Did you try the new jap.mp4 video?
For anyone else that may be interested, I have just learned that the PS3 Sequential Frame format in the above linked video uses the same standard frame packing SEI (45) as x264 does. The only difference is the PS3 stream has the SEI before every frame (with current_frame_is_frame0 flag set for every other frame) while x264 only inserts the message once for each GOP.
Yes, very correct in pointing out the state of the art which was described in the previous pages.
Rather than waiting tough, in order to move forward, I would edit the file as you suggested.
If I had the technical skills, to parse the container and the stream and to see what needs to be amended/added and here. I would do it because if the patched video file works, we would have found the algorithm.
If we had the algorithm, even with my limited programming skills, I would be able to write a program to convert files to the PS3 video format.
An inefficient program but it would do.
Mathew is confident that he can create files that are very similar to the Sony one and is confident that he can identify the differences. That process should yield the precise conversion algorithm.
The prize is huge, really: being able to have the PS3 to play any 3D video in stereo natively, rather than relying on very expensive TVs.
The reason why I say the prize is huge is because HDMI 1.4 displays are truly inexpensive now. I was giving the example of my setup where the VIP Gamer + old-school short through HD DLP projector have a cost which is multiples less than a multi-format 3D TV. Apart from big cost savings, there other benefits of playing S3D natively on the PS3 from USB: a display size as big as a wall, you mount it in a minute and, after watching the S3D video, you can play a bit with the Move controller with friends without having to worry that you interrupt the flow of light between the projector and the wall.
The PS3 community would be very happy.
Apart from all the currently available content in S3D, yesterday I went to an official Sony shop and asked the official expert there how can the Sony user will store and play the 3D videos made by Sony HDRTD10 new 3D camcorder.
He said categorically that the PS3 would not play such videos from the USB disk and the user would have to create a 3D Blu-ray.
But of course, the PS3 has got the ability to play 3D videos natively from the USB disk. Converting to this MP4 stereo 3D Sony format is a problem begging to be solved.
the performance of this file is different than the others.
On the PS3 is played like VLC plays it on the PC: the PS3 shows 2D and in rapid succession the left and the right eye images.
Another idea (this was suggested to me) to identify the format is the following:
- opening an open-source video-player in a debugging tool such as Visual Studio,
- See how this tool opens the Sony file and decodes/plays the 1st few frames.
In this way, we can mark all those bytes in the Sony file which have nothing to do with the special Sony 3D format (the pure video stream and parts of the container that we know cannot contain the special bytes).
In this way we can focus on relatively small area of the file.
If we do the same with the file generated by Avidemux, by copying the video and audio streams, we should see better what bytes are different. At this point of the analysis, they might be few.
Just an idea.
Very good progress Mathew.
It seems to me that, concerning the information that MedioInfo can capture, the two video are the same.
Which of the videos is the one you compared? One of those you gave us to test or a new one?
So, from your post before the last one, I understand that you are now moving towards debugging the extra info in each frame.
Please keep us updated.
Do we know what tool PigPines used for his analysis, the analysis that BlackShark pointed our attention to?
Also, was that analysis complete? I have the impression that BigPines implemented the changes he identified but it still did not work on the PS3.