View Full Version : AutoGK: latest betas and b-Frames
Robyfly
19th July 2005, 15:31
I hope this in not answered before. I tried to search but didn't find one answer.
Triying to encoding a pal dvd using 2.12 beta with XviD with Sigma/Mediatek support i found that in the resulting avi were present B-Frames
Log:
[11/07/2005 0.10.21] AutoGK 2.12b
[11/07/2005 0.10.21] OS: WinXP (5.1.2600).2
[11/07/2005 0.10.21] Job started.
[11/07/2005 0.10.21] Input dir: D:\Estazione DVD\Alexander2
[11/07/2005 0.10.21] Output file: D:\Estazione DVD\Alexander2\Alexander2.avi
[11/07/2005 0.10.21] Output codec: XviD
[11/07/2005 0.10.21] Audio1: Italiano AC3 6ch
[11/07/2005 0.10.21] Subtitles1: Italiano wide
[11/07/2005 0.10.21] Format: .AVI
[11/07/2005 0.10.21] Target size: 1400Mb
[11/07/2005 0.10.21] Custom resolution settings: minimum width of 640 pixels
[11/07/2005 0.10.21] Custom audio settings: Original
[11/07/2005 0.10.21] Standalone support enabled: MTK/Sigma
[11/07/2005 0.10.21] Started encoding.
[11/07/2005 0.10.21] Demuxing and indexing.
[11/07/2005 0.13.27] Processing file: D:\Estazione DVD\Alexander2\VTS_01_1.VOB
[11/07/2005 0.13.28] Processing file: D:\Estazione DVD\Alexander2\VTS_01_2.VOB
[11/07/2005 0.13.28] Processing file: D:\Estazione DVD\Alexander2\VTS_01_3.VOB
[11/07/2005 0.13.28] Processing file: D:\Estazione DVD\Alexander2\VTS_01_4.VOB
[11/07/2005 0.13.28] Processing file: D:\Estazione DVD\Alexander2\VTS_01_5.VOB
[11/07/2005 0.13.28] Source resolution: 720x576
[11/07/2005 0.13.28] Found PAL source.
[11/07/2005 0.13.28] Source aspect ratio: 16:9
[11/07/2005 0.13.28] Analyzing source.
[11/07/2005 0.19.10] Source has percentage of interlacing in motion areas: 0,08
[11/07/2005 0.19.10] Source is considered to be progressive.
[11/07/2005 0.19.10] Output will contain 134997 frames
[11/07/2005 0.19.10] Preparing subtitles.
[11/07/2005 0.21.07] Audio1 size: 259,616,256 bytes (247.59 Mb)
[11/07/2005 0.21.07] Overhead: 899,968 bytes (0.86 Mb)
[11/07/2005 0.21.07] Video size: 1,207,490,176 bytes (1151.55 Mb)
[11/07/2005 0.21.07] Running compressibility test.
[11/07/2005 0.21.07] Writing the following script to D:\Estazione DVD\Alexander2\agk_tmp\Alexander2_comptest.avs
===========================================================
LoadPlugin("C:\PROGRA~1\AutoGK\DGMPGDec\DGDecode.dll")
LoadPlugin("C:\PROGRA~1\AutoGK\filters\autocrop.dll")
LoadPlugin("C:\PROGRA~1\AutoGK\filters\vsfilter.dll")
LoadPlugin("C:\PROGRA~1\AutoGK\filters\RemoveGrainSSE2.dll")
movie = mpeg2source("D:\Estazione DVD\Alexander2\agk_tmp\Alexander2.d2v")
cropclip = autocrop(movie,mode=0,wmultof=4,hmultof=4,samples=10,aspect=0,threshold=34,samplestartframe=0,leftadd=0,rightadd=0,topadd=0,bottomadd=0)
fixed_aspect = 1.42222222222222
c_width = width(cropclip)
c_height = round(height(cropclip) / fixed_aspect)
input_par = float(c_width)/float(c_height)
out_width = 704
out_height = round(float(out_width) / input_par)
hmod = out_height - (floor(out_height / 16 ) * 16)
out_height = (hmod > 4) ? (out_height + (16 - hmod)) : (out_height - hmod)
new_aspect = (float(out_width) / float(out_height)) / fixed_aspect
autocrop(movie,mode=0,wmultof=4,hmultof=4,samples=10,aspect=new_aspect,threshold=34,samplestartframe=0,leftadd=0,rightadd=0,topadd=0,bottomadd=0)
LanczosResize(out_width,out_height)
RemoveGrain(mode=2)
VobSub("D:\Estazione DVD\Alexander2\agk_tmp\Alexander2.idx")
SelectRangeEvery(300,15)
===========================================================
[11/07/2005 0.28.12] Duration was: 7 minutes 5 seconds
[11/07/2005 0.28.12] Speed was: 15,87 fps.
[11/07/2005 0.28.13] Compressibility percentage is: 68,51
[11/07/2005 0.28.13] Chosen resolution is: 688x288 ( AR: 2,39 )
[11/07/2005 0.28.13] Predicted comptest value is: 70,10
[11/07/2005 0.28.13] Running first pass.
[11/07/2005 0.28.13] Writing the following script to D:\Estazione DVD\Alexander2\agk_tmp\Alexander2_movie.avs
===========================================================
LoadPlugin("C:\PROGRA~1\AutoGK\DGMPGDec\DGDecode.dll")
LoadPlugin("C:\PROGRA~1\AutoGK\filters\autocrop.dll")
LoadPlugin("C:\PROGRA~1\AutoGK\filters\vsfilter.dll")
LoadPlugin("C:\PROGRA~1\AutoGK\filters\RemoveGrainSSE2.dll")
movie = mpeg2source("D:\Estazione DVD\Alexander2\agk_tmp\Alexander2.d2v")
cropclip = autocrop(movie,mode=0,wmultof=4,hmultof=4,samples=10,aspect=0,threshold=34,samplestartframe=0,leftadd=0,rightadd=0,topadd=0,bottomadd=0)
fixed_aspect = 1.42222222222222
c_width = width(cropclip)
c_height = round(height(cropclip) / fixed_aspect)
input_par = float(c_width)/float(c_height)
out_width = 688
out_height = round(float(out_width) / input_par)
hmod = out_height - (floor(out_height / 16 ) * 16)
out_height = (hmod > 4) ? (out_height + (16 - hmod)) : (out_height - hmod)
new_aspect = (float(out_width) / float(out_height)) / fixed_aspect
autocrop(movie,mode=0,wmultof=4,hmultof=4,samples=10,aspect=new_aspect,threshold=34,samplestartframe=0,leftadd=0,rightadd=0,topadd=0,bottomadd=0)
LanczosResize(out_width,out_height)
RemoveGrain(mode=2)
VobSub("D:\Estazione DVD\Alexander2\agk_tmp\Alexander2.idx")
===========================================================
[11/07/2005 1.48.32] Duration was: 1 hour, 20 minutes 18 seconds
[11/07/2005 1.48.32] Speed was: 28,02 fps.
[11/07/2005 1.48.33] Expected quality of first pass size: 68,09
[11/07/2005 1.48.33] Running second pass.
[11/07/2005 3.54.48] Duration was: 2 hours 6 minutes 15 seconds
[11/07/2005 3.54.48] Speed was: 17,82 fps.
[11/07/2005 3.54.49] Splitting video into: D:\Estazione DVD\Alexander2\Alexander2.cd1.avi
[11/07/2005 3.56.18] Splitting video into: D:\Estazione DVD\Alexander2\Alexander2.cd2.avi
[11/07/2005 3.57.48] Job finished. Total time: 3 hours 47 minutes 27 seconds
Unfortunately i lost the DRF report, it was on average 3.50 DRF and B-Frames present, but i made another encoding with version 1.96 and same settings (apart subtitles) here is the DRF report:
DivX DRF Analyzer v0.9.5 Report!
File Name: D:\Estazione DVD\Alexander3\Alexander3.avi
FourCC: XVID
Codec: XviD0037
Resolution: [ Width: 640 Height: 272 ]
Frame Rate: 25.000 frames per second
The Video has 134997 frames [ 01:29:59 ]
Average Frame quality is HIGH [Average DRF/quantizer is 2.90]
Standard Deviation: Quality is HIGH [Std. Deviation is 0.36]
Image Resolution is HIGH
There are NO frame drops ( NO drops is better )
Recomended Resolution: [624x256] (Target DRF/quantizer=2.8)
Performance Caracteristics:
Macroblocks per frame: 680
The Width is multiple of 32
Kilobits per Second: 1741.49
Kilobits per Frame: 69.64
Kilobits per Macroblock: 0.102
Bits per Pixel: 0.41
Frame Type Statistics :
I Frames: 1.33%
P Frames: 98.67%
B Frames: 0.00%
S Frames: 0.00%
N Frames: 0.00%
(More Advanced Codecs use B and S frames)
Frame Quality Statistics :
DRF=1&2: 15636 11.7%
DRF=3: 110452 82.9%
DRF=4: 7109 5.3%
DRF=5: 0 0.0%
DRF=6: 0 0.0%
DRF=7: 0 0.0%
DRF=8: 0 0.0%
DRF=9: 0 0.0%
DRF>9: 0 0.0%
KeyF/DeltaF: 1.35%
KeyDRF<4: 1702
KeyDRF=4: 98
KeyDRF>4: 0
AverageKeyDRF: 2.98
MAXDRF: 4
AverageDRF: 2.90
Deviation: 0.36
The second encoding presents less resolution (640x272) and no B-Frames but better quality, also looking the movie with the pc monitor and better values present in the report. Also i know on some standalones (some Philips for example) B-Frames are not welcome so i ask if it would be possible to disable B-Vops from AutoGK, I tried to disable it via encoder config but AutoGK 2.12 activated it again.
Thank for your attention and for the great program wich give me many good rips with semplicity :)
unskinnyboy
19th July 2005, 17:36
B-frames are not welcome on your standalone? Are you sure? Most standalones are OK with max. cons. BVOPS=2, and almost everything works fine with max. cons. BVOPS=1 (it is even part of the DXN profile and it should be perfectly compatible with a DivX-certified player - Phillips DVP642, a very common player is DivX-certified).
If you still want to disable B-frames go for an outrageous target size and/or reduce the resolution a lot, so that the compressibility test will return a very high value and AutoGK will automatically disable b-frames to avoid undersizing. You cannot force AutoGK to disable b-frames while keeping the current settings.
BigDid
19th July 2005, 18:04
I hope this in not answered before. I tried to search but didn't find one answer.
Triying to encoding a pal dvd using 2.12 beta with XviD with Sigma/Mediatek support i found that in the resulting avi were present B-Frames
Log:
...
[11/07/2005 0.10.21] Target size: 1400Mb
...
[11/07/2005 0.21.07] Audio1 size: 259,616,256 bytes (247.59 Mb)
[11/07/2005 0.21.07] Overhead: 899,968 bytes (0.86 Mb)
[11/07/2005 0.21.07] Video size: 1,207,490,176 bytes (1151.55 Mb)
...
[11/07/2005 0.28.13] Chosen resolution is: 688x288 ( AR: 2,39 )
[11/07/2005 0.28.13] Predicted comptest value is: 70,10
...
Unfortunately i lost the DRF report, it was on average 3.50 DRF and B-Frames present,...
Hello Robyfly,
It's kind of difficult to compare without the second log but I see that in the first one you have a resolution of 688x288. I have discovered that my SAP (H&B 3220 with a MTK 1389DE) stutters or choke with to high resolution or bitrate (with B-frames)
Now I try not to go above 1500kbp/s or 640*xxx
... i made another encoding with version 1.96 and same settings (apart subtitles) here is the DRF report:
[...
Resolution: [ Width: 640 Height: 272 ]
Frame Rate: 25.000 frames per second
The Video has 134997 frames [ 01:29:59 ]
Average Frame quality is HIGH [Average DRF/quantizer is 2.90]
...
Kilobits per Second: 1741.49
...
Frame Type Statistics :
I Frames: 1.33%
P Frames: 98.67%
B Frames: 0.00%
S Frames: 0.00%
N Frames: 0.00%
...
The second encoding presents less resolution (640x272) and no B-Frames but better quality, also looking the movie with the pc monitor and better values present in the report. Also i know on some standalones (some Philips for example) B-Frames are not welcome ...
Is it the same source??
Now concerning B-frames, I have no knowledge of an option to turn it on or off, but I know how to achieve same results (you have done it also without knowing it)
You need to "sature" the bitrate so that AGK will disable B-frame to achieve the desired final size.
Use your first source, use max width=640 and make tests
Begin with 1400mb final size and let the encode to the end of the first pass ( or the second, don't remember atm); AGK will let you know if it has disabled B-frames to attain desired size;
If not, raise the size by 200mb steps if possible.
If not possible you should obtain identical results in setting the max size (let's say it is 1400Mb) and decrease the max width below 640*xxx
It would be nice to know if the problems (stuttering or choking) are coming from B-frames or too much bitrate, thanks to report back;
Did
Edit: been quite long to write this reply, hello to unskinny boy :)
Robyfly
20th July 2005, 01:03
First thank you for the answers :)
Here is the log of the second encoding made with version 1.96:
[12/07/2005 0.21.20] AutoGK 1.96
[12/07/2005 0.21.20] OS: WinXP (5.1.2600).2
[12/07/2005 0.21.20] Job started.
[12/07/2005 0.21.20] Input dir: D:\Estazione DVD\Alexander3
[12/07/2005 0.21.20] Output file: D:\Estazione DVD\Alexander3\Alexander3.avi
[12/07/2005 0.21.20] Audio: Italiano
[12/07/2005 0.21.20] Subtitles: none
[12/07/2005 0.21.20] Codec: XviD
[12/07/2005 0.21.20] Target size: 1400Mb
[12/07/2005 0.21.20] Custom resolution settings: minimum width of 640 pixels
[12/07/2005 0.21.20] Custom audio settings: AC3
[12/07/2005 0.21.20] Started encoding.
[12/07/2005 0.21.20] Demuxing and indexing.
[12/07/2005 0.23.30] Processing file: D:\Estazione DVD\Alexander3\VTS_01_1.VOB
[12/07/2005 0.23.30] Processing file: D:\Estazione DVD\Alexander3\VTS_01_2.VOB
[12/07/2005 0.23.30] Processing file: D:\Estazione DVD\Alexander3\VTS_01_3.VOB
[12/07/2005 0.23.30] Processing file: D:\Estazione DVD\Alexander3\VTS_01_4.VOB
[12/07/2005 0.23.30] Processing file: D:\Estazione DVD\Alexander3\VTS_01_5.VOB
[12/07/2005 0.23.30] Source aspect ratio: 16:9
[12/07/2005 0.23.30] Source resolution: 720x576
[12/07/2005 0.23.30] Found PAL source.
[12/07/2005 0.23.30] Analyzing source.
[12/07/2005 0.28.55] Source has percentage of interlacing in motion areas: 0,08
[12/07/2005 0.28.55] Source is considered to be progressive.
[12/07/2005 0.28.56] Output will contain 134997 frames
[12/07/2005 0.28.56] Using bitrate spikes control.
[12/07/2005 0.28.56] Audio size: 259,616,256 bytes (247.59 Mb)
[12/07/2005 0.28.56] Overhead: 1,603,072 bytes (1.53 Mb)
[12/07/2005 0.28.56] Video size: 1,206,787,072 bytes (1150.88 Mb)
[12/07/2005 0.28.56] Running compressibility test.
[12/07/2005 0.28.56] Writing the following script to D:\Estazione DVD\Alexander3\agk_tmp\Alexander3_comptest.avs
===========================================================
LoadPlugin("C:\PROGRA~1\AutoGK\DGMPGDec\DGDecode.dll")
LoadPlugin("C:\PROGRA~1\AutoGK\filters\autocrop.dll")
LoadPlugin("C:\PROGRA~1\AutoGK\filters\RemoveGrainSSE2.dll")
movie = mpeg2source("D:\Estazione DVD\Alexander3\agk_tmp\Alexander3.d2v")
cropclip = autocrop(movie,mode=0,wmultof=4,hmultof=4,samples=10,aspect=0,threshold=34,samplestartframe=0,leftadd=0,rightadd=0,topadd=0,bottomadd=0)
fixed_aspect = 1.42222222222222
c_width = width(cropclip)
c_height = round(height(cropclip) / fixed_aspect)
input_par = float(c_width)/float(c_height)
input_par = input_par > 1.4 ? input_par : (4.0/3.0)
out_width = 704
out_height = round(float(out_width) / input_par)
hmod = out_height - (floor(out_height / 16 ) * 16)
out_height = (hmod > 4) ? (out_height + (16 - hmod)) : (out_height - hmod)
new_aspect = (float(out_width) / float(out_height)) / fixed_aspect
autocrop(movie,mode=0,wmultof=4,hmultof=4,samples=10,aspect=new_aspect,threshold=34,samplestartframe=0,leftadd=0,rightadd=0,topadd=0,bottomadd=0)
LanczosResize(out_width,out_height)
RemoveGrain(mode=2)
SelectRangeEvery(300,15)
===========================================================
[12/07/2005 0.35.15] Duration was: 6 minutes 18 seconds
[12/07/2005 0.35.15] Speed was: 17,85 fps.
[12/07/2005 0.35.15] Compressibility percentage is: 59,83
[12/07/2005 0.35.15] Using softer matrix.
[12/07/2005 0.35.15] Chosen resolution is: 672x272 ( AR: 2,47 )
[12/07/2005 0.35.15] Predicted comptest value is: 73,90
[12/07/2005 0.35.15] Running first pass.
[12/07/2005 0.35.15] Writing the following script to D:\Estazione DVD\Alexander3\agk_tmp\Alexander3_movie.avs
===========================================================
LoadPlugin("C:\PROGRA~1\AutoGK\DGMPGDec\DGDecode.dll")
LoadPlugin("C:\PROGRA~1\AutoGK\filters\autocrop.dll")
LoadPlugin("C:\PROGRA~1\AutoGK\filters\RemoveGrainSSE2.dll")
movie = mpeg2source("D:\Estazione DVD\Alexander3\agk_tmp\Alexander3.d2v")
cropclip = autocrop(movie,mode=0,wmultof=4,hmultof=4,samples=10,aspect=0,threshold=34,samplestartframe=0,leftadd=0,rightadd=0,topadd=0,bottomadd=0)
fixed_aspect = 1.42222222222222
c_width = width(cropclip)
c_height = round(height(cropclip) / fixed_aspect)
input_par = float(c_width)/float(c_height)
input_par = input_par > 1.4 ? input_par : (4.0/3.0)
out_width = 672
out_height = round(float(out_width) / input_par)
hmod = out_height - (floor(out_height / 16 ) * 16)
out_height = (hmod > 4) ? (out_height + (16 - hmod)) : (out_height - hmod)
new_aspect = (float(out_width) / float(out_height)) / fixed_aspect
autocrop(movie,mode=0,wmultof=4,hmultof=4,samples=10,aspect=new_aspect,threshold=34,samplestartframe=0,leftadd=0,rightadd=0,topadd=0,bottomadd=0)
LanczosResize(out_width,out_height)
RemoveGrain(mode=2)
===========================================================
[12/07/2005 1.46.41] Duration was: 1 hour, 11 minutes 25 seconds
[12/07/2005 1.46.41] Speed was: 31,50 fps.
[12/07/2005 1.46.41] Expected quality of first pass size: 57,63
[12/07/2005 1.46.41] Trying to adjust settings.
[12/07/2005 1.46.41] Chosen resolution is: 640x272 ( AR: 2,35 )
[12/07/2005 1.46.41] Predicted comptest value is: 60,51
[12/07/2005 1.46.41] Re-running first pass.
[12/07/2005 1.46.41] Writing the following script to D:\Estazione DVD\Alexander3\agk_tmp\Alexander3_movie.avs
===========================================================
LoadPlugin("C:\PROGRA~1\AutoGK\DGMPGDec\DGDecode.dll")
LoadPlugin("C:\PROGRA~1\AutoGK\filters\autocrop.dll")
LoadPlugin("C:\PROGRA~1\AutoGK\filters\RemoveGrainSSE2.dll")
movie = mpeg2source("D:\Estazione DVD\Alexander3\agk_tmp\Alexander3.d2v")
cropclip = autocrop(movie,mode=0,wmultof=4,hmultof=4,samples=10,aspect=0,threshold=34,samplestartframe=0,leftadd=0,rightadd=0,topadd=0,bottomadd=0)
fixed_aspect = 1.42222222222222
c_width = width(cropclip)
c_height = round(height(cropclip) / fixed_aspect)
input_par = float(c_width)/float(c_height)
input_par = input_par > 1.4 ? input_par : (4.0/3.0)
out_width = 640
out_height = round(float(out_width) / input_par)
hmod = out_height - (floor(out_height / 16 ) * 16)
out_height = (hmod > 4) ? (out_height + (16 - hmod)) : (out_height - hmod)
new_aspect = (float(out_width) / float(out_height)) / fixed_aspect
autocrop(movie,mode=0,wmultof=4,hmultof=4,samples=10,aspect=new_aspect,threshold=34,samplestartframe=0,leftadd=0,rightadd=0,topadd=0,bottomadd=0)
LanczosResize(out_width,out_height)
RemoveGrain(mode=2)
===========================================================
[12/07/2005 2.53.35] Duration was: 1 hour, 6 minutes 53 seconds
[12/07/2005 2.53.35] Speed was: 33,64 fps.
[12/07/2005 2.53.35] Expected quality of first pass size: 62,31
[12/07/2005 2.53.35] Running second pass.
[12/07/2005 4.27.58] Duration was: 1 hour, 34 minutes 22 seconds
[12/07/2005 4.27.58] Speed was: 23,84 fps.
[12/07/2005 4.27.59] Splitting video into: D:\Estazione DVD\Alexander3\Alexander3.cd1.avi
[12/07/2005 4.29.07] Splitting video into: D:\Estazione DVD\Alexander3\Alexander3.cd2.avi
[12/07/2005 4.30.10] Job finished. Total time: 4 hours 8 minutes 49 seconds
During this encoding i see that the first pass was runned twice, decreasing the resolutiom from 672 to 640. What is the reason?
The source in the same, but it is not an original dvd, it is a dvd processed with dvd-shrink; in the first encode (2.12)i used the folder made with it and saved on my HD, in the second encode i ripped the shrinked dvd on HD with dvd decrypter in ifo mode.
My question is: why using previous versions of AutoGK, 1.96 included, i never had B-Frames on my encodings, all in a very good quality (i usually encode movies in 1400mb size) and using latest betas with new xvid encoder those are present? Those also present much higher average drf and standard deviation, the difference looking at the movie is not big, but it is visible.
About the standalones i could have made a mistake, i have a kiss dp1000 and a amstrad 3016dx(chip mediatek i think) and i have no problems with b-frames, but a friend of mine did have with a Philips (don't know the model).
I will make more test next days with an original dvd and using only dvd decrypter
unskinnyboy
20th July 2005, 02:46
OK, Here is the deal - AutoGK will try to attain your desired target size at all costs. You chose the minimum width to be 640, so AutoGK started with 704 and ran a compressibility test. The value it got was pretty low (59.83%) for your encode to look decent. So it used a softer quantization matrix (H.263), reduced the h-res to 672 and ran the first pass. It expected 73.90%, but got only 57.63%. Again, not enough for a quality encode and more over you had allowed it to go down as much as 640. So it reduced the h-res to 640 and re-ran the first pass and got 62.31%. It would have reduced the resolution still further if you had chosen 'AutoWidth', but since you limited it to 640, it went onto the second pass and completed the encode. There is no way you can do without b-frames in this case.
Encoding from a DVD9 and a "shrunk" DVD5 makes a lot of difference - the first source is pristine, while the second is already compressed, which introduces minor artifacts and degrades the quality of the material. While it might be invisible to your eye, AutoGK will find it difficult to compress and that is why you are getting low comptest values.
You say the previous versions of AutoGK/XviD disabled b-frames for your encodes, but was it the *same* source? Go back to your previous versions of AutoGK/XviD and try this same "shrunk" Alexander in it. It will behave the same.
BigDid
20th July 2005, 03:28
Again late compared to unskinnyboy but I have details, stay tuned :)
...During this encoding i see that the first pass was runned twice, decreasing the resolutiom from 672 to 640. What is the reason?
Answer is in the tutorial or FAQ, related to difference between compressibility percentage and Predicted comptest value: if more than 10% (or 15% not sure) first pass is rerun.
...My question is: why using previous versions of AutoGK, 1.96 included, i never had B-Frames on my encodings, all in a very good quality (i usually encode movies in 1400mb size) and using latest betas with new xvid encoder those are present? Those also present much higher average drf and standard deviation, the difference looking at the movie is not big, but it is visible.
My previous tip is working in AGK up to 2.09 rev and has to be tested in 2.12 or up because Lenox allowed the undersize possibility in 2.12, it's in the changelog here: http://forum.doom9.org/showthread.php?t=64266&page=1
Also see this thread for better understanding about undersize/codec saturation here: http://forum.doom9.org/showthread.php?p=655374#post655374
In short AGK switched (up to 2.09) the B-frame off and if needed changed min quantizer to 1 instead of 2 to prevent undersize. Actually it will not change quantizer and I dunno for the B-frames; up to you to verify the tip.
...I will make more test next days with an original dvd and using only dvd decrypter
Thanks to report back
Did
unskinnyboy
20th July 2005, 04:18
Ahhh, didn't know about the min quant change len0x put in (and now took out). Good to know. :)
Robyfly
20th July 2005, 10:08
.... There is no way you can do without b-frames in this case.
Encoding from a DVD9 and a "shrunk" DVD5 makes a lot of difference - the first source is pristine, while the second is already compressed, which introduces minor artifacts and degrades the quality of the material. While it might be invisible to your eye, AutoGK will find it difficult to compress and that is why you are getting low comptest values.
You say the previous versions of AutoGK/XviD disabled b-frames for your encodes, but was it the *same* source? Go back to your previous versions of AutoGK/XviD and try this same "shrunk" Alexander in it. It will behave the same.
Sorry, probably i didn't explained well:
First encode with 2.12 --> B-Frames present
Second encode with 1.96 --> B-Frames NOT present (the one you are talking)
I am completely satisfied by the 1.96 encode (ok, i shoud have set on auto and get less resolution, but i prefere loose a little quality an have a little more resolution)
About previous encodes i was speaking in general: i encoded about 15 dvds with AutoGK, usually setting it wit 2cds size, min 640, and, depending of the lenght and the type of the movie, choosing mp3 at 128/160/192 or ac3 audio track.
Sometimes i had very good resulting avi with 688/704 or 720 resolutions, and drf values very close to 2.8
About shrunk dvds, this is the first case (and the last i think :D )
It seems to me that with old versions with xvid 1.03 in general i get less resolution but more quality, with new beta i get more resolution, less quality, higher drf values and b-frames... may be AutoGK choose more resolution because b-frames allow more compression?
I want to test version 2.12 or the last 2.13 on an original dvd and see what happens
A big :thanks:
unskinnyboy
20th July 2005, 14:38
Sorry, probably i didn't explained well:
First encode with 2.12 --> B-Frames present
Second encode with 1.96 --> B-Frames NOT present (the one you are talking)
Sorry, your issue is clear to me now. I was touching on just the second encode only.
First encode with 2.12 --> B-Frames present
True, and the log concurs.
Second encode with 1.96 --> B-Frames NOT present
Even though the DRF report says 0.00% B-frames, looking at the log there is no indication of AutoGK having turned off b-frames. And as I said before, b-frames will only be switched off when AutoGK sees that a high percentage of compression is possible and if it keeps b-frames, undersizing will occur. In your case, you have a difficult to compress source and so logically speaking, b-frames *should* be present. Now why the DRF says 0.00% I don't know, but can you make the XviD stats file available? I also recollect a similar situation (http://forum.doom9.org/showthread.php?t=92359) from before.
And yes, for the same filesize, b-frames will give more compression and will allow AutoGK to choose better resolutions.
Robyfly
20th July 2005, 21:35
Unfortunately i deleted the other files :devil:
I looked at the encoder configuration ant it was on Profile Simple@L0
Encoding type: Twopass- 2nd pass
Target file size 1178503
Clicking on more... the profile window is completely turned to grey (disabled) included B-Vops and packed bitstream
I remember in the past i looked at the codec and i found profile AS@L5
I saved some old log on my HD, but i remember only in one case i saw "B-frames switched off", but in all cases i never had B-Frames present in the resulting avis.
Of course that log is not present on my HD :devil: , here is another old log with high compressibility values and relative DRF report:
[18/05/2005 0.44.26] AutoGK 1.95
[18/05/2005 0.44.26] OS: WinXP (5.1.2600).2
[18/05/2005 0.44.26] Job started.
[18/05/2005 0.44.26] Input dir: D:\Estazione DVD\Qualcuno
[18/05/2005 0.44.26] Output file: D:\Estazione DVD\Qualcuno\Qualcuno.avi
[18/05/2005 0.44.26] Audio: Italiano
[18/05/2005 0.44.26] Subtitles: none
[18/05/2005 0.44.26] Codec: XviD
[18/05/2005 0.44.26] Target size: 1400Mb
[18/05/2005 0.44.26] Custom resolution settings: minimum width of 640 pixels
[18/05/2005 0.44.26] Custom audio settings: CBR MP3 with bitrate: 192Kbps
[18/05/2005 0.44.26] Started encoding.
[18/05/2005 0.44.26] Demuxing and indexing.
[18/05/2005 0.46.30] Processing file: D:\Estazione DVD\Qualcuno\VTS_02_1.VOB
[18/05/2005 0.46.30] Processing file: D:\Estazione DVD\Qualcuno\VTS_02_2.VOB
[18/05/2005 0.46.30] Processing file: D:\Estazione DVD\Qualcuno\VTS_02_3.VOB
[18/05/2005 0.46.30] Processing file: D:\Estazione DVD\Qualcuno\VTS_02_4.VOB
[18/05/2005 0.46.30] Processing file: D:\Estazione DVD\Qualcuno\VTS_02_5.VOB
[18/05/2005 0.46.30] Source aspect ratio: 16:9
[18/05/2005 0.46.30] Source resolution: 720x576
[18/05/2005 0.46.30] Found PAL source.
[18/05/2005 0.46.31] Analyzing source.
[18/05/2005 0.52.09] Source has percentage of interlacing in motion areas: 0,00
[18/05/2005 0.52.09] Source is considered to be progressive.
[18/05/2005 0.52.09] Output will contain 146778 frames
[18/05/2005 0.52.09] Decoding audio.
[18/05/2005 0.54.20] Normalizing audio.
[18/05/2005 0.56.34] Encoding audio.
[18/05/2005 1.04.40] Using bitrate spikes control.
[18/05/2005 1.04.41] Audio size: 140,909,184 bytes (134.38 Mb)
[18/05/2005 1.04.42] Overhead: 1,742,976 bytes (1.66 Mb)
[18/05/2005 1.04.42] Video size: 1,325,354,240 bytes (1263.96 Mb)
[18/05/2005 1.04.42] Running compressibility test.
[18/05/2005 1.04.42] Writing the following script to D:\Estazione DVD\Qualcuno\agk_tmp\Qualcuno_comptest.avs
===========================================================
LoadPlugin("C:\PROGRA~1\AutoGK\DGMPGDec\DGDecode.dll")
LoadPlugin("C:\PROGRA~1\AutoGK\filters\autocrop.dll")
LoadPlugin("C:\PROGRA~1\AutoGK\filters\RemoveGrainSSE2.dll")
movie = mpeg2source("D:\Estazione DVD\Qualcuno\agk_tmp\Qualcuno.d2v")
cropclip = autocrop(movie,mode=0,wmultof=4,hmultof=4,samples=10,aspect=0,threshold=34,samplestartframe=0,leftadd=0,rightadd=0,topadd=0,bottomadd=0)
fixed_aspect = 1.42222222222222
c_width = width(cropclip)
c_height = round(height(cropclip) / fixed_aspect)
input_par = float(c_width)/float(c_height)
input_par = input_par > 1.4 ? input_par : (4.0/3.0)
out_width = 704
out_height = round(float(out_width) / input_par)
hmod = out_height - (floor(out_height / 16 ) * 16)
out_height = (hmod > 4) ? (out_height + (16 - hmod)) : (out_height - hmod)
new_aspect = (float(out_width) / float(out_height)) / fixed_aspect
autocrop(movie,mode=0,wmultof=4,hmultof=4,samples=10,aspect=new_aspect,threshold=34,samplestartframe=0,leftadd=0,rightadd=0,topadd=0,bottomadd=0)
LanczosResize(out_width,out_height)
RemoveGrain(mode=2)
SelectRangeEvery(300,15)
===========================================================
[18/05/2005 1.12.03] Duration was: 7 minutes 19 seconds
[18/05/2005 1.12.03] Speed was: 16,70 fps.
[18/05/2005 1.12.03] Compressibility percentage is: 93,20
[18/05/2005 1.12.03] Chosen resolution is: 720x400 ( AR: 1,80 )
[18/05/2005 1.12.03] Predicted comptest value is: 87,48
[18/05/2005 1.12.03] Running first pass.
[18/05/2005 1.12.03] Writing the following script to D:\Estazione DVD\Qualcuno\agk_tmp\Qualcuno_movie.avs
===========================================================
LoadPlugin("C:\PROGRA~1\AutoGK\DGMPGDec\DGDecode.dll")
LoadPlugin("C:\PROGRA~1\AutoGK\filters\autocrop.dll")
LoadPlugin("C:\PROGRA~1\AutoGK\filters\RemoveGrainSSE2.dll")
movie = mpeg2source("D:\Estazione DVD\Qualcuno\agk_tmp\Qualcuno.d2v")
cropclip = autocrop(movie,mode=0,wmultof=4,hmultof=4,samples=10,aspect=0,threshold=34,samplestartframe=0,leftadd=0,rightadd=0,topadd=0,bottomadd=0)
fixed_aspect = 1.42222222222222
c_width = width(cropclip)
c_height = round(height(cropclip) / fixed_aspect)
input_par = float(c_width)/float(c_height)
input_par = input_par > 1.4 ? input_par : (4.0/3.0)
out_width = 720
out_height = round(float(out_width) / input_par)
hmod = out_height - (floor(out_height / 16 ) * 16)
out_height = (hmod > 4) ? (out_height + (16 - hmod)) : (out_height - hmod)
new_aspect = (float(out_width) / float(out_height)) / fixed_aspect
autocrop(movie,mode=0,wmultof=4,hmultof=4,samples=10,aspect=new_aspect,threshold=34,samplestartframe=0,leftadd=0,rightadd=0,topadd=0,bottomadd=0)
LanczosResize(out_width,out_height)
RemoveGrain(mode=2)
===========================================================
[18/05/2005 2.49.55] Duration was: 1 hour, 37 minutes 51 seconds
[18/05/2005 2.49.55] Speed was: 25,00 fps.
[18/05/2005 2.49.56] Expected quality of first pass size: 89,98
[18/05/2005 2.49.56] Running second pass.
[18/05/2005 5.09.43] Duration was: 2 hours 19 minutes 47 seconds
[18/05/2005 5.09.43] Speed was: 17,50 fps.
[18/05/2005 5.09.43] Splitting video into: D:\Estazione DVD\Qualcuno\Qualcuno.cd1.avi
[18/05/2005 5.10.48] Splitting video into: D:\Estazione DVD\Qualcuno\Qualcuno.cd2.avi
[18/05/2005 5.11.46] Job finished. Total time: 4 hours 27 minutes 20 seconds
DRF:
DivX DRF Analyzer v0.9.5 Report!
File Name: D:\Bittorrent\Uploads\Qualcunodaamare\Qualcuno720x384.avi
FourCC: XVID
Codec: XviD0037
Resolution: [ Width: 720 Height: 384 ]
Frame Rate: 25.000 frames per second
The Video has 146778 frames [ 01:37:51 ]
Average Frame quality is HIGH [Average DRF/quantizer is 2.19]
Standard Deviation: Quality is HIGH [Std. Deviation is 0.39]
Image Resolution is HIGH
There are NO frame drops ( NO drops is better )
Recomended Resolution: [800x432] (Target DRF/quantizer=2.8)
The filesize can be smaller!
Performance Caracteristics:
This video seems to have too FEW Keyframes. May degrade image quality and seeking.
Macroblocks per frame: 1080 ( Poor Playback in Slow Computers, PIII450 or better required )
The Width is not multiple of 32. May degrade performance in some systems.
Kilobits per Second: 1758.84
Kilobits per Frame: 70.35
Kilobits per Macroblock: 0.065
Bits per Pixel: 0.26
Frame Type Statistics :
I Frames: 0.65%
P Frames: 99.35%
B Frames: 0.00%
S Frames: 0.00%
N Frames: 0.00%
(More Advanced Codecs use B and S frames)
Frame Quality Statistics :
DRF=1&2: 117094 80.3%
DRF=3: 28737 19.7%
DRF=4: 0 0.0%
DRF=5: 0 0.0%
DRF=6: 0 0.0%
DRF=7: 0 0.0%
DRF=8: 0 0.0%
DRF=9: 0 0.0%
DRF>9: 0 0.0%
KeyF/DeltaF: 0.65%
KeyDRF<4: 947
KeyDRF=4: 0
KeyDRF>4: 0
AverageKeyDRF: 2.20
MAXDRF: 3
AverageDRF: 2.19
Deviation: 0.39
BigDid
20th July 2005, 21:58
...I saved some old log on my HD, but i remember only in one case i saw "B-frames switched off",...
I have searched the AGK changelog with no success, the "B-frames switched off" in the log (after the comptest not after first pass, my mistake) appeared between 1.96 and 2.08.
If you want a rev which still has the no-undersize option, I would not recommend 2.08, rather use 2.09.
Did
unskinnyboy
20th July 2005, 22:08
Simple@L0? No wonder you didn't get any b-frames. It is a restrictive profile and disables all advanced MPEG-4 features. Maybe 'bitrate spikes control' forces Simple@L0. Test with and without forcing 'bitrate spikes control' and then check the DRF. AS@L5 will allow b-frames.
BigDid
20th July 2005, 22:34
Simple@L0? No wonder you didn't get any b-frames. It is a restrictive profile and disables all advanced MPEG-4 features. Maybe 'bitrate spikes control' forces Simple@L0. Test with and without forcing 'bitrate spikes control' and then check the DRF. AS@L5 will allow b-frames.
Hi USB,
I am afraid the AS@L5 default profile disappeared in the 2.08 or 2.09 rev replaced with the unrestricted profile (Edited:which also allows B-frames)
:0_0: (2 big eyes smiley, should be :s )
Did
unskinnyboy
20th July 2005, 23:27
Hi USB,
I am afraid the AS@L5 default profile disappeared in the 2.08 or 2.09 rev replaced with the unrestricted profile (Edited:which also allows B-frames)
:0_0: (2 big eyes smiley, should be :s )
Did
I know that..did notice that in some earlier tests. But is that the case even with 'bitrate spikes control'? I doubt it (not in a position to test atm). I *think* previously AS@L5 and now 'unrestricted' is chosen only without it. OK - I better shut up now till I go home and get to the bottom of this.
unskinnyboy
21st July 2005, 03:07
Bleh! Bitrate spikes control just enables 'DXN HT NTSC' profile (should have guessed). MTK support enables 'MTK NTSC' profile. No idea what forces Simple@L0, because that seems to be the only reason why b-frames are not used. Will wait for len0x to answer this. :(
Robyfly
21st July 2005, 14:49
ok, for next tests i have to wait few days, what version you suggest to use?
I have 1.96, 2.09, 2.12 and 2,13
BigDid
21st July 2005, 16:53
ok, for next tests i have to wait few days, what version you suggest to use?
I have 1.96, 2.09, 2.12 and 2,13
If you want AGK to be able to switch B-frames off stay with 1.96 or 2.09
Did
unskinnyboy
21st July 2005, 17:30
If you want AGK to be able to switch B-frames off stay with 1.96 or 2.09
Did
And why would it switch off b-frames with v1.96 or v2.09 with his current settings?
BigDid
21st July 2005, 17:45
@ Unskinnyboy
...Also see this thread for better understanding about undersize/codec saturation here: http://forum.doom9.org/showthread.php?p=655374#post655374
In short AGK switched (up to 2.09) the B-frame off and if needed changed min quantizer to 1 instead of 2 to prevent undersize. ...
In the log of the quoted thread it is written:
...[5/7/2005 11:15:20 AM] Compressibility percentage is: 134.27
[5/7/2005 11:15:20 AM] Switching b-frames off
[5/7/2005 11:15:20 AM] Chosen resolution is: 720x304 ( AR: 2.37 )
[5/7/2005 11:15:20 AM] Predicted comptest value is: 100.43
[5/7/2005 11:15:20 AM] Running first pass. ...
That was in 2.08 related to undersize and should still be valid in 2.09. In 2.12 or after it has to be tested, the only thing sure is no undersizing avoiding (by switching to quant1) anymore.
Hope it is clear, not enough coffe this morning :D
Did
unskinnyboy
21st July 2005, 19:15
Doh..OK, I am so confused I don't even understand the purpose of this thread anymore :D. So many logs in here, we are all refering to something else everytime. So OK, *that* log, yes, got it. I was refering to the other log where the first pass was re-run.
BigDid
21st July 2005, 19:49
Doh.. I don't even understand the purpose of this thread anymore :D. ...
@ USB
I kind of agree, I will wait for Robyfly to make the tests and come back with results or more questions :)
Did
Robyfly
22nd July 2005, 09:28
Just trying to explain the purpose.. :D
In my little experience i found that usually my rips with 640 min res, and 1400mb size using stable releases are very good and also good drf values (to confirm).
All rips are without b-frames.
With same settings with new betas b-frames are present, i found more resolution but, imho, a worse video quality and worse drf values (to confirm).
Now i know that it is not possible a manual switch-off of the b-frames, could be reasonable to ask if it would be possible to have this option in the hidden options? Or it is one illogical question or a not possible option?
In yours encodings, using betas and stable releases, did you find always (or often) b-frames? How is the quality?
Speaking in general and about movies with 90-120 minutes lenght my settings are wrong, or there is something strange in the program?
If i make more test, have i to re-install the program without the bitrate spikes control?
But first of all, :thanks: for explaining a lot of things.
unskinnyboy
22nd July 2005, 14:51
In my little experience i found that usually my rips with 640 min res, and 1400mb size using stable releases are very good and also good drf values (to confirm).
All rips are without b-frames.
This is the part which I haven't understood till now (can someone please one explain this magic?). How is that all the encodes didn't get b-frames? 640 min-res and 1400 MB target size is not the kind of settings where you will get undersize (ok, depending on the material you might get for a few encodes, but surely not all?), so I am assuming the lack of b-frames might be due to a profile like Simple@L0 which was forced, but I don't know which option forces that profile. I am also skeptical about your claim about the quality, but that's not important atm.
Now i know that it is not possible a manual switch-off of the b-frames, could be reasonable to ask if it would be possible to have this option in the hidden options? Or it is one illogical question or a not possible option?
In yours encodings, using betas and stable releases, did you find always (or often) b-frames? How is the quality?
Eventhough b-frames themselves are encoded at a higher quantizer (lower quality), for a given target size if there is an opportunity for using b-frames, then it only will improve the overall quality of the encode. If you are using AutoGK, it is prudent to leave that decision to it.
I have a feeling that there is something else which is wrong with your encodes. b-frames can't be the cause for bad quality (there is some good info on b-frames in this forum. Search and read about it a bit).
Robyfly
27th July 2005, 14:48
Ok, i'm curious and i made 4 test using version 2.13 and 1.96
I choosed the italian version of Untamed Heart, which seems to be very well comprimible.
I have all logs, drf. gspot analisys and take note of the encoder profile and b-vop option after each encode, so if necessary tell me and i will put them on.
For each rip i set min res at 640 1400mb size (apart 1 case at 700mb) and installed the program without bitrate spikes control.
----------------------------------------------------
Test1: 2.13 1400mb,640 min res, mp3 128 MTK profile
tried this to have a final avi without b-frames
[24/07/2005 1.32.56] Compressibility percentage is: 101,29
[24/07/2005 1.32.56] Chosen resolution is: 720x400 ( AR: 1,80 )
[24/07/2005 1.32.56] Predicted comptest value is: 95,08
and in fact after first pass i had
[24/07/2005 3.18.18] Switching b-frames off
[24/07/2005 3.18.18] Re-running first pass.
Average Frame quality is HIGH [Average DRF/quantizer is 2.50]
Standard Deviation: Quality is HIGH [Std. Deviation is 0.50]
Bits per Pixel: 0.27
Frame Type Statistics :
I Frames: 0.64%
P Frames: 99.36%
B Frames: 0.00%
Codec XviD: Profile MTK PAL B-VOPs not active
------------------------------------------------------
Test 2: 2.13 1400mb,640 min res, AC3 (only 2 channels) MTK profile
tried this to have B-Frames
[25/07/2005 0.27.33] Compressibility percentage is: 97,88
[25/07/2005 0.27.33] Chosen resolution is: 720x400 ( AR: 1,80 )
[25/07/2005 0.27.33] Predicted comptest value is: 91,88
and i had b-frames:
Average Frame quality is HIGH [Average DRF/quantizer is 2.98]
Standard Deviation: Quality is MEDIUM [Std. Deviation is 1.00]
Bits per Pixel: 0.26
Frame Type Statistics :
I Frames: 0.64%
P Frames: 50.14%
B Frames: 49.22%
Codec XviD: Profile MTK PAL B-VOPs active
------------------------------------------------------
Test 3: 1.96 700mb, 640 min res, mp3 160
tried this to see what AutoGK do, i had b-frames
[26/07/2005 0.54.48] Compressibility percentage is: 68,69
[26/07/2005 0.54.48] Chosen resolution is: 640x352 ( AR: 1,82 )
[26/07/2005 0.54.48] Predicted comptest value is: 68,69
Average Frame quality is MEDIUM [Average DRF/quantizer is 3.64]
Standard Deviation: Quality is MEDIUM [Std. Deviation is 1.03]
Bits per Pixel: 0.15
Frame Type Statistics :
I Frames: 0.64%
P Frames: 50.09%
B Frames: 49.26%
Codec XviD: Profile AS @L5 B-VOPs active
------------------------------------------------------
Test 4: 1.96 1400mb, 640 min res, AC3 (only 2 channels)
In this case i had B-Frames switched-off again!
[27/07/2005 0.28.29] Compressibility percentage is: 107,73
[27/07/2005 0.28.29] Switching b-frames off
[27/07/2005 0.28.29] Chosen resolution is: 720x400 ( AR: 1,80 )
[27/07/2005 0.28.29] Predicted comptest value is: 75,33
Average Frame quality is HIGH [Average DRF/quantizer is 2.43]
Standard Deviation: Quality is HIGH [Std. Deviation is 0.49]
Bits per Pixel: 0.26
Frame Type Statistics :
I Frames: 0.65%
P Frames: 99.35%
B Frames: 0.00%
encoder: Codec XviD:Profile AS @L5 B-VOPs not active
-------------------------------------------------------
This is not finished... :p
i conserved on mi HD an old log made with version 1.95 of the same movie with similar settings: min res 640, mp3 192 (equivalent size of Ac3) 1400mb
and after compressibility test i found this:
[18/05/2005 1.12.03] Compressibility percentage is: 93,20
[18/05/2005 1.12.03] Chosen resolution is: 720x400 ( AR: 1,80 )
[18/05/2005 1.12.03] Predicted comptest value is: 87,48
In this case there is not the message "Switching b-frames off"
but here is the drf:
Average Frame quality is HIGH [Average DRF/quantizer is 2.19]
Standard Deviation: Quality is HIGH [Std. Deviation is 0.39]
Bits per Pixel: 0.26
Frame Type Statistics :
I Frames: 0.65%
P Frames: 99.35%
B Frames: 0.00%
I also found that in every encode (apart 700mb) the real resolution was 720x384, but this probably is due to the fact that maximun horiz. resolution is reached, is it true?
Using b-frames makes DRF Analyzer useless?
:thanks:
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.