You don't :!:
Just load any project file you have around, and it will clear it :!:
Remember, it's not CQMatic that does that. It's TMPEG, which remembers the last "state" it was in :!:
So when you do prediction only, and TMPEG exits, the next time you run TMPEG, it will read from the registry whatever the last run was. In this case, the last prediction, which is set with the prediction time slices :D

-kwag

@ all.. :)

RE: CQM v1.1.07 .. ..

I can't say how painfully, and excrusiating it has ben to finally figure this
all out w/ CQM and encodeing w/ ES..

This may be a bug, but NOT a bug hehe..

If you don't leave the Audio Source: [ ] field cleared, and have some
audio filename inside, you'll be encoding the whole test clip (all the frames)
I've restled with this for days, but just coudln't put two together.

Now, I have, but, I'm still having some minor issues w/ hitting the first
encode w/ CQM and thoes short frames.. every so often, I TMPG will pop
up real quick, and CQM will finish. I guess, something is not cleared 100%
percent of the time :!: But, w/ a little more patiance, I'll get through each.

Just wanted to point out, the you have to make sure you Audio Source
filed is left bank, thats all :)

Well, I'm finally able to use v1.1.07 now.

-vhelp

It's a bug in TMPEG.
The audio field must be cleared.
Try it directly with TMPEG (Without CQM) and you'll get the same result ;)

-kwag

oh, 8O ok then.. tanx!!

FWIW, as you know, I'm currently testing w/ TMPG v2.520
-vhelp

.
.
Bug or not.. that is the question..

RE: Short vs. Complete length vs. Total PASS'es

Yes, that is the question :!:

When I encode w/ shorter length sources ie, 1 VOB (usually 20+ minutes)
I have these LONG number of passes. My goal was to use smaller size
source files, to make shorter wating time, CQM prediction test runs, but
this is proving to be an error on my part. Its not happening.

Is there a missing link w/ this shorter vs. longer vs. total passes ??
Or, are there any other advises or tips to give, to reduce the amount of
time. Just trying to cut corners, that's all :)

Hard drive space is limited for me. I have it crammed w/ other VOBs for
testing and stuff. I'd hate to have to delete them for every test.

-vhelp

When you are prediction a very short clip ( CQ above 80 ), or an extremely long movie (CQ below 50), TMPEG is working on a very "flat" scale. Not linear at all :!:
That's why it takes so long. Because a change in CQ value, produces almost NO file size difference. That's the problem, and there's no way around it, but wait, and wait, and wait :lol:

-kwag

.
.
ok then.. :grrr:

-vhelp

hi Kwag..

I don't know if this is a bug or glitch or just the way CQM works..

When I do short 'prediction' test w/ CQM like 30 to 20 minutes or so, it can
take a very long time, and many iteratatins (passes) before reaching it's
disired CQ.

But, if I feed it a full length source, such as those I fed in earlier, ie, "Dogma"
128 minute movie, and it only takes 2 passes and 12 minutes to complete,
is the suppose to happen this way ??

Or, is it something else altogether ??

hehe, and is this Linear too ??
-vhelp

Yes. It's the different sampling sizes I take that affect the time. The problem is really the "Seeking" to non-sequential frames. When TMPEG requests a "jump to" a non sequential frame, there's a time lag before the actual frame is delivered. Multiply this "lag" by all the samples taken, and there's your delay ;)

-kwag

So, the real benefit is WHOLE movie length, not partials :wink:

Ok, gotcha!!

I'm currently contenplating another source for testing. Do you have any
preference for any of those one's maybe that you did, but that you're curious of what others' are getting from such ??

Please let me know, as I've cleaned up a few more gigs and I'm ready for
another reasonable testing stage :!:

-vhelp

Yes, because CQ also falls in the linear CQ range on longer movies. Not on the "flat" (Low CQ) areas :) No I don't have any preferences, but the more testing, the better :D

-kwag

hi,

not sure if this is a bug in CQM or in the setup on my computer (have reinstalled anything last weekend) ...

I've want to do the prediction for a movie captured via tv-card.
Saving a tmpgenc project (as text project file) and loading it into CQM, entering the movie time (77 min) and avg bitrate (1260), switched to prediction only and pressed the start button

CQM tok only 19 minutes to finish the prediction using 8 cycles but no output file (m1v) has been created. Not to say that I understand that all, but the log output of CQM looks strange for me. CQ is raising but the file size difference is the same in every cycle (I thing this is because no sample has been encoded, but why?). While processing the tmpgenc window is show only a fraction of a second for each cycle.

Btw: If i start encoding the movie in tmpgenc without CQM, tmpgenc displays an calculated total time of about 5 hours. The avs script used does not contain a call to a Sampler like function and the tmpgenc source range is cleared.

Any ideas ?

Thx.

Hi holgerschlegel,

Something is not set correctly in your .avs script or in your project (.tpr) file.
The file size difference can never be negative, so TMPEG is bombing out somewhere. Try to load your project in TMPEG, and hit "start" to encode. See what error you get in TMPEG.

-kwag

I've "test-started" the encode in TMPEG before saving the project to the .tpr file and TMPEG displays a remaining time of about 5 hours after a few minutes of encode. Havn't tried to encode the whole movie, but the first minutes are working. I'll encode the complete movie in the afternoon.

Holger

I have the same problem like holgerschlegel! I wanted to encode 3 different movies now and the Current CQ always was 89,88! I dont know where the Problem is, but I think it isn't a problem of CQMatic!

Just change your path in your avs script to something simple like:

Mpeg2Source("F:\Videos\movie.d2v")

@kwag: After getting past this small obstacle with the avs script Cqmatic is working very nicely. When my Mpeg2Source line was more complicated I kept getting the same CQ. 89.88!!!! Maybe you should mention something about it in the instructions. Just a suggestion. Anyway thanks for the great program. Finally accurate prediction! :lol: :lol: :lol:


DVision

Hi DVision, and welcome :)
Check out version 1.1.11 ;)

-kwag

This time I worked without any AVS-Script and loaded the .d2v into TMPGEnc! The same Problem! I will try the new CQMatic Version now!

It's the same in the new Version!

I don't know where the problem is!!! :cry: :cry: :cry:

After I looked in my log again, i saw that the "Movie Time" and the "Average bitrate" are changed! But I'm sure, I made all right! Very stupid!