Bitrate - Need to be educated...

[NIIDeep]

Okay I've read LordSmurf's "AVI vs. Mpeg" article and it talks about max bit rates for certain resolutions and formats. There is a plateau in which increasing bit rates has diminishing returns.

However, I still do not understand the concept of how the bit rate is calculated and what it pertains to. In ATI MMC 9.1, I understand the diff between CBR and VBR but the actual values to choose, I am not sure what to do. I could just take LordSmurf's word for it but I'd like to learn. How did he calculated those theoretical numbers?

So, if anyone can explain to me what bit rate is and how it is calculated that would be great. I also welcome any web site links that will educate me

Thanks.

[andkiich]

https://www.videohelp.com/calc.htm

Look at the bottom of this page.

[NIIDeep]

andkiich,

Thanks for the link... it answered one part of my question.

As I see it there are two context of bit rate so to speak. One, pertaining to the link you sent, is the bitrate used to figure out the maximum bit rate that can be used to fill a given media of X MB/GB space for a given length of video (plus sound). So as the length video continues to increase while the media space remains the same, the bit rate (being inversely proportional) decreases. The longer your video, the lower the bitrate at which you can capture.

But.... there must be a limit, a minimum bit rate before quality is compromised.

LordSmurf kinda alluded to this in the "AVI vs MPEG" article. For a given resolution there is a minimum bit rate that will maintain quality. Any increases in bit rate beyond the "minimum" for a given resolution would have diminishing returns if any in term of quality. Meaning your quality won't get that much better with super high bit rates (or at least higher than the "minimum"). But any bitrate less than this "minimum" would compromise quality. It's that special magic bit rate number.

So I can interpret this as.... if the length of video is so long such that for a given media size, your bit rate is brought below the "minimum" bit rate (specified by LordSmurf's article) then you should starting thinking about dividing the video and puttig them on multiple media (i.e. a second CD, DVD, etc.) to meet the minimum bit rate requirement to maintain full quality.

My question is..... how did LordSmurf calculate the "minimum bit rate" in his article?

[chicola]

You can't decisively call it a "minimum bit rate" because the degradation is gradual.

Lowering the bit rate will give you worse, and worse and worse results.

The threshold depends on your eyes!

By definition: You think that something has QUALITY when it's characteristics meets your expectations.

So that precise threshold depends on each person. Surely, we can all agree that a really bad (compressed) image is a really bad image but the threshold is another matter.

[NIIDeep]

Well .. I can agree with you on that... but is there anyway to calculate a "recommended theshold" just for a guideline?

I mean.. when someone decides to capture a source to put on some media... and the capture application has a setting for bit rate..... how does one know what to set it to? It shouldn't be a trial and error process... set bit rate x... burn... watch... no good... set bit rate y... burn watch.. okay but could be better... set bit rate z... burn... watch... hmmm seems like no difference...

what about a person who wants to optimize usage of their media... trying to cram as many high quality video segments on to a particular media, best bang for your buck scenario.....

There MUST be some math we can do to say... okay this is a recemmended guideline for the threshold. I'll deviate from this up/down.... to suite my needs.

all the bit rate calcs are working off the assumption that a person wants to completely fill up a media with one video clip. There's none.. I don't think that will warn them... well if you're video is too long.. you're going to get crappy results or if you put too many videos segments on a media... you're sacrificing bit rate to a point where you quality is crappy....

I hope I'm making my question clear.

LordSmurf seems to have come up with some recommended threshold values... I've not been able to find anything like this anywhere else ... so I would like to know how does one come up with these numbers?

I am *assuming* that it can be calculated based on some theories.... I'm sure he didn't just make them up.... unless... he painstakenly did the trial and error process while using some measurement tool to come up with his results... then this could explain... why no one has a magic formula to calculate this stuff....

This is something I haven't thought about.... but it could very well be what was done.


anybody else have any insight? I welcome any and all opinions.

Thanks...

[Capmaster]

I'll echo what chicola said. Perceived quality is very subjective. For example, my wife and I have drastically different perceptions of what is "good enough" and what is "crap". Funny enough, she is the one who is way more critical. I can watch a movie and be happy with it, at the same time she is saying "didn't you see the blocks in that scene?"

There are all sorts of artifacts that creep in when lowering the bitrate ...like macro blocks, motion in still objects, hesitation, fuzzy edges, etc. Where each one registers in your brain is a personal thing.

I'd highly recommend the trial and error method because no matter how much advice you get here about "good" and "bad" bitrates, they are subjective, and you'll just end up going the trial and error route eventually

[ZippyP.]

Every person will have a different minimum bitrate that they will try to stay above. It depends on the resolution, type and quality of the source, encoding method (CBR, VBR), output type (divx, mpeg) and ultimately how fussy they are. You can ask what someone would use in a given situation if you give all the details. Finally though, you will just have to learn from experience what your threshold will be.

[chicola]

NIIDeep,

Trust us (we're not trained professionals! ). Don't waste your time trying to find the "holy graal". Use some of the most used thresholds and iterate a little until you are confortable.

You'll never reach an optimum.

[NIIDeep]

well I'm not "really" trying to reach the holy grail... per se... I was just curious to understand the reasoning behind, specifically, the numbers behind LordSmurf's bell curves in his article "AVI vs. MPEG".

I just want to try to understand things rather taking in people's suggestions, picking up suggested "magic" bit rate numbers. Where do they come from?

BUT... I think I've been pursuing this from the incorrect perspective. The best conceptual explanation I can come up with bitrate is a analogy with bitrate regarding MP3 audio.

With ideal means... you would want infinite bit rate to "perfectly" reconstruct the analog signal. But since we're digitizing, bit rates can only be finite. The higher the bit rate the closer to the original you'll get but you'll never reach it. Thus you have to compromise and find that bit rate in which gives you the perception that it's "perfectly" or "good enough" reconstructed. This perception is subjective and unique from individual to individual. Ex... some people think they can tell the difference between MP3 audio ripped at 128 Khz, 192 Khz, 256 Khz, etc..
For MP3, I think 192 Khz is good enough.

For video... well.. I guess I have to figure it out for myself.

I hope I have the right thinking regarding this. So I'm just going to go with this for now.

I still wonder about LordSmurf's bell curves.... they seem to imply a rather mathmatical justification to the optimal bit rate. Perhaps I'm reading too much into it. Oh well..

Thanks for everyone's help.

[ZippyP.]

For MP3, I think 192 Khz is good enough.

I agree with you, but there seems to be a lot of 128 kbps MP3's around. Seems like some people either don't care about quality or are listening to music on a really crappy system. Same thing goes for video, many are happy with VCD quality which is at the low end of the spectrum.

If you really want an explanation of Lordsmurf's bitrate recommendations then you should probably go to the horse's mouth.

[NIIDeep]

If you really want an explanation of Lordsmurf's bitrate recommendations then you should probably go to the horse's mouth.

I did... but he said he couldn't explain more beyond what's already on the web site.

It's okay.. I think the bell curves and stuff were arbitrarily created from his perspective, anyways. The general consensus is to do the "trial and error" method... and so I'm embarking on this path.

Thanks.

[victoriabears]

Simple question (Ha!)

I take a 2 hour long VHS capture it direct to Mpeg, edit it using TMPG author(Which fits my editing needs fine thanks) end up using only 75% of the dvd capacity, could I then re-encode the edited Mpeg at a higher bit rate top fill the DVD to near capacity and get better quality.

Or does it have to be captured to avi, then edited then encoded to mpeg to get the "best" image?

Thanks

[TheFamilyMan]

When considering "what is the appropriate bit rate", you also have to look at the source being converted. Loosely speaking, MPEG2 compresses by taking a reference frame and then appling several "deltas" to that frame to make the succeeding frames. If there is very little motion, the amount of data needed for the deltas is small. If there is lots of motion, lots of data is needed for the deltas. And there you have it...if your source footage is from an unstabilized hand held analog camcorder, using 4000 max bit rate for 352/480 resolution doesn't cut it quality wise due to all the motion and VHS video noise (which appears as motion to the MPEG2 compression). On the other hand, good quality captures of most TV shows are the opposite. Bottom line - the quality of your source footage has as much to do with the necessary bit rate as does its resolution - they're just ain't no absolute answer to this question....

[ZippyP.]

Or does it have to be captured to avi, then edited then encoded to mpeg to get the "best" image?

If you want to re-encode then it is best to capture using a lossless (or close to lossless) codec. Mpeg2, Divx and Xvid are lossy formats, high in compression but with a loss of information. Since uncompressed video takes up far too much space, a compromise is to capture with the huffyuv codec (or similar) which is near lossless. After editting you can create your mpeg which is sized to fill up the disk. Every compression causes a loss in quality so doing it only once is the best. Hope that helps.

[victoriabears]

Hey thanks, I have a 300gb hard disk set up so initial size not a problem, what I wish to produce for this project is the very best I can, it is only an hour long, so again thanks.

Using Tmpgenc express should do a very good job I think.

[victoriabears]

Can you make any suggestions regarding the low level of sound I have on this concert recording, there is a bit of sound noise but not too bad, its the lack of volume, will anything from Tmpgenc express help or would demuxing it help?

Thanks a lot.