|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Why a new file format?There are a number of reasons for the creation of the new file format.
Why using ADV like file format?Making the AAV file format very similar to the ADV format used by the Astronomical Digital Video System (ADVS) was the most natural choice for me. Already having the source code to record in ADV format and the tools (Tangra3) to read and process an ADV file it was natural to make AAV simply an ADV file for 8 bit video data. With this however the state channel and system metadata had to be modified to fit better the usage of an analogue video camera and recorder that doesn't offer for example GPS coordinates of the observation. AAV ver.2 File FormatAAV version 2 is an extension of the ADV version 2 file format. In order to determine the AAV format version check the AAV-VERSION System Metadata (also called File Metadata) tag of the ADVv2 file. The information below describes how a processing software should handle AAV version 2 files. This image shows a typical set of File Metadata tags recorded by the latest version of OccuRec and how to interpret the most important tags is explained below. As AAV recordings are product of analogue video cameras there will be always a NATIVE-VIDEO-STANDARD tag. This should be used only for information purposes. The combination of NATIVE-FRAME-RATE and EFFECTIVE-FRAME-RATE tags can be used to calculate an integration rate. For example an EFFECTIVE-FRAME-RATE of 6.25 and a NATIVE-FRAME-RATE of 25.0 leads to calculating a x4 frame integration 25.0 / 6.25 = 4.00. This should be the integration rate used by the video reduction software for processing the AAV video recording. The next two very important tags are the OSD-FIRST-LINE and OSD-LAST-LINE indicating the first and the last line (from top to bottom) where the VTI-OSD timestamp is located. As AAV files represent stacked video frames with preserved timestamps and because there is no GPS timestamps recorded, the user will need to visually read the timestamps off the screen and enter them in the video reduction software. The processing software could either split the frame into odd and even video field or could only split the even and odd lines between OSD-FIRST-LINE and OSD-LAST-LINE. The image below shows the original OSD lines from an AAV frame and the lines when split into even and odd parts. ![]() By definition the timestamps in an AAV frame will be of the first and the last video field inside the stacked frame from an integration interval. In the example above with a 25 fps PAL camera 19:37:52.9522 is the end timestamp of the first video field and 19:37:53.0522 is the end timestamp of the last video fields. We know this because it is evident that IOTA-VTI has been used for timestamping. If a different video time inserter is used then the interpretation of the timestamp from the first and last video fields could be different. The actual interpretation is something for the end user to figure. In our case above the exposure started at 19:37:52.9322 and finished at 19:37:53.0522, which means it took 120 ms, which corresponds exactly to 4 PAL frames. The video reduction software should ask the user to enter a timestamp to be used for the integrated 120 ms frame and may guide the user in this process. It is also important to note that by design the very fist and very last frame inside an AAV file should always use no integration. Depending on how the file may have been additionally chopped however, this may not always be the case. The purpose of the no-integration frames is to determine the order of video fields inside the frame. Usually the top (odd) field should be the first one and should have the smaller timestamp. However some frame grabbers could have a defect and could combine video fields incorrectly. If this is the case then calculations from the split timestamps above may show that the frame duration was different than the expected one from the integration rate. This could be because the timestamp of the very first and very last video fields in the integration period may not correspond to the actual chronological first and last video fields. In those very rare cases the non-integrated frames can help identify this problem. For such videos there will be a timestamp bias of one video field. Individual AAV frames may have the following status channel values:
It is also worth noting that:
What are the values recorded by OccuRec in the Status Channel of the AAV (ver.1) file?
|