Bitrates: CQmatic Bug Reports - Page 3

Krassi · #41 07-23-2003, 05:01 AM

Quote:

Originally Posted by Krassi

Right now i'm redoing with Start-CQ=60 to give you more details about sample size etc.

CQ=60.00: Difference: 0.862563
CQ=51.75: Difference: 0.974991
CQ=51.75: Difference: 0.974991
...
m1v.Sample-filesize with 51.75-CQ: 2,392,064 Bytes
m1v-filesize with 51.75-CQ: 3,973,120 Bytes

Any idea

Fiz · #42 07-23-2003, 05:08 AM

@krassi,

I have the same problem starting with a CQ of 60.

Code:

CQMatic Versoin 1.0 Beta 1
Copyright Softronex Corporation, 2003
All rights reserved.
Ready!
Project: E:\Encodes\test.tpr
Creating CQMatic.tpr
E:\Encodes\test.m2v
Execute.
Movie Time: 98
Average Bitrate: 996
Executing Prediction Phase...
Using CQ of 60.00
Encoder started...
Encoder end.
Difference = 1.126464
Using CQ of 67.59
Encoder started...
Encoder end.
Difference = 1.030481
Using CQ of 67.59
Encoder started...
Encoder end.
Difference = 1.030481
Using CQ of 67.59
Encoder started...
Process canceled

I cancelled it in the end. I haven't tried by starting with a different base CQ, I'll do that now.

Fiz.

Krassi · #43 07-23-2003, 08:38 AM

Quote:

Originally Posted by Fiz

I cancelled it in the end. I haven't tried by starting with a different base CQ, I'll do that now.

Got it working with another CQ at start

J-Wo · #44 07-23-2003, 08:52 AM

I'm having the same problem as Krassi and Fiz where after 2 prediction samples, CQMatic keeps sampling at the same CQ! I find that if I cancel and restart with a CQ closer to the "stuck" value, that prediction completes as normal. Very weird bug.

DKruskie · #45 07-23-2003, 08:58 AM

I'm having the same problem also with the resampling of the same cq after 2 tries.

David

NismoSX · #46 07-23-2003, 10:37 AM

Hey! Same problem with repeating CQ here too!

kwag · #47 07-23-2003, 11:35 AM

Working on it ....

-kwag

kwag · #48 07-23-2003, 11:45 AM

Please download this: www.kvcd.net/cqmatic-test.exe
Try to use the same starting CQ that continued in a loop.
See if this fixes that.

-kwag

Krassi · #49 07-23-2003, 11:55 AM

Quote:

Originally Posted by kwag

See if this fixes that.

The behavior is different

. The second CQ was 55.88, third was 53.38.

kwag · #50 07-23-2003, 12:21 PM

Quote:

Originally Posted by Krassi

Quote:

Originally Posted by kwag

See if this fixes that.

The behavior is different

. The second CQ was 55.88, third was 53.38.

Great

sorry

The condition shouldn't repeat

I'm updating the regular cqmatic.exe to the same version as that test file now ( BETA 2 )

-kwag

Krassi · #51 07-23-2003, 12:23 PM

I'm sorry to inform you that its still there. It only needed about 10 tries to find the same 51.75-value where it hangs. Sorry.

kwag · #52 07-23-2003, 12:30 PM

Quote:

Originally Posted by Krassi

I'm sorry to inform you that its still there. It only needed about 10 tries to find the same 51.75-value where it hangs. Sorry.

Thanks

Priority #1 to fix today
Stay tuned

-kwag

kwag · #53 07-23-2003, 01:30 PM

Here it is. www.kvcd.net/cqmatic-1.0b3.exe
Please try it again

I believe that if you start with the CQ that always continued to loop from the start, that's the actual CQ to encode, because it's within the precision boundaries. But

was running another iteration, instead of exiting at that point

I can't test it, because I haven't seen that condition in my encodes

So please give us feedback

-kwag

Krassi · #54 07-23-2003, 01:39 PM

Quote:

Originally Posted by kwag

I can't test it, because I haven't seen that condition in my encodes

So please give us feedback

You bet

Doing a test right now. Second CQ is 57.24, third 55.86...56.18...final 56.18.
Have a beer, Kwag

Well, i have to double-check the new CQ. Its a bit higher than before (56.18 instead of 51.01). Maybe its more accurate now

.
Thanks Kwag

kwag · #55 07-23-2003, 02:01 PM

Quote:

Originally Posted by Krassi

Quote:

Originally Posted by kwag

I can't test it, because I haven't seen that condition in my encodes

So please give us feedback

You bet

Doing a test right now. Second CQ is 57.24, third 55.86...56.18...final 56.18.
Have a beer, Kwag

Well, i have to double-check the new CQ. Its a bit higher than before (56.18 instead of 51.01). Maybe its more accurate now

.
Thanks Kwag

Thanks

I'll post BETA 3 now

I think that bug was totally squashed

-kwag

Krassi · #56 07-23-2003, 03:48 PM

Quote:

Originally Posted by kwag

I think that bug was totally squashed

Thats also what i thougth. Man, this bug has a real big armor

.
It was coming back when encoding a long movie (not the one i encoded previously

.
Maybe this has something to do with start-CQ of 60 or with PAL

EDIT: This time Difference in log doesn't stay at the same value with the same CQ, it varies a little bit.

kwag · #57 07-23-2003, 04:06 PM

Quote:

Originally Posted by Krassi

Quote:

Originally Posted by kwag

I think that bug was totally squashed

Thats also what i thougth. Man, this bug has a real big armor

Do this mean you are experiencing the problem again

With BETA 3

Krassi · #58 07-23-2003, 04:09 PM

Quote:

Originally Posted by kwag

Do this mean you are experiencing the problem again

With BETA 3

Yep, exactly. With Beta 3

.

kwag · #59 07-23-2003, 04:37 PM

http://www.kvcd.net/cqmatic-1.0b4.exe
Grab it, and let me know

-kwag

NismoSX · #60 07-23-2003, 06:43 PM

Is it normal that one prediction cycle is taking almost 20 minutes to complete? This with a 140 min. movie. It's going real slooooow....
Using the same script in ToK only takes 3-4 minutes on the first cycle.
This with a XP 2500 Barton. (Just upgraded)