A few weeks ago, I put together a “practice” scene selection menu just to see how it was done. Specifically, I wanted to create a menu with the ability to either play an entire title or select individual chapters within that title. At the end of each individually selected chapter, the display would return to the menu.
There were several excellent “how-to” articles on this subject in the archives and, following these articles; I found it fairly easy to create the desired menu.
There was, however, a Joker in the deck. I had a lot of trouble with so-called flash frames. That is, after carefully setting chapter markers and in and out points, I would get random instances of material from adjacent scenes showing up at either the beginning or end of a chosen scene.
This was particularly crazy making because my title was nothing more than a slide show created in Vegas 5. There were no transitions between events and it was a simple matter to identify the precise timecodes where one event stopped and the next event started.
As I struggled with my flash frame problem, there were several posts on this Forum raising DVDA timecode issues and suggesting that these issues could be could be related to the mpeg encoding process. After some research, I believe that there is a connection here and it was the cause of my problems.
I ran my title through an mpeg stream analyzer. The results showed that none of the chapter points I set in Vegas corresponded to the location of an I-frame in the rendered title. In fact, every chapter point was somewhere within a 15 frame Group of Pictures and always corresponded to a B-frame. Specifically, the chapter point would fall on the one B-frame in the GOP that displayed information from both chapters.
If the results of this one test are universally true, it would mean that most, if not all, DVDA in and out points will not correspond to the start of a GOP. Therefore, on playback, the mpeg decoder will start decoding at some point within a GOP. I believe that this will cause flash frames to be displayed. This condition will be particularly noticeable if the chapter points fall between very different scenes.
Now, as I’ve said before, a little knowledge is a very dangerous thing and, on this subject, I have as small a portion of understanding as anyone. If my analysis is factualy incorrect, I would really like to know.
Jan
There were several excellent “how-to” articles on this subject in the archives and, following these articles; I found it fairly easy to create the desired menu.
There was, however, a Joker in the deck. I had a lot of trouble with so-called flash frames. That is, after carefully setting chapter markers and in and out points, I would get random instances of material from adjacent scenes showing up at either the beginning or end of a chosen scene.
This was particularly crazy making because my title was nothing more than a slide show created in Vegas 5. There were no transitions between events and it was a simple matter to identify the precise timecodes where one event stopped and the next event started.
As I struggled with my flash frame problem, there were several posts on this Forum raising DVDA timecode issues and suggesting that these issues could be could be related to the mpeg encoding process. After some research, I believe that there is a connection here and it was the cause of my problems.
I ran my title through an mpeg stream analyzer. The results showed that none of the chapter points I set in Vegas corresponded to the location of an I-frame in the rendered title. In fact, every chapter point was somewhere within a 15 frame Group of Pictures and always corresponded to a B-frame. Specifically, the chapter point would fall on the one B-frame in the GOP that displayed information from both chapters.
If the results of this one test are universally true, it would mean that most, if not all, DVDA in and out points will not correspond to the start of a GOP. Therefore, on playback, the mpeg decoder will start decoding at some point within a GOP. I believe that this will cause flash frames to be displayed. This condition will be particularly noticeable if the chapter points fall between very different scenes.
Now, as I’ve said before, a little knowledge is a very dangerous thing and, on this subject, I have as small a portion of understanding as anyone. If my analysis is factualy incorrect, I would really like to know.
Jan