avi to dvd ideal bitrate question

[45tripp]

It's been a while since I've done an avi to dvd conversion.
So I was looking it up a bit.
I was wondering if someone uses a formula to derive the best mpeg2 bitrate to use according to the sources' bitrate? Or is the goal usually just to fill a single dvd(4.7)?
In the past i did a 2 Avis to one dvd but can't remember how well it turned out compared to one-to-one. I'll have to find the disks.

For examples sake, take the info on the video of an avi I might convert soon:

Codec : DivX 5
Bit rate : 1778 KBps
Width : 640
Height : 336
Aspect ratio : 16/9
Frame rate : 25
Bits/(Pixel*Frame) : 0.331
Duration: 1h 30min 21sec


Also say the bitrate of your source is low.
In that case, if you are using a full dvd do you just go for CBR encoding?

What would be a good cutoff point between moving between VBR and CBR?

ty

[jagabo]

You typically need 2 to 3 times the bitrate when converting from MPEG4 (Divx, Xvid) to MPEG2. It will depend on things like if you're upsizing or downsizing, encoding 4:3 fullscreen, 16:9 letterboxed in a 4:3 frame, or anamorpic 16:9, the quality of your source, etc.

But if file size isn't critical just use a constant quality encoding. Pick the quality you want and you'll automatically get the right bitrate for the video regardless of frame size, 4:3 vs 16:9, video complexity, etc. Not all encoders support this though. TMPGEnc Plus and CCE do.

[45tripp]

You typically need 2 to 3 times the bitrate when converting from MPEG4 (Divx, Xvid) to MPEG2. It will depend on things like if you're upsizing or downsizing, encoding 4:3 fullscreen, 16:9 letterboxed in a 4:3 frame, or anamorpic 16:9, the quality of your source, etc.

Thanks.
Maybe you could give a specific example.
I'd say generally upsizing, anamorphic, good quality.

If I could rephrase.
When your source avi is of good quality and has video bitrate below 1300 Kbps, and of an average film duration, how much difference in quality will you get going beyond the 2x (bitrate) mark when converting to mpeg2? Considering that if stiking to 2x you could get 2 avis to one dvd.


On a slightly different note:
I tried out mainconcept with the above clip. It took 2 hours. I didn't touch much setting wise.
Then I used BitrateView for the first time.
Afterwards my normal conversion with cce and avisynth serving it up.

AVISource("A.avi", False)
STMedianFilter(8,20,4,8 )
ConverttoYUY2
ColorYUV(off_y=-20, gain_y=32, cont_u=0,opt="coring")
Lanczos4Resize(720,552)
AddBorders(0,12,0,12)

I saw STMedianFilter used somewhere and wanted to try it out. Just playing with the colour a bit.
Conversion took 3h for VBR 2 pass (1+vaf).
Pretty slow I think. How much do the above filters slow down the process? I wouldn't think by much.

Also fed result into Bitrate View.
The concept clip spiked considerably over the maximum bitrate setting. The CCE didn't.
The concept clip had Q levels below bitrate through out.
The CCE clip had the Q level above the bitrate through out.
I don't really understand bitrate view or q levels. First time using.
But I read the lower the Q level the better.
Is that right?
Is it a bad thing that the cce clip had q levels way higher than the mainconcept clip?

[45tripp]

And to repeat the question:

"Also say the bitrate of your source is low.
In that case, if you are using a full dvd do you just go for CBR encoding?

What would be a good cutoff point between moving between VBR and CBR? "

[Scorpion King]

According to https://www.videohelp.com/dvd "What is DVD" the total bitrate including video, audio and subs can be max 10.08 Mbps (10080 kbps).

If you're using no subs, and the audio is 256Kbps then the video can be max 9.824Mbps. For a 1hr dvd you could use CBR 9.824Mbps, but if your video was longer than 1 hour then it would probably benefit by using VBR instead. You would then set the average to whatever your bitrate calculator said.

Using VBR for a 1 hour or less video when you are trying to fill the disk would put the max and average at the same value. That will work but might not make a whole lot of sense. (That's why I do it all the time )

Of course this would leave no room for menus. lol

Good luck.

[45tripp]

According to https://www.videohelp.com/dvd "What is DVD" the total bitrate including video, audio and subs can be max 10.08 Mbps (10080 kbps).

If you're using no subs, and the audio is 256Kbps then the video can be max 9.824Mbps. For a 1hr dvd you could use CBR 9.824Mbps, but if your video was longer than 1 hour then it would probably benefit by using VBR instead. You would then set the average to whatever your bitrate calculator said.

Using VBR for a 1 hour or less video when you are trying to fill the disk would put the max and average at the same value. That will work but might not make a whole lot of sense. (That's why I do it all the time )

Of course this would leave no room for menus. lol

Good luck.

Thanks.
I always work off a 9000-9200 kbps max.

But my question isn't about the technical limitations of what I'm doing.

I'd like to know, to put it as plainly as I can,
how much diff there is in converting a 1000 kbps Xvid video to a 2850 kbps Mpeg2 video or to a 6200 kbps Mpeg2 video, quality wise? Perhaps a highly subjective question, but I'm open to highly subjective answers. Looking to draw on other peoples experience.
Then,
assuming it's worth taking a 1000 Kbps Xvid up to a 6200 Kbps Mpeg2, is it worth going VBR or is there no advantage in doing that over using CBR?

More specifically,
does anyone have any rules, or is able to deduce such rules, for the ideal way to select which is the best (quality/space consideration) way to pick your final mpeg output bitrate based on the birate of your mpeg4 source?
i.e. (speculation)
if your source is below the 1000 Kbps mark it's hardly worth converting it to a full 4.37 dvd, but better to keep it to 1/2 a disk saving space, as there are essentially no gains in going for the higher bitrate. The bitrate is wasted?

I don't know.
That's why I'm asking.

Also a reminder,
I asked something about bitrate view and q levels.
just in case it's forgoten.

help appreciated

[Scorpion King]

Others here can answer your question far better than I.

The one thing you must remember is that all this is up to you. You're in control of your files and you need to just do it how you want to. That is all thats really important anyway. And that is what makes this hobby fun.

We do something the best we can based on what we thing and know and are happy with the results until we learn something new then we go about it in a different fashion. Thats Fun!!!!

My theory:

Every video file I have created has garbage in it, ie blocking, noise, etc. Some of my files are better than others because the source was better, even though the bitrates/etc are the same. Reencoding these again will require different bitrates to 'retain the quality' they have now because of the garbage they contain, besides the complexity and action in the original file (what makes up the actual video). Because this garbage adds complexity.

When you add complexity to a video the necessary bitrate will also increase that is needed to retain the quality during the reencode. Not using the needed rate will guarantee you a quality hit. There won't be enough bits to retain the details of the garbage and the video, which will introduce even more garbage.

If you don't have room for the necessary bitrate then you can either take a quality hit and enjoy the conversion, or try and filter out some of the garbage (complexity) by smoothing/deblocking/etc. You will most likely loose detail in the process, but you will remove some of the complexity of the file making it possible to reencode with a less bitrate because of the added ability of the encoder to compress the file. Hopefully only removing the garbage that is.

I might be wrong, but I don't care.... I'm having fun

Good luck.

[jagabo]

does anyone have any rules, or is able to deduce such rules, for the ideal way to select which is the best (quality/space consideration) way to pick your final mpeg output bitrate based on the birate of your mpeg4 source?

Mostly, I just use a Divx/DVD player so I don't have to convert. When I do convert I encode with constant quality. That is the simple solution to your problem. YOU decide what QUALITY you want. The encoder decides, at each frame, how much bitrate is needed to deliver that quality. Since you have CCE try encoding "1-pass VBR", set the Quantization Factor to the quality you want (I usually use 15, the smaller the value the higher the quality), set the Min and Max Bitrate values to legal values for DVD (I usually use 1000 and 8000), then encode. It's fast and simple.

Doing this you'll find that the overall average bitrate varies depending on the source. With Q=15 I typically see about 2 to 3 times the source's bitrate. That quality level will add a very small amount of macroblocking but it won't be noticed at normal playback speeds.

This is especially useful for 22 or 45 minute TV episodes. Encode them all with the quality you want then burn as many episodes as will fit on each DVD.

[45tripp]

When I do convert I encode with constant quality. That is the simple solution to your problem. YOU decide what QUALITY you want. The encoder decides, at each frame, how much bitrate is needed to deliver that quality. Since you have CCE try encoding "1-pass VBR", set the Quantization Factor to the quality you want (I usually use 15, the smaller the value the higher the quality), set the Min and Max Bitrate values to legal values for DVD (I usually use 1000 and 8000), then encode. It's fast and simple.

Thanks.
It was very helpful.
I looked up CCE and the Q factor and found a great guide on doom9.
It'll help me take a more practical 'scientific' approach to the problem.

I might be wrong, but I don't care.... I'm having fun

Can't argue with that 8)

For me it's an equation with Time, Quality and Space. In that order!
Jagabo's tip has led me to a method that will help me find whatever balance I may be seeking.

You mention complexity. A pretty loose term.
Are fast motion vids the most complex? Is the sum of movement the main force behind 'complexity'? Where do animations stand complexity wise?
I guess it doesn't matter much.

Thanks guys for the help.

[jagabo]

You mention complexity. A pretty loose term.
Are fast motion vids the most complex? Is the sum of movement the main force behind 'complexity'? Where do animations stand complexity wise?

As far as MPEG compression is concerned, here are some properties that require more bitrate:

1) The amount of detail in each frame. A flat single color frame requires very little bitrate. The more details there are the more bitrate the codec will require.

2) The changes from frame to frame. This can be things like motion, fades, smoke, fog, fire, water, etc. A large part of MPEG compression comes from not having to encode static parts of the image over and over again. You have a key frame (basically a JPG picture) followed by several frames that only encode the differences between frames. If there are lots of differences between frames more bitrate will be required.

3) Static video noise. This is really the same as #1 and #2 but one might not think of it.

4) Bright scenes require more bitrate that dark scenes.