Talk:OpenSceneGraph

From Wikipedia, the free encyclopedia

I have removed the advert notice, since the information presented here is consistent with the other pages of in the 3D graphics topic. Sometimes feature lists can sound like adverts, but they are really just an enumeration of standard features of the domain --Tiganu 16:17, 26 October 2006 (UTC)

I originally added the tag, and it was not so much out of concern about which features to list, rather, how the information is presented. There are quite some claims of high performance - notably in the very first sentence of the article - which would probably require (third-party) comparisons to other projects to justify inclusion in an encyclopaedic article; also, I'm a bit uncomfortable with wordings such as "...allowing terabyte database to be explored at a solid 60Hz" and "Hobby becomes Obsession". Finally, the article may be in a grey area regarding copyright - the content seems to have been largely lifted from here, here and here - which may be a problem, since the OSG wiki apparently does not specify a free-content licence (I may have missed something here). In any case, I think a rewrite from scratch would be the best idea for this article. Changed to {{tone}} tag for now; feel free to remove if you disagree. — Peter L <talk|contribs> 19:25, 26 October 2006 (UTC)

Proposal for fair rewriting of article, if current tone is to be maintained, write a comparison article such as Comparison of massively multiplayer online role-playing games, titled something like Comparison of 3D Scenegraph APIs looking into items from category "3D Scenegraph APIs": First define a Gold standard (test) based on a set of full featured serious games that exercise all claims in various ways, as these would require source code written by different unrelated people that can recompile for each system under test, in these circumstances an independant third party would be required to judge fairness of implementation, probably requiring that programming against each API be open-source software to allow inspection (even if API itself does not need to be open source). Second explain plausibility and mechanism by which each performance advantage is obtained. Third perform statistically valid test runs to exclude measurement errors. Fourth perform blind experiments to remove observer bias: each software implementation should look the exact same across all implementations, only performance, data size, and quality settings should be recorded against an anonymous implementation, and the software should not be optimised for specific test paths, but instead be generally suitable for random end-user control. Fifth maintain a list of other competing 3D engines that could have these comparable testable "implementations" (i.e. instrumented testable applications looking the same and doing the same things) to test against (such as entries from the category "3D Scenegraph APIs", including Irrlicht Engine, OGRE3D, etc...), and highlight that validity of any conclusion on test results are partial, pending full coverage of other APIs. Even then, over time, the test results are bound to evolve, so at least the tone of the article should reflect that current evidence (if any available, please quote) suggests that this entry has these characteristics. Fgenolini (talk) 14:45, 15 December 2007 (UTC)