Latest VfW Version

Will the new x264 VfW's include a decoder also?

Will the new x264 VfW's include a decoder also? yes, im using it right now i tried a few things im now trying with deblocking with both on 1 also increased the bitrate abit but its a big movie so encoding takes long total about 24 hours for both passes.I asume ur encoding with gordianknot, if so, then just clip ur movie and just encode about 1-2 mins of video, it should take less than 10 mins, can you post screenshots and more info , like the size and bitrate of the video, and your settings, i might be able to help, im currently working on this one one which has ur exact same problem and i've got it to look pretty good

think is with every encode its a different part of the movie. and its not reallu visable when you just make a screen shot cos most of the duration of the movie is drawn and some parts are moving i didnt check my last encode but will try to tomorrow. i got most settings of the guide the same. i used different bitrate 1000+ i tried its about a 830 mb or so at 1200 bitrate i use the hiest settings unhex etc.

Version 5_736bm and 5_736bm_AQ_0.47 of x264vfw Changes: - Added colorspace conversion (removed from x264 svn-733) - Added support for "tesa" motion estimation - x264 core updated to svn-736

nice a new release going to try that tomorrow.

well..this are the settings that got rid of (98% of blocking in a "live" movie (not animated) in an encode I did, it had heavy blocks on dark areas or scenes with almost no motion or no motion at all try this and tell me if u see any improvements bitarte around 800kb/s I-frame boost= 35 b-frame reduction= 10 bitrate variability= 30 min qp= 4 max qp= 31 max qp step= 5 scene cut threshold= 20 ----------------------------------- all partitions exceot 4x4 intra max ref b-frames= 5 mv prediction= auto --------------------------------- 6 RDO umh range 32 max ref frames, 3 mixed refs trellis 2 deblocking -1, -1 and AQ at defaults

i'll use those settings i will only use a higher bitrate 1200 and use 7 instead of 6rdo. i also shall update to the la-*test*-('") version from bugmaster and see how it goes. and i use uneven -multi hex

Bug master what are those short names in the methode. dia = diamond search hex = hex.. umh = uneven multi hex esa = ?? tesa = ?? and what is the best to use then ? diamond is to heavy and takes to long next best is uneven-multi-hex? got another problem i got all settings the same in pass 1 and pass 2 but i get a error : different number of b-frames then 1st pass (5 vs 3) i had max consecutive 5 and max ref frames 3 but it didnt want to encode it i had that on both passes when i change max consecutive to 3 on both passes it started the encode and its encoding now

Bug master what are those short names in the methode. dia = diamond search hex = hex.. umh = uneven multi hex esa = ?? tesa = ?? and what is the best to use then ? diamond is to heavy and takes to long next best is uneven-multi-hex? They all described in hint for this option: - dia: diamond search, radius 1 (fast) - hex: hexagonal search, radius 2 - umh: uneven multi-hexagon search - esa: exhaustive search - tesa: hadamard exhaustive search (slow) I prefer to use "umh" but if you want addition quality increase (about +1 PSNR) at cost of speed I suggest use "tesa" (its quality greatly depend on range; I usually use default range = 16). As for "esa" it also slower than "umh" but don't lead to significant quality increase. got another problem i got all settings the same in pass 1 and pass 2 but i get a error : different number of b-frames then 1st pass (5 vs 3) i had max consecutive 5 and max ref frames 3 but it didnt want to encode it i had that on both passes when i change max consecutive to 3 on both passes it started the encode and its encoding now Are you sure you use same settings at pass 1 and pass 2? Due to this message I think not. You use 3 B-frames at pass 1 and 5 B-frames at pass 2 and that is why you get this error.

yea i triple checked it that i used the same settings. when i like i said put the max consecutive from 5 to 3 on pass 1 and pass 2 than it starts with the encode i have max ref frames on 3.

ThaNeko If this problem is repeatable and not depends on source (so I can repeat it with my sources) than I need version of x264vfw (also standard or experimental) and full set of options for pass 1 and pass 2 (would be good if you can provide them as two command lines; or as registry files of HKCU\Software\GNU\x264). P.S. Do anybody other experience such problem?

i used the one of 28 jan 2008 i also tried the experiment one of 30 jan the settings are in the screen shot in the next post i only used a other bitrate of 1200 and method 7. i only tried it on 1 source i use gnot :) the problem is repeatable for me here. i can make screen shots of my settings and all and the error. i shall add those later because im not at home at i will be in 2 hours or so then i will make the screen shots and post them too. i got pass 1 and pass 2 exactly the same.

as i said a screen shot i put them together so its easier :) i tried to put the method on tesa and didn't change a thing and it started to encode. so so far i only get it with what is in the screen shot. when i have method on umh.

Use as referenses will not be for 5 B frames, when referense frames max = 3! 3<5! 5 B frames can not 5 referense B frames, but option "Use as referenses" its realized! :) Fuf! sorry for my english

so max consecutive cant be more then max ref frames? it only did that on umh i tried it on tesa and it started the encode.

View Full Version: Latest VfW Version