target bitrate insted of using target size

[besik33]

I just started using Auto GK, because is faster and i can save a time. When decoding from DVD I'm still using original audio with video or matroska format, but then it is another story.....

Is it possible to encode in autoGK to the target biterate or to the target bits/pixel ? Now i'm using bitrate calculator to calculate reguired bits/pixel.

THX

[manono]

Hi-

Since filesize directly correlates with the bitrate, asking to be able to use bitrate instead is pointless. The answer's "no", anyway.

Bits/Pixel is only a guideline, and a pretty useless one at that. Different movies compress differently. Different movies will need different B/(P*F) for the same quality. You're wasting your time. If you're concerned about the quality, then use Target Quality 1-pass encoding. Not only will you be assured of even quality throught the video, but the encoding will be done much faster. But don't complain when your filesizes wind up all over the place.

[besik33]

OK, so I change the reguest.

It is possible to add information about avarage bitrate to the bitrate? It seems for me, like rough information about quality. It's better like nothing.

I have to try the target quality - but i'm not familiar with it, and i'm afraid of ratio quality/size :confused:

I had think that 2pass encoding will give me better quality then the one pass.(but u say that will give me only control about size, am i right?)

[jggimi]

You should read the DivX Users Guide -- it is applicable to DivX as well as other MPEG-4 ASP codecs such as XviD. It discusses the various modes of operation in great detail, and is very easy to read, with great graphics and examples.

It is available in English, German, French, and Japanese at http://www.divx.com/divx/windows/codec/guides/

[besik33]

OK, thx.

I have read this and nothing new.

I am not newbie to the encoding, just i don't have experience with one pass encoding(the results wasn't woth of it)

Imagine this situation:
Create video with quantitazer about 2. Let say it will have about 1gb.
Then create same video with 2passes with size 1gb.

Which one will be better?
I think that always will be better 2pass video(even in higher bitrates) because 2pass will give more bitrate to more complicated scenes and low bitrate to dark scenes(not so complicated). Som scenes wil have quantitazer about 1 and somes about 3.


With target quality i have always same quality. High quality of complicated and high quality non-complicated scenes.

Am i right or not?

[jggimi]

Q=2 is a poor example, as that is effectively 100% (or 98%, depending on ASP codec) quality. In that example, both encodings would have the same quality .... maximum ... and the only difference will be the 2-pass encoding takes twice as long.

You seem to desire bitrate rather than size to specifiy your encoding. They are the same.size = bitrate X time

bitrate = size / timeIf you wish, you can use Gordian Knot, which offers both methods for the convenience of the user.

[mael]

i'm on besik33's side!
i want to enter the bitrate instead of the size

it would be sufficiant to include a calculator so you can enter the bitrate and then the proper size is calculated.

for now i have to use my own size calculator (BASH script) because it seems there are only bitrate calculators out there

[len0x]

OK, thx.

With target quality i have always same quality. High quality of complicated and high quality non-complicated scenes.

Am i right or not?

You are not - target quality is all about average quantizer (since it can't be anything but integer) and the way it works is: you get a certain percentage of maximum available bitrate needed per frame, but that amount varies from frame to frame. High motion frames may have several times higher maximum bitrate than low motion frames even though quantizer is the same. So one pass is no different that 2 pass in this matter, the only difference, one might say, that average quantizer is calculated after first pass to fit the size and used during second pass (not quite of course, but you get the idea).

[Carraway]

...asking to be able to use bitrate instead is pointless.

It's not pointless. There are a myriad of reasons to want to encode in average bitrate, which is why virtually every VFW encoder I can think of offers such an option.

One example: if you're batch encoding a lot of videos and you want each clip to have a certain filesize in proportion to the clip's length (i.e., you want each video to be allocated 10MB for each minute of length, meaning a 1:30.000 clip would be 15MB, a 3:45 clip would be 37.5MB, etc.).

With constant quant, a dark, static one minute clip will be a smaller filesize than a noisy, high-motion one minute clip -- so it fails. With final filesize, you'd have to manually check each clip's individual length to get the proper file size -- this is tedious.

Another example: you're encoding clips for the web. You want them all encoded at approximately the same bitrate so that users' download times are proportionate to the length of the file they're downloading. It's also germane to streaming, although since AutoGK wasn't designed for that, I realize that argument doesn't hold much water.

Average bitrate is a hybrid of file size and constant quant: it offers a reasonable expectation of both the final quality and final filesize of a clip without offering certainty in either aspect. It doesn't provide a guarantee of quality the way constant bitrate does, but it also doesn't have the same wild filesize variability. Nor does it provide exact filesize, but you can be reasonably certain that it will give X number of MB per second of video length. It has its niche, and it is a big time saver on batch encodings where you want a ballpark of quality AND filesize for each clip, and don't want to take the time to perfect settings for each clip individually.

Anyway, it's not pointless. I know it won't ever be implemented in AutoGK, but you dismissed the original poster's question without much thought, and I thought it deserved more than that.

[manono]

besik33 (the OP) was using flawed reasoning, believing that figuring the B/(P*F) actually had anything to do with anything, which it doesn't. If I was dismissive, it's because I meant to be. Encoding by bitrate just shouldn't happen, I don't believe.

...which is why virtually every VFW encoder I can think of offers such an option.

So, go use one of them. You can set the bitrate in VDubMod and batch encode. What do you need AutoGK for anyway? How many times does it have to be repeated; it's for beginners, and for producing AVI for standalones. Beginners don't need to know about bitrate. They encode to a CD or 2, or to a portion of a DVDR. The fact that its use has expanded to also include lazy people in no way means that its features should also expand to allow any screwy request someone might dream up. AAC audio, MP4, Matroska, give me a break. If you want to encode to AVI for a particular size, use 2-pass encoding. If you want to encode to AVI for a particular quality, use Target Quality. If you want something else, use something else. For 99% of the people out there, it fills their needs admirably.

Average bitrate is a hybrid of file size and constant quant...

No it's not. It's entirely about file size. Sure, AutoGK will make some adjustments if you choose a file size either too large or too small (resolution, matrix, resizer), but there's only so much it can do. I may be extrapolating from what you wrote, but it sounds to me like you want to make a given series of batch encodings for a fixed bitrate and all for the same resolution, quality be damned. I find none of your arguments the least bit convincing. If I seem even more dismissive of you than I was of besik33, it's because you know more and should be making more sense, and because I mean to be. It's not up to me anyway. If any of your arguments resonate with len0x, then who knows?

[jggimi]

And, as I already stated in an earlier post, GK has this capability in its bitrate calculator. You can specifiy an intended file size, or you can specify a target average bitrate.

In both cases, the encoding will be multiple pass, not fixed bitrate, and will have the same result. Bitrate is file size divided by time.

[Carraway]

I may be extrapolating from what you wrote, but it sounds to me like you want to make a given series of batch encodings for a fixed bitrate and all for the same resolution, quality be damned.

That's not right, and sorry if I wasn't clear. I'll elaborate on my hypothetical situation.

Say I was backing up two extras on a DVD, but they were vastly different in quality.

Clip 1: One minute, dark, static.
Clip 2: Three minutes long, noisy, bright, high motion.

Here is how AutoGK would handle these clips with each of the three size options.

Constant Quality: AutoGK would assign both the same resolution, and the same quality. However, the final file sizes of the clips would be absurdly different. It's not out of the question for two clips of similar length, using the same average quant, to have final file sizes that are an order of magnitude apart (i.e. the first might be 4 MB while the second is 40 MB).

There's no way to determine this disparity before encoding, so I would likely end up having to re-encode Clip 2, either lowering the resolution and quant and using single-pass again, or inputting my desired file size and using two-pass.

And this is with only two clips. If I were backing up 20 extras off of a DVD, it would be a huge hassle.

Predefined size: I could not use this option without having some sort of prior knowledge of the two clips. I cannot blindly choose a file size for each clip without knowing the length or compressibility, so there's really no way to adequately use this option.

Average bitrate: Hypothetically, say I choose an average bitrate of 1000 kbps. AutoGK would do its normal two-pass process, run a compressibility test, and for the first clip it would bump up the resolution and encode. Final file size would be 8.5 MB. For clip #2, it would run the compressibility test and find its compressibility low, so it would knock down the resolution. The final filesize would be around 25.5 MB and the resolution would be considerably lower in Clip #2 than Clip #1.

Both would be the same "quality" because AutoGK would adjust the settings and resolution to accommodate the comptest values, which is something single-pass simply doesn't offer. And the file size disparity would be much smaller and more reasonable. Neither would need to be re-encoded.

...for a fixed bitrate and all for the same resolution, quality be damned.

Just to sum up -- not a fixed bitrate, an average bitrate; not the same resolution, different resolutions chosen by AutoGK; not quality be damned, AutoGK would adjust the clips so all have the optimal quality for their size.

Average bitrate is a hybrid of file size and constant quant...

I didn't mean it was literally a hybrid, I meant it was a middle ground.

"Target Quality" mode counts on the user to supply a resolution and quality for the clip -- it doesn't take file size into account, and rightly so. But when batch encoding, this can and will lead to a disparity in clip size.

"Preferred Size" option is good when you know the clips you're inputting, but horrible when batch encoding since you can't possibly know the length/compressibility of each clip beforehand.

With an average bitrate option, it still allows AutoGK to use its compressibility tests to determine the best quality without needing to know the length of the clips you input. It's a little bit of "Target Quality" because you're ballparking the final quality of your video, and a little bit of "Preferred Size" because you know exactly the size/length ratio.

So, go use one of them...If I seem even more dismissive of you, it's because I mean to be...asking to be able to use bitrate instead is pointless...screwy request someone might dream up.

Geez.