2nd .bin/cue half the size as 1st .bin/cue

[factor2000]

Sorry if this is duplicated--have tried to find some similar problem but all involve a second half which is too LARGE. My problem is a second which is too small:

Using DVD2SVCD1121/TMPGEnc258, I rendered a 2 hour movie as I have done about 100 times before.

It rendered OK, but when it cut the segments, the first segment was right on the money--about 800mb. The second was about 1/2 that size--less than 400mb. Both .mpgs played OK in PowerDVD--no missing parts--but the first segment was over 80 min while the second was about 50min. TMPGEnc used a CQ value of about 82.

What went wrong and with which program? I do not recall making any settings changes other than the bitrate, which I always set the bracket at about 2 minutes longer than the movie. (e.g., if the movie is 115 minutes, I set the 2 disk/800mb bracket to about 100-117 minutes)

Thx for any advice. Yes, I have used CCE but prefer TMPG cuz it shows progress with actual frames.

edited by mod This is a forum not a place to request help by email. Mark

[winxi]

Hello!

It´s the same problem as with the last image being too large. Using TMPGenc constant quality you will always have the risk of getting too big or too small images.
Use either TMPGenc CBR, 2 pass VBR or CCE multipass to be get a 100 % correct image size.

greets, winxi

[factor2000]

May I go further with you on this subject?

I guess inquiring minds want to know--HOWCUM? I set TMPGEnc to an initial CQ value of 80 and motion search-highest/very slow. I actually prefer a CQ setting to a VBR setting as, IMHO, CQ gives the best rendering, albeit, might take a little longer.

TMP samples the movie (in this case, about 4-5 times--sometimes using the identical clip twice) and fine tunes the CQ settings. In the movie I did, it assigned a CQ of 82 and ended up with a .mpv file of about 1.3gb, which it splits into 2 bin/cues of 800mb & 500mb each. Obviously, the CQ should have been set a few points higher.

I would have thought that TMPG would be smarter than that and that with so many samples, would be closer to the actual CQ.

Do you have any idea what it is about TMPGEnc that allows it to be right on the money most of the time, yet miss the mark by such a large margin on others? Do you know if this can be correctable?

Thx
edited by mod

[markrb]

Once I again I ask you not to post an email address. This is a forum and as such replies belong here. If somebody wishes to email you they can click on your profile. By placing an email in the body of the message it appears as if you are asking for an email response.

Please do not do it again.

Simply put using CQ it is very difficult to predict the final size. Using CBR or VBR is much easier to predict.

In short use CQ at the chance of wasting time.

Mark