Ogre Forums

The question I'm thinking about is: Do you even need history for the assets?
Well, I'm not a modeler so I might be wrong, but if I do not overlook something, there are only two types of changes which occur:

• a "minor" change, like changing scale a bit, changing the facing direction or fixing a glitch in the animation etc.
  -> you do not really care about the old version*
• a "bigger" change, like made a hi/low-poly version of the mesh, changed sinbads textures to look more realistic etc.
  -> you probably would want to treat it as a new model, as it serves different needs

Is it (easily) possible to strip unneeded history of a svn/hg/git repository?

*of course want to have some backup, i.e. a version known as "stable", but you do not need a complete history

We could probably do without history for art assets, but the problem is that storing them outside of the repository complicates the process of building and trying Ogre for our users.
(And no, removing files from the repository history is probably not an easy thing to do.)

Ok I see, we should spit this thread up. I was thinking about the rapid prototyping repository/collection, but yeah my post is relevant for the current assets, too.

@CABAListic you are right it is annoying enough to download the dependencies separately.( not that i would know a better solution)
I just looked at post-clone hooks. They would make it possible to separate the assets but still get them automatically downloaded. Too bad they do not get cloned.

About the problem of having more than one package stored outside the repository, what about make some kind of program like QtSDK downloader / Android SDK Manager? you download the tool, and you can select what kind of packages you want to use: prebuilt-sdks, source sdks, dependencies, sdk-art, prototyping-arg, hd demo...

On linux I would go with a script that invokes a wget or something, maybe automate it in the make/cmake script.
I recall that they are using this method on Makehuman too, because the art assets started taking up too much space too.
Here's what they did: http://makehuman.blogspot.com/2011/08/i ... -debs.html

Of course, you might argue whether we really want to bother devs with such a more complicated checkout procedure (even though it goes automatically, there is a larger chance of something not working).

Working without history of art is possible, though it's risky. You will need to do profound testing before overwriting a file, to be sure nothing is broken. For a texture this might be fairly trivial, but I'm thinking in terms of, for example, changing something in the armature of a model, which at first sight seems to work well. Only later you discover that some of the skeletal animations that already existed for the model are wrong, or an application attached something to a certain bone, that now is gone or in the wrong place. It can get quite tricky, and changing it back or fixing it could sometimes take a long time.
For meshes we could of course store versioned Ogre mesh XML files in svn, but we would also have to keep the source files (eg. Blender files) if we want to change something later on (it's a pity you can't convert back and forth). Blend files are binary..

Maybe just make a tempory repository that can be deleted when everything is correct. When things are approved then commit them to ogre repository that will save bloating.

I agree that the Ogre source should only contain a few prototyping resources, and improving those (eg. the changes Kojack proposed at the start of this thread) along with maybe some new ones could be done just once. The resources could be improved and versioned somewhere else (like a temporary repository) and then included in the Ogre repo when they are finalized. This would incur only one increase in repository size (I have no idea whether that is acceptible or not).

A more elaborate and fancy demo, that is going to be separate anyway, would really benefit from a versioned art repository (either included in or separated from the demo source code). To promote contributions to this project the repo should be accessible by multiple people.

I've got some model info from Sinbad (the person, not the mesh). It's also got some cool historical info on ogre.

knot.mesh was 'created' (as in, I used the tool palette and tweaked stuff) by me in 2000 with 3DS Max. It was the first mesh ever rendered in Ogre
jaiqua.mesh was included in XSI v3 and was free for non-commercial use
penguin.mesh came from Zi (omniter), but I don't know whether he created it or found it
barrel and woodpallet were created by Nfz and I believe he said they were public domain (might be able to find a thread for that)
razor was created by Adrian Cearnau, the first developer on Ogre apart from me, no explicit license except permission to use in Ogre demos
rzr-002 was created by Dennis Verbeek, no explicit license except permission to use in Ogre demos
fish was a mash-up by me (my only bit of animation!). Started with http://toucan.web.infoseek.co.jp/3DCG/3 ... delsE.html and modified, these are free for commercial use and derivation
I can't remember where I got tudorhouse, I think it was a free model on TurboSquid.

Nfz has been gone for 4 years, so it's probably hard to verify his ones. But barrel and woodpallet would be very easy to remake. (Barrel is used in the shading system demo, woodpallet is used in the instancing and deferred shading demos)
I'll do some snooping on turbosquid, see if I can spot the tudorhouse. Or I can email some irrlicht people, see if they know (since irredit uses it too).
Razor and rzr-002 should be considered non commercial use only I guess, since they were allowed for ogre demos.

Is this it?
http://www.the3dstudio.com/product_deta ... oduct=8152
Here it is on turbosquid: http://www.turbosquid.com/3d-models/bui ... 3ds/272296

After some more comparison I think there are some differences, but I think it's from the same series. The "fw43" resembles the name of the texture of the ogre model (fw12) closely. I have not found the fw12 model origin, though I think we can assume it's the same author and it has the same license (being free for non-commercial use).

EDIT: I'm almost certain the source of the tudor house is: http://www.medievalworlds.com/
The model included in Ogre seems not to be available anymore.

If you search for "house" on turbosquid you can find lots of free (medieval) houses by the way (for those interested in quickly prototyping some medieval town).

It's fascinating indeed that by tracking the origins of Ogre resources you track a part of its history.

would there be any interest in me bringing in this little boat dock? I had it for another project, and just made it a bit more game friendly, moved all the planks to one texture (512 x 256) including the endgrain textures. Its not a complicated model at all, but I did at least try building it with structural integrity in mind, lol.

yeah I think so as a port was been discussed for a pirate ship to be docked at.

shadowfeign wrote:would there be any interest in me bringing in this little boat dock?

I think this is a good example of good material. Anything that can be used more generally and is not too specific to one project.
Nice model, by the way

alright, I guess I'll clean it up a bit more and post some model files.

I cleaned it up a tiny bit, and did a small amount of simplifying geometry. Hope it's useful for someone. Use it however you want, attribute not required.

Thank you. Much appreciated!
I'll be sharing some of my stuff as soon as I have the time to clean them up and finish some of them.

Kojack wrote:Don't forget that none of this is actually approved by the ogre dev team. There has been a lot of concern about ogre getting bloated. Currently the media directory that comes with ogre is 58.4MB.

The dev team shouldn't be worried that an improved demo is created by the community. It doesn't have to be shipped with ogre. Neither do the current models / demos.

skyforger wrote: The dev team shouldn't be worried that an improved demo is created by the community. It doesn't have to be shipped with ogre. Neither do the current models / demos.

I think Kojack was considering the size of the mercurial repository and of the download when someone checks it out or downloads the sdk+samples. If it is not shipped with ogre, but as a standalone demo, there is indeed nothing for the team to worry about. Kojacks remark was more as a warning because the "improving sample resources" was getting mixed up with the "creating a nice demo" idea.

shadowfeign wrote:I cleaned it up a tiny bit, and did a small amount of simplifying geometry. Hope it's useful for someone. Use it however you want, attribute not required.

Hey, thanks for sharing this ;-)

Do you mind if I upload this on http://ogrebin.org ?
We're trying to set-up a repository for all Ogre-related binaries (http://www.ogre3d.org/forums/viewtopic.php?f=5&t=68442).

Alexiss wrote:Do you mind if I....

Go for it, use it however you want. Call it a personal lesson in humility. Sell it, put it in something commercial, claim it as your own, and I'll do my best not to cringe even if you make thousands off it(which I highly doubt). but to be honest, letting go of art is something I'm really gonna need to do once I start at a studio. Besides, it is after all, a pretty small piece of artwork, and with ogre released MIT now, I'm not sure I could feel right about giving anything less back. But that's me personally, and it takes a big leap, but in the end, I know it's the right decision.

Today I tried converting the Flick mesh to Blender.
I was fairly succesful at exporting the static model. She's a pretty useable high res mesh at 11000 polys, which makes her suitable as a protagonist or important character in a demo for higher end hardware. It was a good reason to try out the Blender 2.6 to Ogre exporter (as I've mostly used Blender 2.4 for exporting up till now)

I still need to fix up some materials, add transparency etc
Her rig cannot be used as it's a bit too complex (too many bones and vertexes with more than 4 bones), so I need to give her a new one. I quickly tested with one but it still needs a lot of tweaking and I'm not really a big fan of rigging.
So if anyone wants to help me with this..

She would be a nice subject for testing shaders on.
I'll post the model when it is ready.

That looks cool.

When I exported Sintel, she looked kind of undead with just standard vertex lighting. Great as a rendered model, but odd without tons of shaders.

I'd help with her rig, but I spent over an hour last night just trying to apply 4 bones to a cylinder in blender (half that time was trying to make the cylinder have more vertical segments). I gave up and watched tv instead.

Hey duststorm, I'd like to help with the rig, so post the .blend and I'll see what I can do.

Here it is.

http://www.mediafire.com/?xzxkwbc2hxa1lbn

The licensing and credits for the file are as follows:

This model was originally made by Denny Lindberg and is released under the
Creative Commons 3.0(Attribution-NonCommercial-ShareAlike 3.0 Unported) license.

Big thanks and appreciation to the author.

It's based on the Flick v.01.4 model that is downloadable here. My modified file is saved in Blender 2.6 and it works there (after some small tweaks).
I changed it around a bit to be (almost) exportable to Ogre.
What I did: merged all relevant parts of her to one mesh (currently I did not add insides of mouth, and glass lenses on eyes and compass). The objects I did not merge were added to layer 2. Note that eye lenses were originally parented by the eyeballs, redo this parent relationship to have them in the right places.
I removed the blender materials because the blender2ogre export script crashes with an error on the original materials. Blender materials are now default stubs. The submesh groups are kept intact. I did mock up some ogre materials for her quickly, they are included in a .material file in the zip. I however need to add transparency to hair and eyelashes, maybe eyelashes and hair can be merged in one submesh.
Originally all meshes were parented by the armature, I undid this relationship.
One thing I forgot to do: to simplify the rigging modifier. Select Flick and remove the Armature.000 modifier, be sure to untick "Multi modifier" checkbox of the Armature modifier above it and remove the "UpperBody" from the vertex group input field. We only want the armature applied once to the whole body for our purposes.

At the moment she is poseable using her original complex rigify rig. Lots of it is controlled using IK (Inverse Kinematics) bones, like her feet: there are no visible bones in the legs, just pose the position of her feet and the legs follow automatically.
Maybe you'll notice that she has 286 bones, even though you don't see all of them. Select the armature and in the armature properties you can make visible the other bone layers, which will reveal, among others, her leg bones.
286 bones is too much to be exported to Ogre, but you can remove all the non-deformational bones and still have an intact (but more difficult to pose) armature that does export to Ogre. You will still need to fix the hair, eyes and sword as they are not assigned to any bones, which will result in all their vertices being at the origin and thus not visible in Ogre.
To remove all non-deforming bones: make all the bone layers visible, then in the outliner go through the whole Flick_Armature tree, select each bone with the bone properties tab open. If Deform checkbox is unselected, delete the bone. I did this and was able to export her to Ogre. She had 156 bones which is a lot, but animating her worked.

A better idea, however, would be to rig her with a simpler skeleton. A good starting point would be the human meta-rig. I already added one to the blend file and it's in a separate layer. She is not yet posed to it, and still posed to the original rig. To detach her from the original rig you only need to remove the armature modifier. This is where the work starts I guess. I already tried automatically weighting her to the meta-rig, but there is a lot of manual weight painting work to be done.

Bone weights in blender are stored using vertex groups, you could delete all Flick's vertex groups to start rigging from scratch, or try and reuse some of them.

It would be cool, since she is so detailed, to improve the meta-rig with facial poses and mouth and eye movement. Like our Sinbad character. I'm however not experienced enough at all to be able to do that. I'm not even patient enough to weight paint the meta rig to her.

We must decide on quality of the demo. Will it be like

We have a nice particle system there. Oh, and notice the use of shader. Jagged edges? Well, it's only demo.

(current OGRE demos) or more like

Wooahhh! How they did that?!!! It looks like Crysis! But hey, it's 3 FPS only on my new PC!

(current PC games)? Of course we must provide graphic settings, but what about models polycount? Should we feature programmer art or only very good models. Mixing hi- and lowpoly models looks bad. Very bad. Also, should we use Hydrax & Caelum/SkyX?

Lots of very good models are on blendswap, but those needs converting. At least Blender-OGRE pipeline will be tested

And you know what? If we manage to place our call for content on blendernation.com as news, we can get a lot of good stuff.

Before we do an external call for content, there should first be a concept and it should be more concrete.

I think the purpose of the demo would be to be pretty high end, meaning that we can spend quite a lot of polygons on important objects. After all, Ogre already has technical demos so we don't need another one.
Disabling some shaders or effects will probably suffice to get it to run on lower end systems.
For those who still can't run it, you can always put a 1080p version of it on youtube or vimeo.

It being an über demo showcasing what ogre is capable of in a non-technical demo I think we should include projects like Caelum/SkyX, HydraX, PagedGeometry, RTSS, compositors, etc

Not only in-engine technologies, but also content-creation projects focused around Ogre could be used and advertised. I strongly advocate a Blender and Ogitor setup for doing it. That way, proper documentation and tutorials could also be created, because the current documentation is either non-existent or really out of date.

Getting useful resources from blendswap seems a good idea to me. We just need to document where everything came from and have a credits list and an overview of the licensing of each used asset.
Exporting static objects (most of them will be static) is really not that much work, unless you start with normal maps, ambient occlusion, and everything else you might want for a modern day asset.

A pirate scene would probably work very well with HydraX and SkyX, but I don't know if it is already possible to run those on OpenGL. In my opinion this demo should really be able to run at least on Win/Linux/MAC.
On the other hand, if SkyX/HydraX don't have openGL shaders yet, maybe this is a good opportunity to create some!

This is of course just my view on it.