[Opensim-dev] thinking about a viewer

Wed Aug 6 03:34:48 UTC 2014

Diva: I've been thinking along the lines of your #4 suggestion. The base
scene viewer should be architecturally like an X11 server -- just takes
commands to draw things. For a 3D world, the necessary graphics functions
would be more than frame buffers and overlapping windows but would enable
layering of 3D content from many different sources. And then, since
JavaScript workers can be loaded dynamically, rather than embedding an
avatar renderer in the viewer, the avatar renderer would be fetched from
that avatar asset server. That way, different implementations of avatars
could stand side-by-side without having to worry about viewer versions.

All the other support services would be broken out -- chat, inventory,
mapping. They wouldn't be part of the scene viewer but would be other
JavaScript pages that could be in separate windows or displayed on a panel
in the view (needed for the people wearing goggles).

Also, with this loose organization, everything doesn't have to run on the
same computer.

As modular and extensible as possible.

On Tue, Aug 5, 2014 at 5:20 PM, Michael Emory Cerquoni <
nebadon2025 at gmail.com> wrote:

> I am not sure that WebGL is at a level where it could displace or replace
> the standard SL viewer even as a simple viewer only, some experimenting
> with WebGL that we have done shows that it is very very difficult to render
> a typical SL/OpenSimulator scene in the web browser, the performance is
> also not great even on the most top of the line hardware, personally I
> think using WebGL would be a big step down graphically and functionally, I
> like the idea of a easy to use no-nonsense don't have to install anything
> viewer, conceptually it sounds great, but not at the expense of a degraded
> experience which is what I suspect we would end up with if we went down the
> WebGL path today.  I honestly do not have a good solution to this problem,
> especially if we are talking about Cross platform, its going to be some
> form of OpenGL and it will likely be difficult to get Call of Duty style
> graphics cross platform, though its not impossible, valve has shown it can
> be done, but its going to be a lot of work, Radegast is a good start to an
> alternative viewer however graphically it still can not compete with even
> the SL viewer in terms of render capability, however i suspect it does
> better than any other non LL viewer does at the moment, things like
> Windlight will be very difficult to replicate, even plain sky and clouds
> are very difficult to do well.   I am not sure I have seen any opensource /
> cross platform engines or systems that compare to the latest DirectX style
> game graphics, most of the open source cross platform stuff you see,
> including WebGL seem to be stuck in the 2005'ish era graphics at best.  I
> would love to see any examples anyone has that would prove me wrong there.
>
>
> On Tue, Aug 5, 2014 at 7:15 PM, Diva Canto <diva at metaverseink.com> wrote:
>
>>  Hi Mister Blue,
>>
>> You ask a great question.
>> Here are some of the things I would like to see in a future virtual world
>> viewer:
>>
>> #1 - Open source -- goes without saying
>>
>> #2 - Multi-platform -- it should run on the Web, standalone, and tablets.
>> JavaScript maybe? Or something that compiles to it.
>>
>> #3 - VR-enabled -- it should be ready to run on the Oculus and other VR
>> displays and inputs
>>
>> #4 - Modular with a small core -- the core viewer should be as simple as
>> possible, possibly just a non-interactive *viewer* without avatars, and all
>> features (such as avatars, chat, build, etc) should be added as plugable,
>> optional modules with corresponding server-side support. In fact, the
>> viewer-side modules should be downloaded from the server-side as people
>> visit the server (a la JavaScript on the Web) -- that's the maximum
>> flexibility, as it allows for all kinds of additional features and
>> differentiation. Voice chat is an important feature for many, but should
>> also be optional.
>>
>> #5 - Modern graphics -- I don't really know what this entails, but it
>> would be nice to have environments and avatars like those in Call of Duty.
>> This may be at odds with #2.
>>
>> I'm not sure if this requires a whole new server side or if we can use
>> OpenSim for a while. People have added WebSockets support, and that's ripe
>> for being improved.
>>
>> If you start this project, I hope I have time to help! I would like that
>> very much.
>>
>>
>>
>> On 8/4/2014 6:08 PM, Mister Blue wrote:
>>
>> I've been seriously thinking about creating a new viewer project. But
>> with Maria's survey[1] and all the different virtual world announcements
>> happening and the lively OpenSimulator vs Unity discussion on the
>> opensim-user list, I'd like to open up the discussion to the OpenSimulator
>> development community[2].
>>
>>  LL is making a new world and viewer (closed and proprietary as far as I
>> know). HighFidelity is making a new world and viewer (open source with
>> commercial 'grid services'). Unity 5 will have a multi-platform web viewer
>> and a new multi-user backend (pricing not announced). There are various
>> wonderful open source virtual world projects (realXtend, Virtual World
>> Framework, ...) as well as many emerging technologies (xml3d, html5,
>> asm.js). There are several forks of OpenSimulator (Aurora, ArribaSim, ...)
>> that have made many enhancements to the base system. Add to that several
>> successful virtual worlds build on OpenSimulator (inWorldz, Anvination,
>> Kitely, ...) as well as a few packaged distributions of OpenSimulator
>> (Diva, ...). Add to that the many SL/OpenSim third party viewers based on
>> the LL viewer. And that's just in our sphere. There is much more happening
>> in the gaming and education and training communities.
>>
>>  So, what could a new viewer add to the mix? Whatever a new view does,
>> it approaches the question of what OpenSimulator should be three years from
>> now.
>>
>>  For me, there are two basic choices[3]: evolutionary change or
>> revolutionary change.
>>
>>  Evolutionary change says to build on existing OpenSimulator. Make due
>> with the existing LLLP (Linden Lab Legacy Protocol) and improve vehicles,
>> make installation and use easier and add an easier to use and improve
>> viewer.
>>
>>  Revolutionary change would be striking off on a new virtual world
>> architecture. It could have LLLP support but only for downward
>> compatibility and to keep that community and content. But other questions
>> arise: How would one build a viewer/virtual world where a HiFi avatar could
>> stand next to a SL avatar? How could content be delivered to a viewer so it
>> is displayable but is not in the original, copyable form? How to leverage
>> the distribution and power of 'the cloud'[4]? How could one make
>> hypergridable grids across the many virtual worlds?
>>
>>  What do you  think? Evolution or revolution?
>>
>>  -- mb
>>
>>  [1]
>> http://www.hypergridbusiness.com/2014/08/survey-better-vehicles-search-most-wanted-in-opensim/
>>  [2] I see this expanding to other forums eventually.
>>  [3] With a lot of gray area in between
>> [4] I'm always amazed by Google Maps -- it displays the map with traffic
>> overlays and I can zoom in and out with almost immediate response. Why
>> can't a virtual world viewer do that?
>>
>>
>>  _______________________________________________
>> Opensim-dev mailing listOpensim-dev at opensimulator.orghttp://opensimulator.org/cgi-bin/mailman/listinfo/opensim-dev
>>
>>
>>
>> _______________________________________________
>> Opensim-dev mailing list
>> Opensim-dev at opensimulator.org
>> http://opensimulator.org/cgi-bin/mailman/listinfo/opensim-dev
>>
>>
>
>
> --
> Michael Emory Cerquoni
>
> _______________________________________________
> Opensim-dev mailing list
> Opensim-dev at opensimulator.org
> http://opensimulator.org/cgi-bin/mailman/listinfo/opensim-dev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://opensimulator.org/pipermail/opensim-dev/attachments/20140805/f01a2985/attachment-0001.html>