YATTA

From Wikipedia, the free encyclopedia

This article is about the anime programming tool, for the pop song see Yatta

YATTA, Yet Another Telecide Tool for Anime, is a tool used to IVTC video, usually anime.

It is used by a number of groups to achieve high quality MPEG-4 encodes of anime and other material. It enables the user to perform a series of video cleanup tasks, including but not limited to inverse telecine overriding, freeze frame assistance, and sectioned filter application.

The primary focus of YATTA is to perform IVTC (inverse telecine), which is the act of eliminating "combing" (so-called artificial interlacing) on the video source caused by the 3:2 pulldown process and lowering the frame rate to that of film. In YATTA's case, this is accomplished in two steps: by first aligning telecined fields back into full frames using a "current, current, next, next, current" (ccnnc) pattern, then "decimating" (removing) the duplicate frame left by the second "next" match. In addition to pure IVTC, YATTA also allows the combination of IVTC and traditional deinterlacing for sources mixing telecined and truly interlaced material, enabling variable frame rate manipulation to create VFR Matroska video files.

The secondary focus is on source improvements through a variety of means. The simplest involves freeze framing to replace bad frames, and moves into more intricate sectioned filtering.

[edit] YATTA Interface

This article or section uses first-person or second-person inappropriately or excessively.
Please edit this article to use the more formal tone expected of an encyclopedia, per Wikipedia's style guidelines. (help, talk)

The main part of YATTA consists of the IVTC element. This is possibly the most important aspect and can be relatively time consuming. It has 2 key elements. The pattern component, and the threshold components.

Now you've already created your yap project (via ymc), and are ready to roll. The first step, if you've never dealt with the source in question before, would be to jump around the source and get a general idea of the VMetrics that you'll encounter. This is to acquire a VMetric number which will rest just above the peak normal scene value to allow smooth playback, but stop on critically wrong frames (scene changes, stray interlacing, chroma blending).

Once you've determined an appropriate value, place it into the VSearch window. Using abosolute value is usually the simplest way to go through the source, and the rest are for more specialized applications. At this point you want to begin playing through the source. Your key viewing points are: Mouth combs Scene change combs (prior to) Small movement combs

All of these are combs, which can be incorrectly matched by telecide naturally, and are what you need to visually pay attention to. The vsearch metric isn't going to stop the video on these frames, as they are generally very low as far as Vmetric is concerned (which is why they aren't matched correctly in the first place).

Now you will find it will also stop on frames which are non IVTC correctable via the switch interface, these situations are covered later.