Good ideas. I will work on them next week-end as I have a new work and lesser time.


Yes. You can check it easily once finished the sample(s) encoding, do change your audio values, and all other values will change at same time.
Of course, I think across next week end, I will end to fix all this issues.


I will change it also. It is the easier thing Im going to do.

Take your time :bowdown:

Noob question about drag&drop (don't work in my case)
 

Hi,

First of all, I know what is drag&drop, and a windows interface ! :P

The problem is that I can't add more than one file to procalc: it always sustitute the only file I have in the list...

And one small bug: if you change a path (hc path for example), i don't take effect until you restart procalc.

Thanks,
Fabrice

Re: Noob question about drag&drop (don't work in my case
 

I added my 8 files in a single drag'n'drop operation and that works.
As there is no "clear the list" button, I think that Luis do not think about adding the files one by one.

But I drop the app because of some results I found weird. For instance in full mode I dont understand why a movie that is twice long as the other should have a bitrate doubled ! There is really no reason to do that.
I have to verify more closely the "complexity" mode but I will do that with fewer movies.

Re: Noob question about drag&drop (don't work in my case
 

In fact there is and you know why there is phil ;-).
Let's think of 2 movies.
Movie A - 120mins. mainly very low action movie with a lot of long dark scenes.
Movie B - 90mins. mainly very high action movie with a lot of very bright scenes.
It is reasonable to say that movie B will require more space on the DVD to have the same visual quality of movie A.
So maybe you could end up with ~2.8Gb (minus audio size) for movie B and ~1.5Gb (minus audio size) for movie A.
I have seen this happen with a couple of movies of mine.
And visually compared, they seem to have the same visual quality when I watch them on a 16:9 32" TV.

Re: Noob question about drag&drop (don't work in my case
 

The problem is that you do not understand what really happens.

I have two movies that have the SAME complexity (pure theory).
One is 120 minutes, the other is 60.

Let imagine that in pure "time" mode, the avg bitrate should be 3000.
With this, movie1 will use 2 GB, movie2 will use 1GB (still pure theory, this numbers are not the real ones).

In pure complexity mode, the two movies will receive also 3000 as their complexity is the same. File size is still 2GB and 1GB.

In "full" mode, ASPA will allocate a bitrate to movie1 that is the double than the one for movie2 !!
(4000 and 2000).

With this movie1 will take 2.5GB and movie1 will have only 0.5.
But they still have the same complexity. Movie2 has no reason to be handicaped like this.

This is what happens to my movies yesterday : 1 episode is 1h23, all 7 others are 42 minutes. Complexity is quite the same (let say that is 1 is the complexity of first episode, complexity for others is in the range 0.8-1.2).

• Average bitate in time mode was 1370.
  In complexity mode, all goes from 1100 to 1500.
  In full mode, first episode receive a bitrate of 2001 and all the others had less than 1000.

Do you still think that is normal ?

According me, "full mode" is a bad idea, or there is something badly implemented into it.

And, and that is what complexity mode will do. Length of the movie don't have to enter into consideration.

Putting in more straight words the question is : why a 2 hours movies should receive a better bitrate than a 1 hour movie ?
This can be easily tested :
Take a movie, create these scipts :
Enter this into ASPA and see what is found in each mode.

Now do a calc with three time the first script. The result on the DVD of script1 + script2 or 3x script1 is the same. But ASPA won't give you at all the same quality to your two DVDs.

Phil,

could you test my spreadsheet with your samples? Encode a 3% sample of each episode with SelectRangeEvery(500,15) and fill in the values. Number of frames means the total number of frames in the clip. You need to fill in that, the avg bitrate of the sample and the audio bitrate.

http://www.saunalahti.fi/sainki/kbps_eng_multi.xls

(Of course, you can test your theory as well :wink: )

I'm currently doing the test I suggested in my previous post, and I will do yours just after (or this evening).

Note: HC reports BIG peaks in its window. Can we trust that ? Currently it shows 18195 as max !!!

And the result of my suggested test are :

Aspa results with script1 + script 2 (movie - minutes - seconds - avg bitrate)
---------------------------------------
Full mode :
script1 40 59 2465 (can you imagine the quality ?)
script2 81 59 5545 :!:

Time mode :
script1 40 59 4519
script2 81 59 4519

Complexity mode :
script1 40 59 4394
script2 81 59 4581 (quite the same as time mode and both values near the same -> that is normal as both movies have the same complexity but the samples are not exactly the same)

Aspa results with 3 times script1 (movie - minutes - seconds - avg bitrate)
---------------------------------------
Full mode :
script1 40 59 4519
script1 40 59 4519
script1 40 59 4519

Time mode :
script1 40 59 4519
script1 40 59 4519
script1 40 59 4519

Complexity mode :
script1 40 59 4519
script1 40 59 4519
script1 40 59 4519

(as you can see, the formula used by Luis are correct -> all the figures are the same, this is not an error :-))

If CQ_maxbitrate is the mode that is being used, it's a bug. The encoder should automatically raise the quantizer if the set max bitrate is exceeded. CQ_maxbitrate should be used anyway because the avg bitrate of the sample might not be correct otherwise.

Yep, now I see what you mean ;-).
Could you check Boulder's spreadsheet too?
That's what I've been using and it hasn't failed me once.
Just so that we can compare both methods.
That is, if you have time and a bit of patience :).
Cheers

I never dig into HC and I use the ini modifued by luis :
I guess the cq_maxbitrate mode is used ?

Beside this, does "SelectRangeEvery(500,15)" select exactly a 3% sample length for a NTSC 29.97 source ? or is it only for PAL source ?

CQ_maxbitrate does not compensate bitrate allocation when it has to suddenly raise the Q for keeping the maxbitrate.
So if you would encode the same movie clip in regular CQ and with CQ_maxbitrate, both should end up with the same avg bitrate.
And that doesn't happen because CQ_max does not use the spare bits where there is no need for more bitrate.
Hank has that in his to-do list but I guess he is now focused on compliancy and he will implement such features later on.
So regular CQ is sharper on the avg bitrate needed but unfortunately it doesn't take care of spikes.
That's why we have to use CQ_max.
Cheers

No, a CQ mode without max bitrate capping is being used - there is a space before *cq_maxbitrate which disables it.
Just tested, it does. Of course, it's not exactly 3% but very close to it. If you use a GOP length of 12 frames, it would be (400,12) I think as 1% is (1200,12).

Could you post a link where Hank says that? The pdf docs state that the quantizer is raised temporarily if max_bitrate is met. I've never seen CQ go above the max but very, very close to it. In fact, for DVD compliance, the max bitrate must be respected all the time.

Sorry Boulder I had a typo.
Please re-read my edited post.
Cheers

Okay. But now I don't understand your speadsheat : am I supposed to calculate the average bitrate of the encoded sample by myself ?
We use Excel and it should do that better than me. Currently I can give you the time, fps and filesize of the sample, that should be enought :). Can you do the calc yourself with these informations ? I don't have time for understanding your formulas now (i'm at the office ;))

Just tell me what data you need.

Exactly, and that's why CQ_maxbitrate should be used when encoding the samples as well. In fact, the settings should be as close as possible to the actual encoding.

I use Bitrate Viewer or MPEG Stream Analyzer to get the avg bitrate. The method doesn't matter as long as you use the same way for all files that end up on the same DVD. Automatic calculating in Excel doesn't work because the samples aren't exactly 1%, 3% etc.
Yes, bring them on :wink:

I don't have either of them on my PC. Probably on my home's one. This will have to wait for tonight.

Hum... length in minute of the sample, the fps and the filesize, that's all what Excell needs. Having the length is as easy as opening the encoded movie in a media player 8)

If you have time :

script1 :
total length : 40'59" (73704 frames)
sample length : 1'14" (2219 frames)
fps : 29.97

sample filesize : 54.145 MB

script2 :
total length : 81'59" (147408 frames)
sample length : 2'27" (4424 frames)
fps : 29.97

sample filesize : 111.776 MB
(with a peak up to 31450 in HC :D)

For audio bitrate I used 384 in my test with ASPA.

Still it's easier to open the sample in BV or MSA and enter the value in the spreadsheet :wink:
I hope I got it right, avg bitrate 4421kbps for the first one, 4592kbps for the second one. Correlates nicely with the complexities in samples.

http://www.saunalahti.fi/sainki/excel.jpg

Yeah, once you dled, installed (and paid for ?) these tools :D

And that is also close to the ASPA results in "complexity" mode.

I really think that "full" mode is a bad thing (no offence Luis).

That's just one time :wink: Both can be used free of charge, MSA just shows a bouncing logo on the video but that doesn't matter as you just need the figure.

Check the HC.log after you encode the clip.
It will tell you the encoded average bitrate.
Isn't that what you're looking for?
Cheers

That's one option, of course :) Funny it didn't come to my mind even if I played with HC's CQ mode some days ago.

You see, that's what I've been using since I started using your spreadsheet ;-).
I love to see you both working on this multi-movie proportional OPV with HC :D.
Keep it going and I'll try to stay close if I manage to get some time.
Cheers

I don't have any log file. I guess HC does not do it if not requested.

@Phil
The original HCEnc zipfile distributed by Hank with his ftpd, already comes with a file named Template_HC.ini inside.
Change that filename to HC.ini and edit it.
You will see that it have the settings for a logfile.
Edit the path and the name for the logfile.
Open it after you encode search in the end of the logfile and you'll find the figure for the avg bitrate.
BTW, if you keep encoding with that HC.ini, you will see that every encode will be added to that file!
So beware because the file can have the logs for zillions of test encodes :idea:.
Cheers

Also do compare spreadsheet and ProCalc Lite as it should obtain (and do obtain indeed, almost same results, difference only related with floating point values)

Re: Noob question about drag&drop (don't work in my case
 

You should have all your avs files in one folder and drag and drop all at same time. This is the program way.

It is not a fail. It is the behaviour of ProCalc Lite and Boulder's spreasheet also. You don't think in that before, so you didn't see it. I "yes" saw it, but I didn't know enough about it, I could not realize what was the thing that don't worked well (in my mind I mean), so I implemented time and complexity mode, as a ensurance.
My thought was, if we use complexity mode only, the bitrate differences will asign more space to the stream that need it, related, not only with complexity but with movielength also. But I still can't round it completely in my mind.

cq_maxbitrate will give you a stream without spikes. If there is any while encoding, it will cut it to avoid uncompliant DVD results.
The samples are encoded in CQ mode (not cq_maxbitrate), as if I want a complexity analisis, I can't to cut those spikes because the sample wouldnt be representative.

So ProCalc Lite manage HC to encode samples in CQ mode (not cq_maxbitrate).

I didn't interprete that in this way, but if all you are sure I will change CQ for CQ_maxbitrate mode.
But I think that a stream in CQ will have more size that one in CQ_maxbitrate, with same source, despite the avgbitrate be the same.

You don't need the avg bitrate. It is enough with final size as it is an avg bitrate function.
But maybe this is why ProCalc Lite gives you different values (a programtion error) as I used final sample size as value for the calculation.

In my opinion, the complexity should be measured with the final settings in mind. That is, the same matrix, same bitrate boundaries, GOPs etc. That way you can ensure that there'll be a fair result for all clips involved. As Phil's sample showed, there can be huge spikes which can then cause a serious bias towards the clip regarding bitrate. In the final encode, you always have to clamp to the max bitrate anyway :wink:

If you have the framecount and length of your sample, you can obtain avg bitrate with a simple calculation in excell.

Don't worry Phil. In a previous post you can read that could be a programmation error.
What is for you the better way then, complexity mode maybe?
I think fixing full mode it would be still the better way.
I noted (now I explain myself why) that was better to put similar time movies.
Also it is not convenient to put fullscreen together with widescreen, as fullscreen movies eats excesive bitrate.

Hi,

That's true that I didn't try to d&d various selected file... :oops: (And I just see that you put AVSs ... )

I think the problem with the full mode is that you are already using a sample, which have a length proportional to total frame number (5%)
So you encode 2 movies, one 2 time longer than the other, your encoded samples will already consider 2 times more frames, and so, with igual complexity, will have a sample 2 times bigger...

So I think that the complexity mode, which use the sample size, should already take into account the length of the movie and the "complexity", and should be enougth.

Salu2
Fabrice

yes it is.

But Luis there is still NO reason to allocate bitrate according to the length !
Do you have in mind that for the same bitrate, the longer movie will already take more space than the other ? So if you also give to it more bitrate, it's completly amazing.
Return back to my two movies that have the same complexity : with the same bitrate, the 2hours long will take 66% of the DVD and the shorter will have 33% (2/3 - 1/3)
With aspa that gives the double bitrate to the two hours, this one will take 80% of the place (4/5 - 1/5).
Can you tell me what in your mind justify this for two movies with the same complexity ? I want to understand.

Taking two movies that are close in time just reduce this effect but do not justify it.

I think this is included in the notion of complexity but I'm not 100% sure.

That's correct. My determination of complexity doesn't care about aspect ratio, resolution etc. It only cares about how many bits are needed to maintain a similar visual quality level between videos that are to be put on the same disc. A movie with 1.33:1 aspect ratio is actually the same as a movie with 1.78:1 if you keep it anamorphic: both fill the whole 720x576(480) frame. And that's the whole idea why I created the notorious spreadsheet :lol:

I can see the point now. If all here do agree, specially Boulder, I can fix ProCalc Lite in a easy way, just withdrawing ASPA full mode.