Talk:Tarkin

Vaporware?

Maybe I'm missing something, but as far as I can tell:

• There is no encoder or decoder implementation that's remotely suitable for practical use.
• There is no bistream specification, in whole or in part.
• There isn't even any significant amount of preliminary discussion (that I can find) of doing either of the above. All we have is a wishlist of what toolset Tarkin is supposed to use when it's done.
• There is no person, or group of people, who are remedying these problems.
• The extremely minimal work that has been done (see http://trac.xiph.org/browser/trunk/tarkin ) hasn't even been touched in over 5 years. And there was no flurry of activity followed by a cessation of activity. There's just an initial check-in of code, then, no further work.
• Judging by the near-complete lack of information about Tarkin on xiph.org's main website, you'd be at pains to prove that a Tarkin project really exists (and by all evidence, it doesn't).

I think if a commercial company announced an exciting new codec six years ago, and six years of absolute silence followed, we'd assume it was vaporware, and this article may have even been deleted.

You might be tempted to say that developing Tarkin takes a long time, but snow popped up out of nowhere in Summer 2004 from the work of one man, and within a year of the first cvs commit, it was both fast enough to be (somewhat) usable, and good enough quality to be competitive. Nowadays, while still experimental, the speed and quality results are roughly in line with something that'd be suitable for everyday use.

At a minimum, I think the article should take more pains to point out that the tiny amount of abandoned work on Tarkin barely even qualifies as "preliminary," and there is a serious chance nothing will ever come of it. Instead, we have a single sentence mentioning that it's "on hold" buth otherwise I think the reader would be lead to believe we'll see Tarkin any year now. Snacky 01:25, 20 August 2006 (UTC)

Complete speculation but I would say Dirac (codec) or perhaps snow (Dirac is probably more likely as it appears to have a greater amount of support and it also appears to be patent unencumbered (BBC checked) whereas the patent status of snow is largely unknown) would probably end up being adopted by Xiph if either one succeeds and Tarkin will just end up dying and/or Dirac/Snow will also be called Tarkin. BTW, it is interesting and perhaps good that the free software movement appears to be working on wavelet video compression before we really have anything (AFAIK) commercial. (There are some commercial implementations of wavelet video compression but none of them are intended for the low bitrate, high quality broadcast and internet transmission market which snow and dirac and tarkin are intended for) Nil Einne 17:02, 27 October 2006 (UTC)

If nothing else, this page is of historical interest. Was Tarkin one of the first efforts at a wavelet-based codec? 128.194.22.23 22:53, 27 January 2007 (UTC)

The main difference between tarkin and other Wavelet based compression is that other compressions are just something that can be though as MPEG-with -wavelet-instead-of-DCT (ie.: they use wavelet to compress separate frames, and classic technology such as motion vectors between frames, or no interframe compression at all like Pixlet) whereas Tarkin was an attempt to do 3D-Wavelets (ie.: using wavelet both for the 2 dimensions inside a frame, but also for the Time-dimension, between frames). Thus it is overly complex and, besides some discussions on the developpers mailing list and some experimental code, there is no concrete results. It is interesting from a historical point of view. But we won't see concrete application befor a long time, both because of complexity and processing power. (That's also why Snow managed to pop into existence much faster than Tarkin) DrYak 17:33, 28 January 2007 (UTC)

NN-codec?

what the hell is an NN-codec? And why is it grounds for potentially deleting this article? —The preceding unsigned comment was added by 210.84.0.40 (talk) 12:24, 28 January 2007 (UTC).

NN is Wikipedia shorthand for non-notable. Playstationman 14:56, 16 February 2007 (UTC)

Novel Approach

Sure, nothing came out of Tarkin but it began with an interesting idea and should not be forgotten. Perhaps this idea will stimulate others and lead to new strategies in codec design. "encoded as a unit with a 3-D discrete wavelet transform. This is in stark contrast to the more traditional method used in Theora and most other video codecs" —The preceding unsigned comment was added by 24.10.202.87 (talk) 10:03, 1 April 2007 (UTC).