How to get output encoded Frame from Jpeg Encoder for every input frame (no buffering)

How to get output encoded Frame from Jpeg Encoder for every input frame (no buffering)


I am using the sample Jpeg Encoder from sample_encode application of Media SDK.

By default, this encoder is providing the first encoded output frame after consuming 8 input frames. So I changed the AsyncDepth value to 1 (default setting was 4). Now it generates the output frame after consuming 2 input frames but i am looking for out from first frame itself.

After debugging the code further, I found that the AsyncDepth value actually derives the size of encoder task pool (=AsyncDepth*2) which cant be set to 0.

So is there any way through which i can generate the output encoded frame from first input frame itself??



4 Beiträge / 0 neu
Letzter Beitrag
Nähere Informationen zur Compiler-Optimierung finden Sie in unserem Optimierungshinweis.

No response yet. Is there something missing or unclarified in my query?

OK, let me put in other and simple form: 

Is there any mechanism/API/lib in IMSDK through which I can convert any YUV420 frame into a Jpeg frame?




Sorry for the slow response.

The sample code that is using "AsyncDepth*2" and doing a check for even number of tasks was only for creating multi-output streams like MVC.  If the output is JPEG still images, there is no need for this.

You may use ASyncDepth of 1 without issue, but you will need to modify the sample code managing the tasks.  If you remove the " * 2 " from the calling function and remove the check for even value in CEncTackPool::Init() function (in pipeline_encode.cpp).

The limit is really a "sample code" limitation, and not a limitation of the Intel Media SDK API itself.

We'll look at the possibility of updating our sample code in the future.


Thanx Tony. I will check this approach.

Meanwhile a further doubt :

If I set the AysncDepth to 1, and remove the check of "*2", then probably it will create the Encoder task pool with size 1. So I think Jpeg encoder will start giving the first encoded output frame from 2nd input frame onward (i.e it will still consume 1 frame as buffer). Isn't it so? What do you think?

I will try to debug and confirm it at my end also.


Kommentar hinterlassen

Bitte anmelden, um einen Kommentar hinzuzufügen. Sie sind noch nicht Mitglied? Jetzt teilnehmen