Hi Chris,
a similar discussion was already started by me longer time ago. Check this
thread:
http://forum.openscenegraph.org/viewtopic.php?p=5006highlight=#5006
So the thing is, that in order to move from the way how its currently managed,
it requires a lot of power and time. And it seems that
Hi Harash,
yes, this errors are because of not matched osg version. So in order to run it,
you need to get current osg from svn.
Thank you for your help,
Art
--
Read this topic online here:
http://forum.openscenegraph.org/viewtopic.php?p=18681#18681
Hi J.P.,
thanks. Changes merged and submitted to svn. I've changed also FOUND_CUDART to
search also in lib64 path. You can check if it works for you.
cheers,
art
J.P. Delport wrote:
Hi Art,
for osgPPU to find OSG on Debian 64-bit I had to add /usr/local/lib64 to
FindOSG.cmake
Hi J.P.,
yes, I was afraid, that this won't help. I've changed now to look for lib64
before lib. I assume that on 64bit systems lib64 does not exists, and hence
cmake will not find 64bit version. On 64bit system it can exists and hence will
find that one first.
I wonder how to change cmake
Hi folks,
In couple of last days I made huge differences to the code of osgPPU. I would
like to ask you to test current svn before I am going for an official release
(this will be as soon as there is new stable OSG version 2.10 released).
Current code is built against osg v2.9.6.
I've added
.
It works exactly like Art Tevs says, visit the node and then add its
children to a queue. I suppose Art will have already his code
complete, but i post it for other interested readers (or just Art
could have another solution). Just a half hour of coding (ughhh), so
take it as it is: no more than
Hi Paul,
I am not that familar with customizing RenderStage, but it seems it would be
the way to go. However, I am not sure if using own RenderStages is that easy in
osg.
I solved my problem in a different way. So, this is how I do with osgPPU in a
new module called UnitBypassRepeat:
- use
Hi J.P.
No, before the execution the number of iteration is not known. However the
iteration number is known just right before rendering the subgraph. So in
programming language paradigm it is the same as for-loop with known number of
iterations.
In the future one can think on iterating until
Hi J.P., Paul
yes, exactly like Paul said, I know the number of iterations at cull time.
Chaning the number of iterations at rendering time would be either to write a
special RenderStage, as Paul propose, or to write a draw callback which will
iterate the whole execution. However, this seems to
Hi Hui,
have you tried to place an osg::Switch above osgPPU::Processor and use that
switch to disable processor?
Another way would be to set node mask for osgPPU::Processor to 0x0, this should
disable it.
Try this, and let us know if it helps ;)
cheers,
art
Hui wrote:
Hi,
I am using
Hi,
Hui wrote:
I try add a switch as _group and add processor inside. then
_gropu-setAllChildrenOff(); but then the rendering screen become black. But
it I call _group-setAllChildOn(); then rendering screen back to normal.
I tried it also here, and everything works fine. I can switch
Hi Paul,
yes, thanks, it seems this is the way to go.
cheers,
art
Paul Martz wrote:
Hi Art --
Art Tevs wrote:
However, currently any node visitor is going in depth-first-search (DFS)
manner. Does anybody has already tried to implement breadth-first-search
(BFS) in osg
Hi Paul,
ok, I think I have to reject my previous post and will have to say that BFS
with osg is not that easy as it sounds first.
The problem is, that whenever I do iterate over the child list and let accept()
a child the node visitor, I will end up in DFS. Because accept() means to call
Hi folks,
J.P: thanks, but it seems that this is a little bit different problem.
Hmm, it seems that state sorting prevents me of repeated rendering of several
things at the same time. For example: two nodes are being gathered by the
CullVisitor and placed into the stategraph for rendering.
I am using svn/trunk.
Ok, I found it, it calls TraversalOrderBin, not SceneGraphOrderRenderBin,
hence was not able to see it quickly.
Perfect, setting up the sort mode of the renderbin to TRAVERSAL_ORDER helps
to solve my problem. Very usefull thing, thanks!
art
robertosfield wrote:
Hi
Hi folks,
current svn/trunk version has this nice option for rendering the nodes in the
same order as they are traversed by a nodevisitor. This is very nice feature,
which makes my life almost simpler.
However, currently any node visitor is going in depth-first-search (DFS)
manner. Does
Hi Peter,
pp wrote:
Hi Cedric and thx for fast answer.
I replied in the forum already, but waiting for approval.
...
Wonder why your message was still not approved? Peter, please change your
profile according to our forum rules.
I see your email is from our Saarland University, so on
Hi folks,
I need to render a subgraph multiple times per frame. So, not the way how
osg::Sequence work, where it renders a subgraph for specified amount of frames,
but I want to render the same subgraph several times during one frame.
The graph will look for example like this one:
root
|
Hi guys,
thanks for the tipps. I also was thinking about overriding the traverse method
as JS said. So in deed this is what I need. A simpler thing would be also,
what Robert proposed, just to write own traverser.
However, in my case I will also have a circular dependency in the graph. So
Hi Paul, guys,
here is the plugin I am using in such cases. Just unpack the directory under
src/osgPlugins/ and add ADD_SUBDIRECTORY(frgba) into
src/osgPlugins/CMakeLists.txt
It compiles with current svn version. I haven't tested every case well, just
the float format should be well tested.
Hi Paul,
I am not sure how about PNGs or other formats, but in such cases I am using my
own format, which I call FRGBA.
I have written an osg plugin in order to handle files in this format. It just
supports simple gzip compression. It also supports 3D textures and mipmapping.
Also almost every
Hi Robert, Paul
I think having FBOs as a StateAttribute shouldn't be changed. In this way one
can use FBOs also without cameras, which is very appropriate for osgPPU for
example.
However I think we should change how they are handled by the cameras. Currently
there is a problem when mixing
As I said before:
multisampling is not a feature of osgPPU it is rendering related. The patch I
submited previously, has been included into osg. Hence multisampling should
work now. The HDR example can be easy rewritten to support multisampling as
described before in this thread.
cheers,
art
Hi Harash,
Subsampling/Resampling:
Yes, the UnitInResampleOut, do resample the input texture to the size you want
to have it. However it use the default hardware resampling algorithm, so
GL_LINEAR. From your question, I thought you are thinking of some lagorithm
wich do something better than
Hi Michael,
GMan wrote:
1 - render the scene to the framebuffer
Setup a camera with FRAME_BUFFER_OBJECT as rendering target. Then use the
texture you have added to the camera as input into your CUDA environment.
2 - copy the image to a PBO (pixel buffer object)
3 - map
Hi Sebastian,
seb wrote:
...
When I use Processor::dirtyUnitSubgraph(), or Unit::dirty(), nothing happens,
as you can see on the 1st pic, it has not resized my scene
Ok, it seems I have to write a patch for osgppu to allow those methods to
resize also all units. At least some global
Hi Thomas,
carnibirdy wrote:
Actually, it's not doing anything.
It's supposed to pass through 4 units...
UnitTexture-UnitInOut-UnitInOut-UnitInOut
... which does nothing!
Of course, this will do nothing. You do not forward the output of the pipleine
to the main screen. The way you
Hi Thomas
Nice to hear, that you get it worked. If there were some special steps you had
to make to install osgPPU, then let me know, maybe I put them to the website
and to the README file.
OK, now to your questions:
carnibirdy wrote:
I made a pipeline composed of three shaders.
...
I
Hi Sebastian,
I must admit, I have never tried to resize unit's in realtime, however the
functionality for this should be already implemented.
Try to mark the unit graph as dirty. This can be done with
Processor::dirtyUnitSubgraph(). I think this should do the trick. If not, then
try to mark
Hi Thomas,
have you read the README?
carnibirdy wrote:
I downloaded the sources svn co http://svn.tevs.eu/osgPPU/trunk osgPPU and
tried to use Cmake to install it but it appears that some files are missing
or I did something wrong.
Please give the error message, which you get when
Hi Josselin,
as J.P. already said, I think your problem is that single precision float
values are not enough to be precise for the tenth number behind the comma.
There is no OpenGL texture format which can support double textures. There
exists only 32bit single precision float format. Maybe
Hi folks,
The next hundred users are now registered on our forum. Here are current and
last statistics:
June, 2009:
1865 mailing list users.
508 forum users
Overlap of 143
Total amount of users: 2230 users
September, 2009:
1865 mailing list users,
601 forum users
Overlap of 159 users
Total
Hi Johannes,
Hmm, I think it is not possible to use an array of texture samplers in a
shader. An array of any other values, should be possible. However, I would be
glad to hear, if I am wrong.
Your limitation number is the maximal number of texture units supported by your
GPU.
If you really
Hi Eric,
there is a fresh paper about that, what you want to do:
http://wwwcg.in.tum.de/Research/Publications/TerrainRayCasting
Oh at least, I think this is what you want to do?
Cheers,
Art[/url]
--
Read this topic online here:
Hi Miika,
miika wrote:
Alright, thanks. At least I know what it's about now, it was driving me crazy
because I had no idea whether it was because of me or osgPPU or OSG itself. I
guess I'll have to try and hack together something that sort of circumvents
the more elegant structures of OSG
Hi Miika, J.P.,
ok, after hours of thinking and studying of OSG's source code, I have to admit,
that this would not be an easy task to solve that issue.
The problem is that, even if I am able to restore any FBO attribute, FBOs used
by cameras will break the whole thing down :(
Imagine this
Hi Bob,
huh, I would say it depends on your implementation. Currently the way you
describe it requires a volumetric rendering approach.
The thickness calculation seems to be the most tricky part. If you have your
volume data, then use osgVolume for that (I haven't tested it, so I do not know
Hello,
ryoumans wrote:
for the thickness it's actually a fixed data for the geometry (not really
thickness). It's only thickness in the analogue of the x-ray model. It's
geometry dependent data, and for starting out adding this data item and/or
having only the one value will do. I'm
Hi Miika,
OK, i was able to run this. Yes, I can see the issue you are speaking about.
And I think, I understand where the problem is...
To fix this problem, we can try to add firstProcessor as a child of
slaveCamera instead of root, as explained in my previous posts. However, this
Hi Anders,
the results you see are correct. The technique implemented here is a very
simple, faked SSAO. Hence, you do not have to expect much. It is here, just to
show, that everythign is possible. Users are encouraged to implement examples
on their own.
If I will get enough time, maybe I
Hi Steven,
I've tried assigning the attached texture (without running the callback) to a
quad in the scene and it is still blank.
If your RTT texture is black without any post processing, so maybe you forgot
to add any scene below your RTT camera in the scene graph? For example your
Hi Mika,
Is it possible that you provide me with a simple test case implementation to
test it. Because then maybe I would be able to find an answer how to correct it.
However, this causes another problem which I don't even begin to understand
(which may or may not be related to osgPPU).
Hi Asheer,
1) Shouldn't the CPU to GPU data transfer be done with bindBufferInWriteMode?
Oh, I think I maid it not really clear enough. Everytime your write something
to the buffer, you enable the buffer in write mode. So write to the buffer can
be done in CPU-GPU as also GPU-CPU mode.
Hi folks,
I am glad to announce, that our forum broke already the 500 users mark. There
was an unproportional raise of forum members after the last forum server crash
(forum was down for around two weeks). The number of users raised from 400 to
500 in around two weeks after the forum was
Hi Ascheer,
nice to hear, that everything works fine now.
In order to increase performance you could use the GL_BGRA format instead of
GL_RGBA, nVidias drivers seems to work better with in such cases.
Also if you do not require floats, you should use GL_UNSIGNED_BYTE as format
too, so that
Hi Ascheer,
you can take a look into osgPPU::Unit::drawImplementation. There is
PixelDataBufferObejct used to copy the data between GPU und objects attached
with PBO (in this case this could be a CPU data).
In general the use is like this:
// CPU to GPU
Code:
PixelDataBufferObject* pbo =
Hi folks,
Serge, you haven't confirmed the date yet. Is the second week (just one
weekend) in the September is ok for osgInEurope in your company?
Cedric Pinson wrote:
I cheked if there is an event in paris, but i did not find an event at
this period.
Serges can you confirm the date of the
Hi Brian,
yes, you need to disable the OpenGL color value clamping. This can be done by
this, I think:
Code:
// disable color clamping, because we want to work on real hdr values
osg::ClampColor* clamp = new osg::ClampColor();
clamp-setClampVertexColor(GL_FALSE);
Hi guys,
Yes, I wanted to first just change the hd controller, so to use the slap trick.
However, I wasn't able to find the same hd on any of the online stores. So this
wasn't an option any more. Therefor the only options left, were either to send
the disk to some data recovering company or to
Hi folks,
I am unhappy to say, but our server had today at night a total crash. None
of our virtual machines is accessable from the outside and hence forum do
also not work.
I appologize for such circumstances and hope, yes I really hope, that the
web-data doesn't get lost ;(
I installed
Hi Ryan,
Thank you for pointing this issue out!
Yes there is some kind of protection activated, so that the server doesn't get
overloaded by bots. I have disabled the feature now temporarily. If this will
some kind slow down the server, I would have to reenable it again.
Cheers,
Art
Hi Kim,
I know about this crash, however in all of my cases the crash appears in very
rare situation. The bad thing, I cannot reproduce it, hence I wasn't able to
find out why it crashes :(
If you could describe the situation when the crash appears, maybe I could write
then a patch to solve
Hi Paul,
so, you can take a look now into the glow example in the osgPPU's svn
repository. I've added a simple workaround to support depth test, when
rendering glowed objects. This will make sure, that glowed objet does not shine
through other objects.
regards,
art
PaulG wrote:
Hi,
Is
Hi Paul,
yes, there is a way. One has just to use depth valus of the scene when
rendering the glow effect. Should be pretty easy to add, however I was lazy and
did not added this functionality :(
I will take a look today (or in the next days), maybe I could change the glow
example to provide
Hi Vincent,
osgDB Plugins register them self to the db registry. If you take a look into
every plugin, there is some kind of proxy global object. The constructor of
this get initiated as soon as library is loaded, and in the constructor the
plugin register himself to the osgDB. Hence no magic
Hi Paul, Robert,
whenever I get cold, I do following:
- drink a lot of (water, juice, whatever, no alcohol!!!)
- drink black/green tea with honey
- eat lemons, this helps really
- then put a lot of clothes over you over night so that you start to sweat
- sweating means you increase your body
Hi Vincent,
Vincent.B wrote:
Okay, so I implement a myClass in the core, and I declare a static myClass
mycalss in my DLL.
so, I can call with the static implementation, the methods I want in the DLL
from my core.
Not really, you implement a static/global variable, which will then
Hi Paul,
as other Paul said, take a look into osgPPU. There I have an example showing
exactly what you need. The example is simple and very easy to understand. You
can take this as a starting point and then change it for your needs.
Cheers,
Art
--
Read this topic online here:
Hi Martin,
as in my case, I always do derive a non-model class from osg::Group and use it
as a base class of the non-model object. Deriving from Group gives you the
possibility to include the object somewhere in the scene graph if it is
required, so that the nodes to which it belongs can be
Hi Martin,
mgb_osg wrote:
Would visitors still apply to it as with any other node in the graph - I
don't see how that helps?
Of course they would, however nodes are not drawable things, nodes are there
just to organize the scene graph. I thought, you want to include something
Skylark wrote:
Hi Martin,
Why not just devote a bit in the nodemask for such elements? Then you
set the intersection visitor's traversal mask to ignore those elements.
const unsigned int NODEMASK_UNPICKABLE = 0x1000;// whatever
iv-setTraversalMask(~NODEMASK_UNPICKABLE);
That's
Hi folks,
Couple of days ago we have discussed about having a nice opportunity to show
screenshots or videos from different demos using OSG. This opportunity do
already exists and can be found on the Wiki-page of the OSG project.
However, I have also decided to put such a possibility to the
Hello dear osg-community,
we are looking for moderators for the OSG forum and would like to have one or
two additional moderators to include in our team. Number of users is growing
almost every day (current state 340 users) and in order to obey the
forum/mailing list etiquette the user's
Hi folks,
@Cedric, Serge:
Cedric Pinson wrote:
Hi,
I will not be able to make a talk the first week of september, it's the
only week i am not in Paris, murphy's law i guess ;)
Another idea about this meeting could be to synchronize it with free
software event, then we could use the
Hi Sergey,
if you like to see some nice demos created with OpenSceneGraph, then take a
look here:
1. http://www.tevs.eu/project_siggraph07b.html
Demo was created with osg 1.2 (very first release of osgPPU)
2. http://www.tevs.eu/project_i3d08.html
OSG 2.2 (osgPPU v0.2)
3.
Hi Sergey,
sorry, instead of the first one, you take a look better here:
http://www.tevs.eu/project_r4o.html
Is just a little bit different version with more nice renderings ;)
Cheers,
Art
--
Read this topic online here:
Hi Sergey,
Sergey Kurdakov wrote:
But overall idea was to impress others ( esp newbees on a large scale
)...
But isn't that a main aim of all the demos ;)
The demos showed actually new techniques of rendering. To understand exactly
what the new thing of them was, one has to take a look
Hello dear forum users,
in order to establish a nice etiquette in our community, we have decided to
suspend user accounts which do not correspond to forum's rules. The main reason
is that a lot of forum users don't have valid real names specified. The problem
in that is, that your posts are
This is an edited version of a previous post
Hi,
sorry, test continues, please ignore..
regards,
Art
--
Read this topic online here:
http://forum.openscenegraph.org/viewtopic.php?p=11723#11723
___
osg-users mailing list
Hi,
sorry, I have to test something again...
regards,
art
--
Read this topic online here:
http://forum.openscenegraph.org/viewtopic.php?p=11727#11727
___
osg-users mailing list
osg-users@lists.openscenegraph.org
Hi Ed,
as you can read in this Robert's thread:
http://forum.openscenegraph.org/viewtopic.php?t=2498
it does apply to everybody, don't matter if you on the forum or on the mailing
list side! Otherwise forum users are discriminated and we wouldn't like to
discriminate users just because they
Hi Robert, all
So, I just want to sum up, which next steps are needed in order to create a
system which should/might improve the etiquette quality in our community:
Having both a first name signature and the full name elsewhere
displayed on the post is the best of both of worlds. On the
Hi both,
J.P. Delport wrote:
Maybe Art can make the forum to mailing list converter add the details
(that are normally displayed in the forum) to the bottom of the emails?
Would this have any benefit?
I do not know, what I should extra put in the mail sent from the forum. Most of
the
Hi Robert,
robertosfield wrote:
In the real world almost all of us will have a family name and a
personal name and typically the personal name is used in conversation.
Different cultures or even different companies will have different
conventions on which names are put first. In the
mgb_osg wrote:
Except you can't edit it or create an account on it.
And the wiki link takes you to the front page of the site
Guys, please, don't go offtopic :) Yes, there is a wiki page, however a lot of
new users, which are active on the ML or forum, doesn't read the wikis well as
Actually the new feature, where the moderators has first to approve the very
first post of any new member should already keep the bad guys away. This
feature is already in action, since couple of days, as I have posted previously.
I am not a friend of using a lot of anti-spam things like
Hi folks,
I just want to fresh up this topic. I have experimented with the webcam and a
video server, and it worked. So there will be definitively some kind of
broadcasting of the meetings/conference to the www. And as in the another
thread of Jean-Sebastian about ffmpeg-streams, we could even
Hi Serge,
Serge Lages wrote:
Yeah we've been accepted to this new office, so it's OK, we'll move during
June. We'll have a meeting room with approximatively 20 to 30 seats and a LCD
screen (42), and maybe a projector if we ask for it.
So I think that there is no problem for us to host
Test message...
This message should be first filtered out and have been approved before
reaching the mailing list and being visible on the forum.
Please ignore...
--
Read this topic online here:
http://forum.openscenegraph.org/viewtopic.php?p=11131#11131
sorry folks,
please ignore this test message. I have almost completed the new anti-spam
fighting thing ;)
cheers
--
Read this topic online here:
http://forum.openscenegraph.org/viewtopic.php?p=11133#11133
___
osg-users mailing
Hi folks,
due to couple of spam intrusions in our forum any new user and any user with 0
posts, has first to be approved before his/her messages are posted.
As soon as a new user post something on the forum, admins/moderators will be
informed. His/her message is first not visible to the
Robert, yes, it happens that spammers just do register on forums to post
something there. Mailing lists doesn't look that attractive as forums, therefor
there are human controlled spam bots around. In almost 4 months of forum
existing there were 2 spam messages out of around 3000 usual
Hi folks,
if somebody of you is interested in the numbers, I have computed the overlap
between mailing list members and the forum users.
There are currently:
1825 mailing list users.
300 forum users
and 106 users, who are registered on both systems.
Of course this is just an overlap by
Hi,
yeah this are interesting news. To hook up to the thread about 200 forum users
from the last month, it would be interesting to know how much is the overlap
between forum and mailing list users.
Robert, could you still provide me with the list of the names (or emails) of
all mailing list
Oh, in deed, hmm, I didn't knew that there exists such a capability.
OK, I will then write a script which will check how much of the overlap there
is between mailing list and forum users.
As soon as I have the info, I let you all know. Roland, thank you again for
pointing me into the right
Hi Paul,
I've sticked your debugging tips in the forum. So hopefully people will get
read them first, before asking the same questions again and again ;)
cheers
--
Read this topic online here:
http://forum.openscenegraph.org/viewtopic.php?p=10391#10391
Hi Sukender, Serge,
@Serge:
Wow, thats great. I think if we didn't find anything else (i.e. university)
then it would be great to use your opportunities for hosting the event. So
eventually we have to wait until may, so that we then know if your company can
be a hoster for the event.
Hi folks,
yeah, somebody get finally through. From my logs I can not see anything
strange, hence if it was a bot he get through by the first try (which means
this is a almost perfect bot ;) ), or it was a human controlled bot.
Sorry, for that intrusion. The current spam filters are already
Hi guys,
organizing a course is maybe a good idea, but I suppose that this will
automaticly decrease the number of interested people, since they will need to
pay then. If Bob and Paul would like to have such a training course, then why
not. Of course we could split up the courses from the
Hi Robert, Sukender, Serge
@Robert, Sukender:
Actually Sukender meets also my expectations what I was thinking about the
mini-conference. Giving courses does sounds also cool, but I am not sure if we
will have people who are willing to give them. As for us, specially for me, I
do not have
Hi Sukender,
of course this would be also great. Universities have good possibilities for
hosting a conference. And even more, this would be a nice opportunity to show
OSG to young public :)
I would like to hear what Robert do think about it.
However if we decide to make just a user meeting,
Hello, dear osg-users,
so the vote is over and we have the winner city - with 8 votes of 27 Paris won
the race. OK, we now know a city where we would like to have next osg users
meeting/conference.
Venue:
I would like to ask users, who are living in Paris, if they know some nice
venue. I
Hi Guy,
hmm, maybe you confound the Config.h of osgPPU with the Config.h of OSG. These
are different things. Currently in osgPPU/Export.h the Config.h is not used at
all, hence it shouldn't be any problem when building the osgPPU. Including hte
osg/Config.h is also not required, because all
Hi Guy,
Guy wrote:
Forget it, sorry, I found the files :)
The OSG_GEN_INCLUDE is where the CMake generated the solution and the
files are there.
Yes, the files are there, where CMake generate them, when you build OSG.
OSG_DIR is a path, where osg can be found. It can be either an
Hi Guy,
Guy wrote:
Now everything works.
I actually needed the OSG_GEN_DIR since it is where the Config files
are.
You can checkout the newest version. I have removed everything which is not
needed for proper ppu build. There are now much less configuration cmake
variables, which
Hi folks,
this is a last reminder about the running poll of the user meeting/conference
city. Current state is:
Paris, France [ 8 ]
Marseile, France [ 0 ]
Strassbourg, France [ 1 ]
Prague, Czech Republic [ 2 ]
Genova, Italy [ 1 ]
Brussel, Belgium [ 2 ]
Zuerich, Switzerland [ 1 ]
Hi Paul,
hmm, ok I see the issue. However I have no clue how I should correct this or
better to say, there is almost no way to correct this, because the way how
forum posts and mails are handled is different.
Look, whenever a user is reply to a topic it means, he is replying to the main
post,
Hi Oleg,
so, I have now spend some time and I think I've corrected all the CMake issues.
The problem was, that I haven't tested the windows build already for a longer
time. And, my fault, I haven't even tried to build the cuda example on windows
before.
Current cmake setup was now tested with
Hi Daniel,
Have you took a look into this thread:
http://forum.openscenegraph.org/viewtopic.php?t=1778highlight=zbuffer ? (I do
not know if that was you or not, because the user names of the authors are
different)
So you have given two depth buffers in a texture, right? And you want to
Hi folks,
I do just want to inform you, that our forum has reached a mark of 200 users
(just one month after 100 users mark :) ).
My last knowledge about the number of subscription to the mailing list tells me
that couple of months ago there were 2000 mailing list subscriptions.
Of course we
101 - 200 of 481 matches
Mail list logo