• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 98 subscribers
  • Views 427 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

About Pseudo Multi-pass Encoding

Fikret Alim
Intellectual 610 points

Hi,

We have implemented Pseudo Multi-pass encoding on DM365. I have seen that the I frames are forced when a scene change occurs. But, this happens only if the scene change is very obvious, let's say if 90% of the scene changes. It didn't work if a person or a car passes from the scene. My questions are;

1) What is the criteria to decide the scene change? Is there a sensitivity adjustment for detecting the scene change? I want it more sensitive to motion. Can I adjust it to detect the frame as a scene change when a person or a car come into the frame.

2) As you know, the low resolution encoder produces metadata and the high resolution encoder uses this metadata. From the point of multi-pass encoding, is this metadata only used to force I-frame in high resolution encoder, or it even improves the quality of the high resolution encoder even it does not force I frame. I mean, is the produced metadata used to improve the P frames of the high resolution encoder, or is it just used to detect the scene change and put I frame in this case?

Thanks&Best Regards,

Fikret

  • 0
    TI__Expert 3065 points

    Fikret,

    My  answers for your questions:

    1: There is no sensitivity control for multi pass encoding to control Iframe insertion. Low resolution encoder internally decides about scene change based on some logic. Based on motion vector and other frame statistics of current frame, low resolution encoder internally takes the decision. No user control is given for this.

    2. Yes, we use metaData of low resolution to force I frame if scene changes is detected. No other action is taken based on metaData of low resolution encoder.

  • 0 in reply to Ritesh Rajore
    Intellectual 610 points

    Dear Ritesh,

    Thank you for your prompt response and support.

    Is there a minimum percentage of scene change to enable I-frame insertion, because as far as I understand from my tests, almost the whole scene must be changed in order for the algorithm to insert I-frame.

    Best Regards,

    Fikret

  • 0 in reply to Fikret Alim
    TI__Expert 3065 points

    Fikret,

    I don't think scene change is reported only if almost complete frame is changed. It should be somewhere 50-70 % with some other factors. Number of inter/intra MBs in a frame play important role in decision making. Many factors were taken into consideration before reaching to current logic to avoid false alarm.