originally posted by JeroenH:

I found the demos.

Thank you very much

Originally posted by johnoyler:

I like this, I like it alot, but valueless attributes are generally icky. [...] What's wrong with controls="none" and controls="all" ?

Nothing wrong using the unminimized form (which is controls="controls" BTW), but the minimization is nice when hand authoring. The resulting DOM will look the same, so I don't see why so many people see this as a problem. It's just a shortcut.

Why a new "video" element? Wasn't "object" with right mime type perfectly appropriate for video too?

A standard javascript interface for video could be also defined to script the "object". So, again, what's the point for the new tag?

zarlino, did you read the article and see the videos ?
all plugins have the scripting problems, besides requiring different implementations between browser and operating systems.

That filters thing + svg left me drooled !

I hope that music becomes a first-class citizen some time soon.

Here is a small demo of 3D canavs I made:

http://files.myopera.com/grafio/files/demo1.html

(thanks JeroenH for finding the Opera demos)

Post edited Friday, 9. November 2007, 11:09:05

Nice work grafio, how is your performance? I get like 2 frames/second...

By the way: these are the methods and properties of the 3d context.

blend:replace
color:#ffffff
farPlane:100
fov:90
nearPlane:0.1
texture:[object Canvas3DTexture]
ztest:lessequal
save:function save() { [native code] }
restore:function restore() { [native code] }
beginScene:function beginScene() { [native code] }
endScene:function endScene() { [native code] }
translate:function translate() { [native code] }
scale:function scale() { [native code] }
rotateX:function rotateX() { [native code] }
rotateY:function rotateY() { [native code] }
rotateZ:function rotateZ() { [native code] }
drawTriangle:function drawTriangle() { [native code] }
draw3DModel:function draw3DModel() { [native code] }
createTexture:function createTexture() { [native code] }
create3DModel:function create3DModel() { [native code] }
checkIntersection:function checkIntersection() { [native code] }

Well on my PC it's at least 25 fps I think.

It's great to finally see these builds in the wild!

Wow. I am impressed. (I have a widget that can benefit from the new canvas context.)

Any chance these are going to make the final 9.5?
One can dream...

@xErath
you didn't get my opinion. A standard Javascript interface can be created for "Objects". Scripts can test if it is supported as they already do with other scripting features. Something like "if(videoElement.isStandardInterface)".

That said browsers could provide native player functionality for certain MIME types. Also plugins could implement that standard interface. And voilà you don't need to extend HTML in any way.

I played with the build a bit and I quite enjoyed it. Html video element works pretty nice, hopefuly it will be there in the final build.

Good work there, very impressive

Only thing that can improve on that, is having SMIL Timesheets (when ready) integrate the video with the other elements on the page.

http://www.w3.org/TR/2008/WD-timesheets-20080110/

jose

Originally posted by johnoyler:

I like this, I like it alot, but valueless attributes are generally icky. This is probably the wrong place to complain, but I don't feel like signing up for the mailing list just to point it out. What's wrong with controls="none" and controls="all" ?

My god, to be able to do away with crappy flash flv players though. It boggles.

Yeah I came to this forum specifically to find out what Opera Mobile's flash support was, largely to figure out it's true video support.