Explanatory animation about CBR and VBV

leoenc · 20th August 2014, 11:52

Hi all,

This topic applies also to other MPEG codecs, but I believe since H.264 is the most common codec today, it should belong here.

Since I believe it is very important to understand how buffer works when doing encoding for streaming, and since I couldn't find a clear visual explanation for this, I decided to try and make one myself.
Here's my first quick animation draft to try and explain it using the leaky bucket analogy:

Your feedback please!
Do you think it's doing a good job explaining this or does it totally suck?

Did I get anything wrong?

Leo

LoRd_MuldeR · 21st August 2014, 13:53

Nice anymation, though I'd probably do the analogy the other way around

If the bucket represents the buffer, it would be filled at a constant rate (VBV Maxrate, e.g. the network bandwidth) and, depending on the individual frame, different amounts of "water" need to be taken out of the bucket for each frame. VBV ensures we will always have enough "water" left in the bucket to satisfy the next frame. If that's not the case, i.e. we run dry, we have a buffer underflow.

(With your current analogy, the filling level of the bucket is the inverse of the decoder's video buffer's filling level)

nevcairiel · 21st August 2014, 14:16

I guess it depends if you see it on the side of an encoder (ie. constant rate flowing out) or on the decoders side (constant rate flowing in).
It makes more sense to me this way around, because only the encoder actually has any control about the flow into the bucket.

LoRd_MuldeR · 21st August 2014, 14:48

Quote:

Originally Posted by nevcairiel

I guess it depends if you see it on the side of an encoder (ie. constant rate flowing out) or on the decoders side (constant rate flowing in).

You a right. I was thinking more about what is happening at playback time, so looking at it from the decoder side. Both perspectives are equally valid. But I think the animation should at least make clear from which side we are looking at it.

(Or maybe, for a more complete picture of the whole streaming process, the "outlet" of his current bucket could lead to a pipe, leading to another bucket of same size, representing the decoder buffer)

poisondeathray · 21st August 2014, 15:21

Just curious, What did you use for the fluid simulation ?

leoenc · 21st August 2014, 16:19

@LoRd_MuldeR, @nevcairiel - Thanks for your feedback! Didn't think about showing the decoder as well. You're right - I will make it more clear that it's about the encoding.

@poisondeathray - I own a copy of Softimage (using ICE). It's just a really quick simulation though, without fine-tuning the fluid properties, which obviously doesn't look real.

20th August 2014, 11:52	#1 \| Link
leoenc Registered User Join Date: Mar 2007 Posts: 181	Explanatory animation about CBR and VBV Hi all, This topic applies also to other MPEG codecs, but I believe since H.264 is the most common codec today, it should belong here. Since I believe it is very important to understand how buffer works when doing encoding for streaming, and since I couldn't find a clear visual explanation for this, I decided to try and make one myself. Here's my first quick animation draft to try and explain it using the leaky bucket analogy: Your feedback please! Do you think it's doing a good job explaining this or does it totally suck? Did I get anything wrong? Leo

21st August 2014, 13:53	#2 \| Link
LoRd_MuldeR Software Developer Join Date: Jun 2005 Location: Last House on Slunk Street Posts: 13,196	Nice anymation, though I'd probably do the analogy the other way around If the bucket represents the buffer, it would be filled at a constant rate (VBV Maxrate, e.g. the network bandwidth) and, depending on the individual frame, different amounts of "water" need to be taken out of the bucket for each frame. VBV ensures we will always have enough "water" left in the bucket to satisfy the next frame. If that's not the case, i.e. we run dry, we have a buffer underflow. (With your current analogy, the filling level of the bucket is the inverse of the decoder's video buffer's filling level) __________________ There was of course no way of knowing whether you were being watched at any given moment. How often, or on what system, the Thought Police plugged in on any individual wire was guesswork. Last edited by LoRd_MuldeR; 21st August 2014 at 14:09.

21st August 2014, 14:16	#3 \| Link
nevcairiel Registered Developer Join Date: Mar 2010 Location: Hamburg/Germany Posts: 10,046	I guess it depends if you see it on the side of an encoder (ie. constant rate flowing out) or on the decoders side (constant rate flowing in). It makes more sense to me this way around, because only the encoder actually has any control about the flow into the bucket. __________________ LAV Filters - open source ffmpeg based media splitter and decoders

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

21st August 2014, 15:21	#5 \| Link
poisondeathray Registered User Join Date: Sep 2007 Posts: 4,487	Just curious, What did you use for the fluid simulation ?

21st August 2014, 16:19	#6 \| Link
leoenc Registered User Join Date: Mar 2007 Posts: 181	@LoRd_MuldeR, @nevcairiel - Thanks for your feedback! Didn't think about showing the decoder as well. You're right - I will make it more clear that it's about the encoding. @poisondeathray - I own a copy of Softimage (using ICE). It's just a really quick simulation though, without fine-tuning the fluid properties, which obviously doesn't look real.