Re: [whatwg] Dealing with Stereoscopic displays
On Mon, 26 Apr 2010, Eoin Kilfeather wrote: I was wondering if any though had been given to a consistant way of dealing with stereoscopic displays. A use case has come up in a project I am working on which calls for the use of stereoscopic UIs but I can find no metion of the term in either the specs or the mailing list archive. My though was that perhaps it could be achieved by having a browsing context _left_display or _right_display allowing the user agent would render the left/right eye views to the correct display in a technology agnostic way. Apologies if this has already been covered. As others mentioned on this thread, this is the kind of thing that is typically abstracted out of the platform. HTML in general isn't even one-dimensionsal -- it's an abstract language for describing documents and applications at a conceptual level, not a graphics or presentational language. For 3D data, WebGL is probably the most applicable technology; one would imagine that with stereoscopic displays WebGL would just work and render the output in stereoscopic 3D. You'd similarly expect it to just work with holographic 3D displays, non-holographic volumetric displays, lenticular autostereoscopic displays, etc. We wouldn't want to have the author work in terms of a left display and a right display. -- Ian Hickson U+1047E)\._.,--,'``.fL http://ln.hixie.ch/ U+263A/, _.. \ _\ ;`._ ,. Things that are impossible just take longer. `._.-(,_..'--(,_..'`-.;.'
Re: [whatwg] Dealing with Stereoscopic displays
Hi Rob, all, Fair enough :-) I'll have to try better. Rob you give some good examples (WebGL and CSS3) of how an application could be built which correctly renders two views with stereopsis. However, with the exception of Anaglyph methods, a user will need specialised display hardware to properly view the image. So, to clarify, the issue is not the rendering of the stereo views (I'll worry about that later) but rather how those views are targeted to the correct virtual display (for example by alternating the left and right views on odd and even frames). If we take the case of the Blu-Ray 3D specification it is neutral about how the hardware is implemented, but the hardware is expected to respect the flags indicating whether a frame is for the left or right virtual display. In order to work with HTML the UA has to have some awareness of the hardware and way of signalling with view is for which virtual display. My question is, how can this be done in a consistent manner? Given that this usually requires some hardware control, is a good approach to use the device element? I hope this is a little clearer. Best regards, Eoin. On Tue, Apr 27, 2010 at 10:01 AM, Robert O'Callahan rob...@ocallahan.orgwrote: On Tue, Apr 27, 2010 at 8:38 PM, Eoin Kilfeather ekilfeat...@dmc.dit.iewrote: * A user visits the National Museum site and wants to see a time-machine view of objects in the collection with a sense of 3D depth based on their age I think this is the closest you get to an actual use-case :-). The rest is mixed up with information about possible solutions. Also, it's highly unlikely the a user will visit your site with a fully formed desire to view objects in a collection with a sense of 3D depth based on their age :-). But let's say the authors of that site want to visualize objects in the collection with different objects at different depths. It seems to me either WebGL or CSS 3D transforms --- or a mixture --- could be used for this, maybe with some extra information provided to identity the camera positions for rendering the stereo views. Actually, I probably shouldn't be involved in this discussion since I'm monocular :-). Rob -- He was pierced for our transgressions, he was crushed for our iniquities; the punishment that brought us peace was upon him, and by his wounds we are healed. We all, like sheep, have gone astray, each of us has turned to his own way; and the LORD has laid on him the iniquity of us all. [Isaiah 53:5-6] -- Eoin Kilfeather Digital Media Centre Dublin Institute of Technology
Re: [whatwg] Dealing with Stereoscopic displays
On Wed, Apr 28, 2010 at 8:09 PM, Eoin Kilfeather ekilfeat...@dmc.dit.iewrote: If we take the case of the Blu-Ray 3D specification it is neutral about how the hardware is implemented, but the hardware is expected to respect the flags indicating whether a frame is for the left or right virtual display. In order to work with HTML the UA has to have some awareness of the hardware and way of signalling with view is for which virtual display. Sure, but this seems like a UA-specific issue that the Web author should not need to worry about. UAs already coordinate with the underlying software and hardware platform to render Web content. Rob -- He was pierced for our transgressions, he was crushed for our iniquities; the punishment that brought us peace was upon him, and by his wounds we are healed. We all, like sheep, have gone astray, each of us has turned to his own way; and the LORD has laid on him the iniquity of us all. [Isaiah 53:5-6]
Re: [whatwg] Dealing with Stereoscopic displays
Well, I agree that the web author shouldn't worry about how it is achieved, but would it not be the case that the author needs to indicate which view is for which display? That is to say the author would be required to flag the output for correct routing to the virtual display. Is it beyond the scope to the specification to indicate a normative way of doing this? Best, Eoin On Wed, Apr 28, 2010 at 9:17 AM, Robert O'Callahan rob...@ocallahan.orgwrote: On Wed, Apr 28, 2010 at 8:09 PM, Eoin Kilfeather ekilfeat...@dmc.dit.iewrote: If we take the case of the Blu-Ray 3D specification it is neutral about how the hardware is implemented, but the hardware is expected to respect the flags indicating whether a frame is for the left or right virtual display. In order to work with HTML the UA has to have some awareness of the hardware and way of signalling with view is for which virtual display. Sure, but this seems like a UA-specific issue that the Web author should not need to worry about. UAs already coordinate with the underlying software and hardware platform to render Web content. Rob -- He was pierced for our transgressions, he was crushed for our iniquities; the punishment that brought us peace was upon him, and by his wounds we are healed. We all, like sheep, have gone astray, each of us has turned to his own way; and the LORD has laid on him the iniquity of us all. [Isaiah 53:5-6] -- Eoin Kilfeather Digital Media Centre Dublin Institute of Technology
Re: [whatwg] Dealing with Stereoscopic displays
On 04/28/2010 10:39 AM, Eoin Kilfeather wrote: Well, I agree that the web author shouldn't worry about how it is achieved, but would it not be the case that the author needs to indicate which view is for which display? That is to say the author would be required to flag the output for correct routing to the virtual display. Is it beyond the scope to the specification to indicate a normative way of doing this? I think the idea is that rather than the author manually producing different content for each display, the 3D positional information in the underlying format (e.g. WebGL) would be used by the browser to automatically create a 3D view on the hardware available.
Re: [whatwg] Dealing with Stereoscopic displays
On 2010-04-27 00:59, Robert O'Callahan wrote: I think it's interesting to think about what browsers could do with stereo output. We already have three features that could produce useful stereo output today: 1) WebGL 2) CSS 3D Transforms 3)video (assuming there was some kind of 3D video format defined elsewhere) Of those, I think 3D video is perhaps the most significant, followed by WebGL for, e.g. 3D games. But I don't think these are issues that should be addressed by the WHATWG or HTMLWG at this stage, particularly for video. 3D video would depend on the availability of a suitable 3D video codec that can be implemented by browsers. Something like h.264 MVC (Multiview Video Coding) as used by 3D Blu-ray would be nice, but we would need something that is royalty free. I don't know of any other 3D video codec, especially not among the royalty free choices. I think this issue should be revisited in the future if and when such a codec becomes available. But given that the the actual video rendering is handled by the implementation, completely transparently from the DOM API, the current markup and DOM API should be sufficient for basic playback support. 3D Enhancements to the DOM API and markup can be considered later if necessary, such as figuring out how to deal with drawing an HTMLVideoElement playing a 3D video onto a 2D canvas. -- Lachlan Hunt - Opera Software http://lachy.id.au/ http://www.opera.com/
Re: [whatwg] Dealing with Stereoscopic displays
Hi again, Thanks for the replies and informed feedback. I wasn't fully aware of the significance of the replicate proposal, and I like what it potential offers for my use case, but the hardware control problem remains. Essentially the UC is this; USE CASE: A web developer wants to utilise the next generation of stereoscopic displays (for arguments sake we assume that these are going to become ubiquitous as quickly as LCD flat-screens did) for UIs which create an impression of depth (coverflows, time-machines, head-up-displays, etc.) SCENARIOS: * A user visits the National Museum site and wants to see a time-machine view of objects in the collection with a sense of 3D depth based on their age * Her PC is connected to a stereoscopic screen but the web application can't know the details of the implementation: Anaglyph glasses, polarising glasses, lenticular cover etc. * The web page has a device selector with type = stereo_display (?) which detects / gives access to the stereo functions of the display - i.e. turns on whatever feature gives stereopsis * The UA has awareness of a left and right render path for two widows / documents but knows that these are stereoscopically linked (is this sensible ?) * The web application now has two render targets * The web application now generates slightly different left eye and right eye views * The UA renders the two documents in the correct window REQUIREMENTS: * Stereo displays should be discoverable (through device ?) * Stereo displays should be controllable by the UA (again through device ?). * Scripts should have access to both render targets Any suggestions / criticisms? Regards, Eoin. On Tue, Apr 27, 2010 at 3:15 AM, ddailey ddai...@zoominternet.net wrote: No it isn't simple. Allied issues have been discussed here before. As the nature of input devices become richer (e.g. eye movement glasses that give binocular disparity data to the display device) then the nature of the convergence data that defines the scene becomes more relevant to its primary semantics. As SVG and 3D technologies begin to bridge the gap between 2 and 3D (cf. the replicate proposal [1] or [2] ) the distinction between styling and markup so tenaciously held in HTML may cease to be so clearcut. cheers David [1] http://old.nabble.com/A-proposal-for-declaritive-drawing-(%3Creplicate%3E)-to-be-added-into--SVG-td28155426.html [2] http://srufaculty.sru.edu/david.dailey/svg/SVGOpen2010/replicate.htm - Original Message - From: David Singer sin...@apple.com To: whatwg@lists.whatwg.org Sent: Monday, April 26, 2010 8:02 PM Subject: Re: [whatwg] Dealing with Stereoscopic displays I agree that this probably means that web elements that are 'flat' would be styled by CSS with a depth. This is important if other material presented to the user really is stereo (e.g. a left/right eye coded movie). The movie will be set so that the eyes are expected to have a certain 'convergence' (i.e. they are looking slightly inward towards some point) and it's important that if material is overlaid on that. it has the same convergence. Obviously, this is unlike the real world where focus distance and convergence distance are the same (focus distance is fixed at the screen distance), but the brain can get very confused if two things that are both in focus are at difference convergence distances. This is not a simple question, as I expect you are beginning to realize. David Singer Multimedia and Software Standards, Apple Inc. -- Eoin Kilfeather Digital Media Centre Dublin Institute of Technology Aungier Street Dublin 2 m. +353 87 2235928 skype:ekilfeather
Re: [whatwg] Dealing with Stereoscopic displays
On Tue, Apr 27, 2010 at 8:38 PM, Eoin Kilfeather ekilfeat...@dmc.dit.iewrote: * A user visits the National Museum site and wants to see a time-machine view of objects in the collection with a sense of 3D depth based on their age I think this is the closest you get to an actual use-case :-). The rest is mixed up with information about possible solutions. Also, it's highly unlikely the a user will visit your site with a fully formed desire to view objects in a collection with a sense of 3D depth based on their age :-). But let's say the authors of that site want to visualize objects in the collection with different objects at different depths. It seems to me either WebGL or CSS 3D transforms --- or a mixture --- could be used for this, maybe with some extra information provided to identity the camera positions for rendering the stereo views. Actually, I probably shouldn't be involved in this discussion since I'm monocular :-). Rob -- He was pierced for our transgressions, he was crushed for our iniquities; the punishment that brought us peace was upon him, and by his wounds we are healed. We all, like sheep, have gone astray, each of us has turned to his own way; and the LORD has laid on him the iniquity of us all. [Isaiah 53:5-6]
[whatwg] Dealing with Stereoscopic displays
Hello, I was wondering if any though had been given to a consistant way of dealing with stereoscopic displays. A use case has come up in a project I am working on which calls for the use of stereoscopic UIs but I can find no metion of the term in either the specs or the mailing list archive. My though was that perhaps it could be achieved by having a browsing context _left_display or _right_display allowing the user agent would render the left/right eye views to the correct display in a technology agnostic way. Apologies if this has already been covered. Eoin. -- Eoin Kilfeather Digital Media Centre Dublin Institute of Technology Aungier Street Dublin 2
Re: [whatwg] Dealing with Stereoscopic displays
Eoin Kilfeather ekilfeat...@dmc.dit.ie schrieb am Mon, 26 Apr 2010 17:04:47 +0100: Hello, I was wondering if any though had been given to a consistant way of dealing with stereoscopic displays. A use case has come up in a project I am working on which calls for the use of stereoscopic UIs but I can find no metion of the term in either the specs or the mailing list archive. […] What exactly are you trying to achieve ? Also, this is a presentational issue — maybe in scope of CSS WG ? -- Nils Dagsson Moskopp // erlehmann http://dieweltistgarnichtso.net signature.asc Description: PGP signature
Re: [whatwg] Dealing with Stereoscopic displays
On Tue, Apr 27, 2010 at 2:04 AM, Eoin Kilfeather ekilfeat...@dmc.dit.ie wrote: Hello, I was wondering if any though had been given to a consistant way of dealing with stereoscopic displays. A use case has come up in a project I am working on which calls for the use of stereoscopic UIs but I can find no metion of the term in either the specs or the mailing list archive. My though was that perhaps it could be achieved by having a browsing context _left_display or _right_display allowing the user agent would render the left/right eye views to the correct display in a technology agnostic way. Apologies if this has already been covered. Eoin. I do not think such a need has been taken to the Web yet. I am wondering: Would that require a Web page to be defined as actually two basically complete Web pages, separately coded, and then marked as targeted at the left eye and the right eye? That would be a very special use case on the Web and hardly render on every Desktop other than maybe as two Web pages one behind the other. If you need something like this, why don't you implement a demo with two actually separate Web pages and a main page that somehow connects the two pages to your stereoscopic display, targeting one at each channel? JavaScript will help. Then you can find out if there is some technical limit and how it may be done in HTML if there is a need for extension. Cheers, Silvia.
Re: [whatwg] Dealing with Stereoscopic displays
On Mon, Apr 26, 2010 at 6:09 PM, Silvia Pfeiffer silviapfeiff...@gmail.comwrote: On Tue, Apr 27, 2010 at 2:04 AM, Eoin Kilfeather ekilfeat...@dmc.dit.ie wrote: I was wondering if any though had been given to a consistant way of dealing with stereoscopic displays. A use case has come up in a project I am working on which calls for the use of stereoscopic UIs but I can find no metion of the term in either the specs or the mailing list archive. My though was that perhaps it could be achieved by having a browsing context _left_display or _right_display allowing the user agent would render the left/right eye views to the correct display in a technology agnostic way. Apologies if this has already been covered. I do not think such a need has been taken to the Web yet. I am wondering: Would that require a Web page to be defined as actually two basically complete Web pages, separately coded, and then marked as targeted at the left eye and the right eye? That would be a very special use case on the Web and hardly render on every Desktop other than maybe as two Web pages one behind the other. If you need something like this, why don't you implement a demo with two actually separate Web pages and a main page that somehow connects the two pages to your stereoscopic display, targeting one at each channel? JavaScript will help. Then you can find out if there is some technical limit and how it may be done in HTML if there is a need for extension. I would think in most cases you would be using something like WebGL to render the 3D scene anyway, and WebGL already has all the information to render to a stereoscopic display. For normal DOM elements, it might be reasonable to have a renderDepth attribute, and have the browser render the left/right versions properly to convey the depth, but that seems more like a gimmick than anything useful. Given the precision required, I don't think separately coding left/right web pages would be useful at all. -- John A. Tamplin Software Engineer (GWT), Google
Re: [whatwg] Dealing with Stereoscopic displays
I think it's interesting to think about what browsers could do with stereo output. We already have three features that could produce useful stereo output today: 1) WebGL 2) CSS 3D Transforms 3) video (assuming there was some kind of 3D video format defined elsewhere) What are the use cases for stereo output? Would those features be sufficient? Rob -- He was pierced for our transgressions, he was crushed for our iniquities; the punishment that brought us peace was upon him, and by his wounds we are healed. We all, like sheep, have gone astray, each of us has turned to his own way; and the LORD has laid on him the iniquity of us all. [Isaiah 53:5-6]
Re: [whatwg] Dealing with Stereoscopic displays
I agree that this probably means that web elements that are 'flat' would be styled by CSS with a depth. This is important if other material presented to the user really is stereo (e.g. a left/right eye coded movie). The movie will be set so that the eyes are expected to have a certain 'convergence' (i.e. they are looking slightly inward towards some point) and it's important that if material is overlaid on that. it has the same convergence. Obviously, this is unlike the real world where focus distance and convergence distance are the same (focus distance is fixed at the screen distance), but the brain can get very confused if two things that are both in focus are at difference convergence distances. This is not a simple question, as I expect you are beginning to realize. David Singer Multimedia and Software Standards, Apple Inc.
Re: [whatwg] Dealing with Stereoscopic displays
No it isn't simple. Allied issues have been discussed here before. As the nature of input devices become richer (e.g. eye movement glasses that give binocular disparity data to the display device) then the nature of the convergence data that defines the scene becomes more relevant to its primary semantics. As SVG and 3D technologies begin to bridge the gap between 2 and 3D (cf. the replicate proposal [1] or [2] ) the distinction between styling and markup so tenaciously held in HTML may cease to be so clearcut. cheers David [1] http://old.nabble.com/A-proposal-for-declaritive-drawing-(%3Creplicate%3E)-to-be-added-into--SVG-td28155426.html [2] http://srufaculty.sru.edu/david.dailey/svg/SVGOpen2010/replicate.htm - Original Message - From: David Singer sin...@apple.com To: whatwg@lists.whatwg.org Sent: Monday, April 26, 2010 8:02 PM Subject: Re: [whatwg] Dealing with Stereoscopic displays I agree that this probably means that web elements that are 'flat' would be styled by CSS with a depth. This is important if other material presented to the user really is stereo (e.g. a left/right eye coded movie). The movie will be set so that the eyes are expected to have a certain 'convergence' (i.e. they are looking slightly inward towards some point) and it's important that if material is overlaid on that. it has the same convergence. Obviously, this is unlike the real world where focus distance and convergence distance are the same (focus distance is fixed at the screen distance), but the brain can get very confused if two things that are both in focus are at difference convergence distances. This is not a simple question, as I expect you are beginning to realize. David Singer Multimedia and Software Standards, Apple Inc.