[x264-devel] question regading encoded keyframe size at 720p

[Carl Petersen]

Thanks. Makes sense.

If I set f_rf_constant_max, I can reduce keyframe size to honour a smaller
vbv. If that is not set, how much is quality allowed to be reduced to try to
honour vbv?

In other words, in there a concept of a default value for f_rf_constant_max?


-----Original Message-----
From: x264-devel [mailto:x264-devel-bounces at videolan.org] On Behalf Of
Henrik Gramner
Sent: Thursday, October 15, 2015 10:33 AM
To: Mailing list for x264 developers
Subject: Re: [x264-devel] question regading encoded keyframe size at 720p

On Thu, Oct 15, 2015 at 2:51 PM, Carl Petersen <cpetersen at broadsoft.com>
wrote:
> Hi Henrik,
>
> Yes. Vbv buffer size is 2000 and vbv max_rate is 1024. About a year
> ago buffer size was 350 and it was increased to 2000. I have a basic
> understanding of how vbv works but I still can't grasp on what values
> to choose for our application. 350 was too low. There were frequent
> artifacts that seemed to correspond to keyframes. Increasing it to
> 2000 alleviated the problem but I still cant fully comprehend the
> impact of that. I can sort of appreciate that it can increase keyframe
> size. What I don’t understand is why 350 was too low. Was that vbv
> overflow? And I don’t know what value to pick for vbv buffer size.

x264 will utilize the large buffer size to create big I-frames since that is
beneficial for quality, but it is not ideal for real-time applications.
Having a smaller buffer size solves it but will cause I-frames to be bit
rate starved resulting in poor quality, which is why periodic intra refresh
was created.

For zero-delay streaming with a fixed bandwidth you should use periodic
intra refresh and single-frame VBV (a VBV buffer size that is equal to the
VBV max rate divided by the frame rate) with zerolatency tune.
_______________________________________________
x264-devel mailing list
x264-devel at videolan.org
https://mailman.videolan.org/listinfo/x264-devel

-- 

Learn why you should attend BroadSoft Connections 2015 
<http://www.broadsoftconnections.com>

This email is intended solely for the person or entity to which it is 
addressed and may contain confidential and/or privileged information. If 
you are not the intended recipient and have received this email in error, 
please notify BroadSoft, Inc. immediately by replying to this message, and 
destroy all copies of this message, along with any attachment, prior to 
reading, distributing or copying it.