predicting filesize of resulting mpeg?

[freebutterflyx]

Some time ago I encoded an avi (2 cd's of 700 MB and a total of 2h05) to DVD with TMPGEnc and I used a constant quality setting of 85 and bitrates between 2000 and 8000. It just fitted on a DVD. (I use TMPGEnc to author them)
Last week I tried to encode another avi (also 2 x 700 MB but a total of 1h50). Now I had to use a constant quality of 94 to have to same size of the mpeg. (I tried it three time with different quality setting.)

Is there any way to predict the resulting filesize? I rather not use a 2-pass conversion because that's very time consuming compared to a constant quality. Also I read that TMPGenc isn't so good in a 2-pass conversion.

[mats.hogberg]

With CQ it's not possible. CBR is fairly exact, VBR with avg bitrate set according to calculations a little less predictable, but depending on encoder not that much off mark. CQ, as mentioned totally unpredictable.

/Mats

[jimmalenko]

That's the thing with CQ - you can't predict the filesize.

TMPGEnc is absolutely fine in 2-Pass - I use it all the time.

IMO if you need to hit a target, you'd be better off calculating the average bitrate you need with VideoHelp.com's Bitrate Calculator, then encoding 2-Pass VBR with the average you just calculated, 8000 Max and 2000 Min.

[bugster]

I rather not use a 2-pass conversion because that's very time consuming compared to a constant quality.

Not as time consuming as using CQ mode 3 or 4 times trying to hit a certain filesize

There is nothing wrong with TmpGenc's 2-pass VBR quality wise. Yes it takes longer than CQ mode, but if you want to maximise disk useage and therefore bitrate, it is the best way to go.

[lordsmurf]

Do not worry about filesize. Just make the video look good. It takes what it takes.

[mats.hogberg]

Not much use encoding to a file that won't fit your disc, is it? Now, you could always shrink it to fit afterwards, but then, I'm under the impression that encoding to the right size to begin with, gets you the best quality for the space available.

/Mats

[bugster]

Do not worry about filesize. Just make the video look good. It takes what it takes.

If the video is 60 mins long, thats fine. If however it is 3 hours, what CQ setting would still allow it to fit on a DVDr?

Don't know? Neither do I Thats why I use 2-pass VBR and take the encode time hit.

[lordsmurf]

Well, the second complaint used by the OP was that it didn't fill the disc. That should never be a concern, just make it look good.

[jimmalenko]

Well, the second complaint used by the OP was that it didn't fill the disc. That should never be a concern, just make it look good.

I want my stuff to be the best quality it can be, given that it fills the entire disc. If this means I've got to drop from Full D1 to Half D1, then I'll do it.

I want to have my cake and eat it too

[monzie]

I want my stuff to be the best quality it can be, given that it fills the entire disc. If this means I've got to drop from Full D1 to Half D1, then I'll do it.

I want to have my cake and eat it too

What???????????

Drop to halfD1? Are you MAD??? INSANE?? JUST CRAZY???

You've lost the plot somewhere....

GOLDEN RULE NO1: Resolution ABOVE bitrate at ALL TIMES.

GOLDEN RULE NO2: BITRATE is SECONDARY to RESOLUTION by DEFAULT.

For first class single pass encoding use CCE or CCE BASIC.

A 2hr DVDr with ac3 audio (448kbs) will need around 4600 CBR or AVE, which is fine for 95% of most encodings.

[jimmalenko]

I want my stuff to be the best quality it can be, given that it fills the entire disc. If this means I've got to drop from Full D1 to Half D1, then I'll do it.

I want to have my cake and eat it too

What???????????

Drop to halfD1? Are you MAD??? INSANE?? JUST CRAZY???

You've lost the plot somewhere....

GOLDEN RULE NO1: Resolution ABOVE bitrate at ALL TIMES.

GOLDEN RULE NO2: BITRATE is SECONDARY to RESOLUTION by DEFAULT.

For first class single pass encoding use CCE or CCE BASIC.

A 2hr DVDr with ac3 audio (448kbs) will need around 4600 CBR or AVE, which is fine for 95% of most encodings.

No, my preference

If I have a 3 hour sporting event capped off TV @ 720 X 576, I'll happily drop it to Half D1 so that the whole thing fits on 1 DVD.

My preference, my decision.

[monzie]

"I'm a Doctor not a mechanic"

[jimmalenko]

"I'm a Doctor not a mechanic"

Those anger management classes are *really* starting to pay off

[monzie]

"Beam me up, Scotty"

[bugster]

GOLDEN RULE NO1: Resolution ABOVE bitrate at ALL TIMES.

GOLDEN RULE NO2: BITRATE is SECONDARY to RESOLUTION by DEFAULT.

Umm, how can I say this without appearing rude. NO. There, said it

I most cases you are right, but for longer encodes (say 2 1/2 hours or more, depending on the type of material and original source) you obviously need lower bitrates. If your average bitrate drops too low (and again, this depends on the material and also personal preferences) you are going to start to see artifacts and perhaps macroblocking. By reducing the resolution to half D1, the picture becomes a little softer and some fine detail is lost, but the artifacts are reduced and macroblocking can disappear.

There are no hard and fast rules such as those above, there are only guidelines, personal preferences and experience.

[lordsmurf]

GOLDEN RULE NO1: Resolution ABOVE bitrate at ALL TIMES.

GOLDEN RULE NO2: BITRATE is SECONDARY to RESOLUTION by DEFAULT.

Umm, how can I say this without appearing rude. NO. There, said it

.

Somebody needs to tell that to Panasonic about their DVD recorders. Lots of resolution ... so you can better see all the noise and macroblocks.

[jimmalenko]

I'd almost go as far to say that BITRATE determines RESOLUTION for me.

I'm by no means a purist, but like my "products" to be watchable, naturally. I've only got a 27" TV, so even VCDs are quite watchable IMO.

Say I have capture x, runtime: 2 hrs 25 min (for want of a better number). I work out the bitrate (video 4127, audio 224) and straight away I can see that Full D1, even with 2-Pass VBR, is going to be a bit of a stretch if it's medium to high action footage. @ Half D1, I could almost do a CBR and get typically decent (IMO, of course) results, but I err on the side of caution and do a 2-Pass VBR anyway, with Max 6000, ave 4100 and min 1000.

- Quality is compromised a little (but not noticably on my setup)
- 1 disc used completely (no changing discs, I stopped that when I stopped making VCDs)
- I know before I start the encode that TMPGEnc *will* hit the size I require
- IMO the best of all worlds, done once, done right, output's good, 1 disc, everyone's happy.

[bugster]

GOLDEN RULE NO1: Resolution ABOVE bitrate at ALL TIMES.

GOLDEN RULE NO2: BITRATE is SECONDARY to RESOLUTION by DEFAULT.

Umm, how can I say this without appearing rude. NO. There, said it

.

Somebody needs to tell that to Panasonic about their DVD recorders. Lots of resolution ... so you can better see all the noise and macroblocks.

Smurfie, are you saying that Panasonic DVD recorder still use full D1 in 4 hour and six hour mode ! ( do Pana's do 6 hour mode?).

That must look pretty awful!

[lordsmurf]

No, not that bad. They just push the important of resolution, IMO, over that over visual quality. But they tend to push the resolution in areas where it should be lower.

[jknutson2]

I'd almost go as far to say that BITRATE determines RESOLUTION for me.

...

Say I have capture x, runtime: 2 hrs 25 min (for want of a better number). I work out the bitrate (video 4127, audio 224) and straight away I can see that Full D1, even with 2-Pass VBR, is going to be a bit of a stretch if it's medium to high action footage. @ Half D1, I could almost do a CBR and get typically decent (IMO, of course) results, but I err on the side of caution and do a 2-Pass VBR anyway, with Max 6000, ave 4100 and min 1000.

Jim, I have been wrestling with this issue of late and logically, I can agree with you. There's a nice explanation of bitrates vs. resolution on a DVDLab tutorial. The chart makes it easier to understand the relationship.

Since I've been encouraged to use trial and error and see what works for me, I decided to take a DV capture (2 hr 8 m) and convert it to both D1 and half-D1 and compare. I used the same bitrate for both (VBR 2-pass avg 4479). If I play back the two using high action scenes (e.g. camera pans during a fight scene), I can't tell any difference. I've tried comparing both from a playback and a pause frame point of view. Given the chart above, I would expect to see a difference, but I don't. Why would that be?

I understand that logically, I should use half-D1 for quality, but trial and error doesn't help resolve the question or provide any deeper understanding.

[jimmalenko]

Since I've been encouraged to use trial and error and see what works for me, I decided to take a DV capture (2 hr 8 m) and convert it to both D1 and half-D1 and compare. I used the same bitrate for both (VBR 2-pass avg 4479). If I play back the two using high action scenes (e.g. camera pans during a fight scene), I can't tell any difference. I've tried comparing both from a playback and a pause frame point of view. Given the chart above, I would expect to see a difference, but I don't. Why would that be?

I understand that logically, I should use half-D1 for quality, but trial and error doesn't help resolve the question or provide any deeper understanding.

Without knowing exactly why that would be myself, I can only offer these possibilities:

- source - if your footage has a nice mix of high action and low action, then your encoder can bump up the bitrate to suit the high action and conversely lower it for the low action scenes - that's essentially the nature of 2-Pass VBR, and to an extent, the CQ mode is similar, but a little different again. It may well be in this case that your encoder can suitably handle your source sufficiently given the bitrate parameters you have set, and so the end results are indistguishable to your eyes. BTW, when I meant medium to high action footage, I'm talking throughout the entire footage, such as sporting events and whatnot. These will really give your encoder a good workout, believe me

- playback device - If you're watching these on a computer screen, or even on, say, a 27" TV there's a fiar chance you may not see any noticable difference. However, put these into a projector setup where the viewable area ends up being about 6ft X 4ft, and I'm certain that you'll see a difference.