XViD oversized using koepis binaries

[SolarBurn]

I'm using the 23/9 binaries off koepis website to encode my movie using the linear scale curve thingie. First time I tried it it was fine used the same values as I did with stats reader and it came out great. However I then did a 2CD encode using H.263 for the first pass and modulated for the second and it comes out around 200-300mb too big any ideas? I'm just doing the whole movie over again with MPEG quantisers atm

K

[Koepi]

Please use the search - this question has been answered SEVERAL times.

Your settings are too restrictive so XviD can't compensate for bitrate usage. Your credits settings could be wrong. (Most likely your quantizer restrictions are too hard).

Regards,
Koepi

[SolarBurn]

sorry must have missed them, I did look, must be blind, I 'll try again

[SolarBurn]

I still can't get it to work.
Settings
Quatization I/P both 1min 31max
0% iframe boost, distance 250 reduction 30%
High/low 0
payback proportionally 250
credits on the last 5000 frames

first pass h.263 second pass modulated

I set the file size to 1077672 and it comes out at 1240436

Any suggestions?

[iago]

Originally posted by SolarBurn
I still can't get it to work.
Settings
Quatization I/P both 1min 31max ...@SolarBurn

Set Min I-frame and Min P-frame quantizers as 2, not 1!

And please think and check everything in your encoding parameters twice before starting such a thread, the name of which simply doesn't reflect the truth.

iago

[Koepi]

You have to set the credits quantizer range and quant heighth (and activate) them on first pass, too.
Sounds to me like you use fixed quant credits without using them in the first pass (?).

[SolarBurn]

I never touch the credits settings after the first pass, and I re-encoded the 1st pass file with the new quantizers. I set them in the 2nd pass mode as they are greyed out for me in first pass mode and then went back to the first pass mode to encode the stats. Will do it all again with the quantizers set to 2-31

[SolarBurn]

with quantizers set to 2-31 and h.263/modulated the second pass is now 1189736. I did the same encode with the same settings only with the first and second pass set to mpeg and the second pass is 1077726 which is pretty much the right size. So why is h263/modulated not working?

P.S. The credits are coming it at just below 5Mb for 3:09 minutes which seems fine

[Koepi]

You're out sir.

"why is it not working?" - because you still setup something wrong.It works perfectly. Just not for you. So the codec is flawed. *caugh* Your assumptions are mistaken.

Gna.
As iago wrote - don't accuse others to be the error-cause if it's in fact your setup.

Koepi

[SolarBurn]

Sorry I didn't mean to blame anyone I was just looking for a solution, is there any settings that could change when I swap the codec over from h263/modulated to mpeg/mpeg that aren't obvious, submenus or anything? The first pass size for the h263 mode is only 974Mb would this affect how the codec behaves if I ask it to create a larger file in modulated mode?

Sorry

SB

[-h]

Sorry I didn't mean to blame anyone I was just looking for a solution, is there any settings that could change when I swap the codec over from h263/modulated to mpeg/mpeg that aren't obvious, submenus or anything? The first pass size for the h263 mode is only 974Mb would this affect how the codec behaves if I ask it to create a larger file in modulated mode?

You want the second pass file size to be larger than the first pass? That's a bad idea, since it can only be achieved by using quant=1 frames, which wreak havoc with the overflow and force some frames to look worse than the original quant=2.

I'd just hang onto the first pass file - quant=2 looks pretty good!

-h

[SolarBurn]

Well I'm gonna use the mpeg/mpeg encode I did, I didn't notice the first pass size until I looked at it just now. But I just wanna find out what's going wrong with the h263/modulated encode that I did, whether it be me using the wrong settings or the codec getting confused somehow, or whatever variable that has been introduced into the system which has caused the erroneous output so that it doesn't happen again.

[SolarBurn]

I did another h263/modulated enocode with the same settings only this time I set the filesize to about 550mb and it came out as the correct size. So it would seem that it only misses the target when it is set to be larger than the first pass size. Are there any settings that would alleviate this?

[Koepi]

I didn't see that at all that you want to increase the size. This is _impossible_ to achieve, at least to achieve it within desired filesize.
Use another quantizer type(as you did with MPEG), use a bigger resize (e.g. 720x304 instead of 640x272),... the usual stuff.

Regards,
Koepi

[SolarBurn]

I didn't realise at the time either :) Would a warning built into the codec be possible? just to stop dumbasses like me posting stupid threads like this? TDX forbids me to use a higher resolution ;) but I'll probably do it over again anyway with the new binaries with the too_small_limit fixed, I didn't want to change the build just incase it was related to that, might see if mod_hq does the same thing aswell

SB