Google Summer of Code 2007

Post by **sinbad** » Wed Feb 28, 2007 10:04 pm

Yes, I'm still very much pondering this, but the use of several SMs should still be considered very coarse-grained. Yes, the idea is to connect them, and at different points so that the same SM can be connected to another multiple times (e.g. cave system emerging into landscape at multiple points). Every SM will have a concept of top-level 'localities' - pretty large scale concept, with a local coordinate system (for addressing precision issues). I plan to come up with a referencing scheme so that with just a single locality reference you know how to find the SM that manages it and the locality in the SM - that's your top-level world segment so to speak. Within that the SM is still in charge. Localities will be connected by spatial relationships or connectors - I don't want to call them portals because that implies a certain type of structure - perhaps 'wormholes' would be an ok term

Localities as a whole will be paged in and out either by distance or connection or some other scheme - the top-level class controlling this. The 'master' won't be called a SceneManager, it needs to be something more grand than that - perhaps UniverseManager

One thing I'd like to do though is make this class not need to be altered in the main - which means some clever thinking about connectors between localities across potentially many scene managers without being restrictive.

Anyway, that's a bit of a brain dump, nothing is being written on this until Eihort is out the way and I get some pencil-chewing time. But if I do get the chance to take it as far as I'd like to, this is almost a total rethink of SceneManager, perhaps even enough to consider it Ogre 2.0

Post by **Kencho** » Wed Feb 28, 2007 10:13 pm

Sinbad, that looks definitely awesome!

Back to the SOC07 topic... when and how should mentors apply?

Post by **sinbad** » Wed Feb 28, 2007 11:05 pm

If we get accepted I'll post a call for mentors like I did last year.

Kentamanos · Post by **Kentamanos** » Wed Feb 28, 2007 11:13 pm

This definitely sounds awesome. I think this would be HUGE boost to Ogre's capabilities. I completely understand the need to get Eihort out of the door, but this is exciting future stuff.

Once the "Universe Manager" API is defined, someone will figure out a good data format for it (XML document possibly) and create loaders. Then tools will start coming out of the woodwork to support that format (like World Minion type applications). That's when things will really take off.

I agree that something like this would probably be significant enough to make it Ogre 2.0

.

Post by **Kencho** » Wed Feb 28, 2007 11:26 pm

sinbad wrote:If we get accepted I'll post a call for mentors like I did last year.

Thanks

I've re-read the last year's and noticed the dates were pretty different (May last year, March this year!)

Post by **Praetor** » Thu Mar 01, 2007 12:24 am

Yes, big schedule change. We should know soon if Ogre is accepted. Then we'll need to start student applications real quickly.

Post by **Kencho** » Thu Mar 01, 2007 12:37 am

I've left an announce in my university forums, as well. I think it's not known enough outside of these forums...

Grom · Post by **Grom** » Thu Mar 01, 2007 4:21 am

what other kinds of things could students apply to work on? After instancing and a cel-portal scenemanager, i can't think of what else OGRE is in desperate need of.

syedhs · Post by **syedhs** » Thu Mar 01, 2007 5:03 am

Sinbad,

Maybe you can give certain hint of SOC projects so that they are inline with whatever plan that you may have with Ogre. The hint could be in term of the big plan of what you are thinking for Ogre to be in future. I think there is nothing wrong with it, and the candidate will just happily narrowing down their applications scopes.

Post by **sinbad** » Thu Mar 01, 2007 10:42 am

http://www.ogre3d.org/wiki/index.php/HelpRequested

I'll be updating that list if we get accepted.

ShingoHill · Post by **ShingoHill** » Fri Mar 02, 2007 1:09 am

Hi all, I´m Andrés, currently studyng CS in Spain.
I´ve been reading the forums for some time and I´m interested in doing one of these proyects:

A camera system based on the idea presented here 3rd person camera system, but with many enhancements offering a complete solution. Understand it, I´ve been taught by Kencho, and now I´m obsessed with cameras as well.

Texture atlases support. A series of classes and a toolset to help using texture atlases with Ogre, doing it less traumatic for the artists and programmers.

What do you think about it? Feedback?

Post by **Kencho** » Fri Mar 02, 2007 1:23 am

Hahaha, well, someone interested by the announcement

Do you know of someone else who plans to join SOC07?

People seems to only show interest in these topics when you say it's paid...

Anyways, glad to see you here.

pekar · Post by **pekar** » Sun Mar 04, 2007 4:47 pm

I'm seriously thinking about applying. It seems like an excellent opportunity to gain experience in a field of choice.

I'm a 22 years old CS student from Belgium (yes, it does exist

) . I got interested in engine design and game programming through a school assignment (extensible game engine).

I read something about procedural content on the board of last year, but it didn't seem like somebody took the assignment last year. Seems like a shame to me, because it will become more relevant in the future, and amazing things can be accomplished.

I wondered if other people are already working on this, either in Ogre or in the open source community. What has already been done? How should it be done?

Post by **Kencho** » Sun Mar 04, 2007 10:34 pm

Welcome aboard!

I know some people has been working a bit on their way on that field, though nothing real has come out for Ogre itself. I also think this will be a killer in the future. Indeed, recently read in gamasutra a bit about it from the guys of... Darwinia? Nice stuff

Hope you choose us in case Ogre is accepted as an organization this year again

Welcome anyways

Grom · Post by **Grom** » Mon Mar 05, 2007 1:34 am

I would love to do something, but I have no idea what to do. What does OGRE even need anymore, it seems damn near perfect! I'm looking through the help requested list but nothing pops out at me

Post by **Praetor** » Mon Mar 05, 2007 1:44 am

When we get the notice that Ogre has been accepted as an organization this year (when, not if dammit!), the wiki page with some ideas for last year will be updated. You can also go into and read some of the stuff in the SoC forum to look for ideas. Look for extensions and new things that pop out.

pekar · Post by **pekar** » Mon Mar 05, 2007 3:06 am

Welcome aboard!

Thanks for the hospitality!

Indeed, recently read in gamasutra a bit about it from the guys of... Darwinia?

Fascinating article, thanks for mentioning it. I think I will always remember the words "Pointillist Renderer". Maybe something for a Summer of Code project?

Seriously though, there is a surprisingly honest development diary about Darwinia. It also covers the changes in art through development. I'm not sure, but I seem to remember the terrain was generated during development, and the meshes burned on the cd. IMHO, they had great style in the finished product, especially on the simple sprites.

Hope you choose us in case Ogre is accepted as an organization this year again

I'm thinking about applying to the one laptop per child project. Last year they were looking for a game developer to do a informational game about aids. But I have serious doubts about the attainability of the project, especially on the play-testing side of it (I'm guessing my chances are slim if they're not being manufactured).

On the other hand, procedural content seems like it's very interesting. There are just so many ways we can take this. I'm definite leaning towards Ogre, are there any real reasons you guys wouldn't get accepted as a mentoring organization?

Post by **Kencho** » Mon Mar 05, 2007 3:10 am

I can't see why not either

Post by **pjcast** » Mon Mar 05, 2007 5:00 am

What would really be great, is if there could be a SoC project to enhance and extend Ogre Studio - since it is GPL now

Some procedural plugins would be cool, as would work on the scene editor, a sweet particul editor, or any other number of things that could be done/improved. But, I know it is .Net and limited to Windows atm, but I can hope anyway

My time has been devoted elsewhere, but I have not given up on the editor.

vexd · Post by **vexd** » Wed Mar 07, 2007 7:00 pm

Hi, I realize this is my first post here but I have been reading around here for the past couple of months. I am a 21 year old comp sci student at Edinburgh University in my final year considering organisations for the Google Summer of Code to apply for. The SOC proposals for this year seem really strong and ambitious which is good although it would appear that the suggestions of a good scene editor or a faster particle editor would indeed also be really useful.

Procedural content creation is a hot topic at present with Intorversion studios flouting their city generator for Multiwinia and many other games heading for texture and geometry generation in this manner. The creation of a framework to handle generic procedural algorithms would be kind of nice, leaving it open for users to extend as they see fit.
I am currently working as part of a team applying for the British Dare to be Digital game design competition which we are intending to use OGRE in if we get into the final selection. We had already considered creating such a framework to handle our cloud system with the possibilities of dynamic alteration at runtime to reflect changing weather. This is of course just an example of how such a framework could be used, relying on the underlying randomisation algorithm to create connected trees parameterised with cloud related variables. These could cover things such as thickness/density, spread, depth (of tree), number of roots (for multiple discrete unconnected trees in one object).

Anyway I digress, I have recently finished my dissertation on hyperbolic geometry drawing in two dimensions so have had all manner of graphics and optimisation headaches but I would still like to work in this field if possible. If the Dare application falls through it would be ace if I could still apply myself to the OGRE project in some way, the summer of code would definitely be preferred though

.

Post by **sinbad** » Wed Mar 07, 2007 8:14 pm

For info the SoC application process is now open and I have of course applied on behalf of OGRE. We'll have to see if we get accepted now. I'm going to put a call out to potential mentors now.

MedO · Post by **MedO** » Mon Mar 12, 2007 5:26 pm

Hello, I'm a CS student, 21, and I found Ogre by looking over last years mentoring organisations. I'd love to be able to do a project here, because I like 3d/graphics programming, and have e.g. already done some basic experiments with procedural geometry / textures in the past (some even back in the day when I was still using QBasic, heh).

Just want to say I'm definitely interested. Hope you get accepted, even if I can't take part, because this project definietly looks like it deserves it.

Post by **sinbad** » Thu Mar 15, 2007 4:00 pm

I've updated our Ideas List now that we've been accepted for SoC 2007. I've added some things and refined others. Just to explain which ones I've picked from assaf's list;

1. DirectX 10.
2. Geometry shaders (both DirectX10 and OpenGL).

Yep, those are in the list (refined a bit)

3. Upgrade the particle system so it will be a GPU based particle system ï¿½ that uses shader (Look at the render monkey example - particle system ï¿½ to get the idea)

I've actually suggested that adding more threading as suggested in the Intel article might be a good project, and particle systems are a good target for that. The trouble with GPU particle systems is that they're nowhere near as flexible in terms of pluggable emitters & affectors etc so they'd be something of a gimmick. Better to put them on a second core and get the parallelism but the same feature support, IMO. Simulating the particle system on the GPU could be done as a separate custom MovableObject but it wouldn't integrate well at all with the exsiting particle system structures.

4. Improve the billboard class so it will not update the geometry every frame ï¿½ but will use a shader that will direct the billboard to the camera.

Yes, and this can already be done as a custom ParticleSystemRenderer fro example. You also have point rendering as an option already. But overall I think it's too small for a SoC project.

5. Finish the billboard clouds ï¿½ people will be able to use it.

Not added because it appears the 'interesting' work is done, it's just fixes and optimisations. Perhaps if the original author comes forward and has a lot of enhancements (ie new features) that could be added, but as a SoC project I don't think cleaning up someone elses stuff is that much fun

6. A GPU based font system - http://www.ogre3d.org/phpBB2/viewtopic.php?t=27646
[/quote]
Added.

7. A more advanced level of detail system

Already there

8. Convert ossimPlanet to OGRE so we can have a nice base to a "google earth" like terrain. http://www.ossim.org/OSSIM/ossimPlanet.html

I've just put a note at the bottom of the page about potential new Scenemanager classes. The issue here is that I already plan to revamp the SM in Shoggoth and it would be unfair to ask a student to work on a moving target and be dependent on me getting the new system stable to work on their own stuff. If someone really, really wants to work on a new SM and understands the risks that's fine but I actually don't want to put it on the list to entice people in and then make it harder than it should be for them. Depends on how my time goes as to when the new SM approach should be ready.

9. A View-Dependent Texture Mapping using Per-Pixel Visibility sample - http://www.ogre3d.org/phpBB2/viewtopic.php?t=26219

This is kinda nice but I'm not totally convinced about its practical use, so I've left it out. If someone came along and proposed it as their own project it could still be considered but I'm still a bit 'meh' about it compared to existing techniques that are available.

10. A basic infrastructure for 2D drawing in 3D ï¿½ this will be a good base of SVG, SWF, XAML support in the future ï¿½ it needs to be something similar to amanith.

Not sure about this one because OgreSWF is covering it.

11. A new render system called "broadcast render system" that will be used to transmit the render system data to the network - and a client that will get the transmitted data and re-render it. This can be used to render when using MS virtual PC or Remote desktop and in multi-screen scenarios.

Nice idea, but I think it's well outside our core mandate so I've excluded it.

Post by **tuan kuranes** » Thu Mar 15, 2007 4:57 pm

I would have some additionnal proposal and change to the WIKI helprequested page:

Changing Skeletal animation LOD section into "Mesh Animation"

Then as subsections :

- Skeletal animation control: As described in Adaptive Dynamics of Articulated Bodies that can be used in data driven simulation, data+user control, user control. Implement some physical controller samples as in animating virtual characters using motion data and simulation in a context dependant way in order to show simplification results in each configuration (data, simulation, data+simulation) (Students with good experience in Physic/Graphics/Animation, Hard subject, but papers tells it all.)

Now shamelessly taking ideas from cryengine2 (read this now), we could implement parts of it and add interface for ogre user to be able to plug its own physic engine (bullet, ode, physx, newton, user one) :

- Parametric Skeletal Animation: doesn't that look like an evolved cal3D ? Ogre can have at least Cal3D level of animation control. (Students with some experience in Graphics/Animation)

- Procedural Motion Warping: we could implement some procedural algorithms, like CCD-IK, analytic IK, example-based IK. Think of needs of nextgen controllers like Wiimote in terms of animation. If you control a hand using a wiimote, you need IK to make a whole animated character to behave "somewhat" correctly. Perhaps could even be applied to tree, foliage interaction. (Students with good experience in Physic/Graphics/Animation, but many, many IK papers on that...)

Btw, while you're at Crytek page, read everything as there's some nice ideas lies overthere :
- Animation Compression (save RAM and VRAM)
- Mesh Destruction
- Offline Rendering (movies and still to advertise projects)
etc...

Have a deep look at Crysis Specials Effects Video HDTV - WMV...

Post by **sinbad** » Thu Mar 15, 2007 6:12 pm

Ok, I've added to that, but I didn't understand your second point.