'Re: [Opensim-users] what version of collada is expected by various standard viewers & opensim'

[prev in list] [next in list] [prev in thread] [next in thread] 

List:       opensim-users
Subject:    Re: [Opensim-users] what version of collada is expected by various standard viewers & opensim
From:       Dr Ramesh Ramloll <r.ramloll () gmail ! com>
Date:       2013-07-06 14:18:16
Message-ID: CAHi41xJjCsQn8fDf4HybQOy844k_MxzO2inDXKtKrmcOz-bW8Q () mail ! gmail ! com
[Download RAW message or body]

Thanks John, Dahlia,
I think I get it, I cannot obviously model the situation in my head
because of many factors that am sure am missing. But overall, I think
it is like a raw data compression issue for optimal its optimal
transmission down a constrained pipe. So in this case, mesh is
analogous to raw data. The optimizing process when done manually will
look like cutting down intelligently a large mesh into smaller
repeatable parts and send each part down the pipe and instructions to
reassemble them at client. Because the reassembly is cheaper than if
the whole mesh object were to be pushed down. What users may see is a
complex world appearing more quickly. Thanks again for the
explanations.
R

On Fri, Jul 5, 2013 at 11:33 PM, Dahlia Trimble <dahliatrimble@gmail.com> wrote:
> you cant really compare number of "object downloads" to number of "mesh
> downloads" as meshes can grow very large in size and memory requirements
> where object downloads are very small and fast. There's also a negative
> impact in simulator physics and memory when using very complex or large
> meshes. Of course if your meshes are phantom then sim impact will be less.
>
>
> On Fri, Jul 5, 2013 at 7:59 PM, John Sheridan <john@pseudospace.net> wrote:
>>
>> Dr. Ramloll,
>>
>> You would probably get a better benefit if you were to upload a few
>> different variations.  For example: a single plant, a 10x10 patch of joined
>> plants, a 20x20 patch, and so on.  Each of the three variations if
>> duplicated in world would only be downloaded once and you would have the
>> flexibility to cover a variety of different areas in different sizes more
>> efficiently.
>>
>> In regard to prim counts, if you were to upload a single plant and then
>> build your patch in world you would likely be using far more prims than you
>> would if you uploaded a pre-joined patch.  So, not counting for vertex
>> counts and all the other goodies that come from a land impact calculation -
>> the following would hold true.
>>
>> 1 plant    = 1 prim     = 1 mesh download, 1 object download
>> 10 plants (linked in world) = 10 prims = 1 mesh download, 10 object
>> downloads
>> 10 linked plant patches = 100 prims = 1 mesh download, 100 object
>> downloads
>> 1 patch (10 plants joined in mesh before import) = 1 prim = 1 mesh
>> download, 1 object download
>> 10 mesh patches = 10 prims = 1 mesh download, 10 object downloads
>>
>> The theory being that the mesh download is much larger than the object
>> download.  The mesh download contains the mesh's geometry, textures, texture
>> maps, and so on wherein the object download contains such things as the
>> object's position in world, the id of the mesh, textures, and so on that are
>> applied to it.
>>
>> I hope this helps.  :)
>>
>>
>>  - John / Orion Pseudo
>>
>> On 07/05/2013 10:07 PM, Dr Ramesh Ramloll wrote:
>>>
>>> Hi Dahlia,
>>> If I understand you correctly, if I upload a grass plant in mesh and
>>> then just multiple copy it inworld, it would not make a difference if
>>> made a grass patch in mesh and then uploaded it. Let me know, am kind
>>> of dabbling with this issue right now. Am thinking in the way for
>>> leaves in a tree.
>>> R
>>>
>>> On Fri, Jul 5, 2013 at 9:43 PM, Dahlia Trimble <dahliatrimble@gmail.com>
>>> wrote:
>>>>
>>>> If an object is made up of many pieces, all of which are the same mesh,
>>>> then
>>>> it would be better to upload the mesh once and just duplicate it. This
>>>> is
>>>> because the mesh is a single asset and the viewer *should* only download
>>>> it
>>>> once. However, each piece is still a single "prim" whether it's a prim,
>>>> a
>>>> sculpt, or a mesh, and will still require a message from the sim
>>>> (ObjectUpdate packet) telling the viewer that it is there. This message
>>>> is
>>>> quite small compared to the size of the mesh asset though and many of
>>>> them
>>>> can be sent quite quickly. If the same mesh was uploaded many times and
>>>> these different uploads were used in the same scene, then each mesh
>>>> would
>>>> get a different Asset ID and the viewers would have no way of knowing
>>>> they
>>>> were all the same and would need to download each copy separately; this
>>>> could put quite a strain on the simulator and the network and should be
>>>> avoided.
>>>>
>>>> Note that in Second Life, "Land Impact" may be adversely influenced by
>>>> many
>>>> copies of the same mesh *even though the mesh only has to be sent to
>>>> each
>>>> viewer once*. This is an oversight in the design of Land Impact and
>>>> OpenSimulator does not have this issue as it does not implement Land
>>>> Impact.
>>>>
>>>>
>>>> On Fri, Jul 5, 2013 at 5:49 PM, John Sheridan <john@pseudospace.net>
>>>> wrote:
>>>>>
>>>>> re : "The various editors I tried require a lot of gymnastics just to
>>>>> get
>>>>>
>>>>> objects that render well in them to be imported into opensim.
>>>>> I am only familiar with ac3d, and Blender somewhat."
>>>>>
>>>>> I've been working with Blender 2.6, and while I'll admit it takes a bit
>>>>> of
>>>>> adjusting if you're moving from prim based modeling it works perfectly
>>>>> with
>>>>> opensim uploading through either Firestorm or Kokua.  Scaling is a bit
>>>>> of a
>>>>> headache but after upload my models show exactly as they did in Blender
>>>>> and
>>>>> oddly enough, they seem to actually load a bit faster.
>>>>>
>>>>> While on the topic of mesh, I did have a quick question to anyone out
>>>>> there regarding instancing.  Awhile back (before the advent of mesh) I
>>>>> seem
>>>>> to remember someone mentioning that sculpts were superior to normal
>>>>> prims as
>>>>> only one copy of the sculpt is downloaded and then applied to all
>>>>> objects it
>>>>> was used on by the viewer.  Would anyone happen to know if the same
>>>>> holds
>>>>> true for mesh?  Say in the case of an model with multiple duplicated
>>>>> pieces,
>>>>> would it make better sense to include one copy of the duplicate piece
>>>>> in the
>>>>> dae and then duplicate it in world or just glob it all together during
>>>>> the
>>>>> import?
>>>>>
>>>>> Thanks, :)
>>>>>
>>>>> - John / Orion Pseudo
>>>>>
>>>>>
>>>>> On 07/05/2013 08:08 PM, Nebadon Izumi wrote:
>>>>>>
>>>>>> The various editors I tried require a lot of gymnastics just to get
>>>>>> objects that render well in them to be imported into opensim.
>>>>>> I am only familiar with ac3d, and Blender somewhat.
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Opensim-users mailing list
>>>>> Opensim-users@lists.berlios.de
>>>>> https://lists.berlios.de/mailman/listinfo/opensim-users
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> Opensim-users mailing list
>>>> Opensim-users@lists.berlios.de
>>>> https://lists.berlios.de/mailman/listinfo/opensim-users
>>>
>>>
>>>
>>
>> _______________________________________________
>> Opensim-users mailing list
>> Opensim-users@lists.berlios.de
>> https://lists.berlios.de/mailman/listinfo/opensim-users
>
>
>
> _______________________________________________
> Opensim-users mailing list
> Opensim-users@lists.berlios.de
> https://lists.berlios.de/mailman/listinfo/opensim-users

-- 
'Consider how the lilies grow. They do not labor or spin.'
Rameshsharma Ramloll PhD, CEO CTO DeepSemaphore LLC, Affiliate
Research Associate Professor, Idaho State University, Pocatello, ID
83209 Tel: 208-240-0040
Blog, LinkedIn, DeepSemaphore LLC, Google+ profile
_______________________________________________
Opensim-users mailing list
Opensim-users@lists.berlios.de
https://lists.berlios.de/mailman/listinfo/opensim-users
[prev in list] [next in list] [prev in thread] [next in thread] 