VP7 Free For Personal Use : 7.0.10.0 - Page 2

Peter1234 · 29th July 2005, 02:09

I tested VP7, fmeg (xvid), and X264 at 720x480 30fps using 900 kbps video encoding rate with the same test clip. VP7 was a clear winner. It had the fewest macro blocks and the sharpest image. The X264 encode was the worst since it would skip frames on fast action scenes and was jerky when played back (on 3GHz P4).

EDIT:

ONE PASS ENCODE TIMES on 3 GHz P4 (30 second video + mp3
audio)

Xvid = 30.3 sec
X264 = 1 min 32 sec
VP7 = 3 min 30 sec

B.F. · 29th July 2005, 04:38

Quote:

Originally Posted by On2Tech

Can you post the settings you used for both codecs? We'd like to duplicate your results.

Van Helsing traler
1:26 long, 720:400, only fast motion action scenes.
Video bitrate 400 Kbit.
Two pass max quality encoding.
Vp7 7.0.8
x264vfw.revision280

Encoding time almost the same (Vp7 faster in 1st pass but slower in 2nd pass).
Vp7 decoder eat more cpu power to decode video.
Vp7 too slowly react on the scene change and picture look less detaled than the x264.

29th July 2005, 04:40

pls, can anyone tell me what should be the estimated bitrates, sharpness settings, etc. for a 2-hr. dvd movie to vp7 encode, please on2 dude, pls give some help!

On2Tech · 29th July 2005, 13:18

Regarding you first question, I am sorry if this sounds obvious but the rate you choose depends on how big you want the final file to be. If you want a 2 hour DVD to fit on a single CD then about 750Kbits/sec is a good starting point and should give you something eminently watch-able, as most real movies are overall easier to encode than movie trailers, which tend to have 50% of the action crammed into 2-3 minutes plus tons of scene cuts, fades etc.

It is always possible to get some improvement in metrics or psycho visual characteristics by fiddling with the various VBR settings, sharpness settings and preprocessing level, especially on short clips, but we have set what we believe to be reasonable defaults that should give you reliably good results.

Things like sharpness are a bit of a matter of personal taste. The default value of 0 will usually give the best metrics response but some people are willing to accept some increase in the level of some artifacts in return for more sharpness. I would not recommend that you try and tune this on a clip by clip basis, particularly if you are going to encode at best quality (note that good quality speed 0 is almost as good on most clips and about twice as fast), because you will waste a lot of time. Find a value that suits your preferences and stick with it.

I would keep the pre-processing set to 0 unless you have a very noisy source. Pre-processing / noise reduction techniques can help improve the visual quality if the source is noisy but will always tend to hurt metrics numbers.

For coding a movie at around 750K I would avoid temporal or spatial re-sampling. These really only come into their own at much lower rates.

CBR (or “Stream from Server”) is designed for case where you want to stream your video over a fixed bandwidth link and it enforces strict buffering constraints. If you are going to play back off something like a CD or hard drive then VBR or “Local File Playback” is your best option.

In general I would suggest that you de-interlace you video before you encode it rather than encoding the raw interlaced video using “interlaced mode”. If you have to encode interlaced video then the interlaced mode optimization switch will help, but interlaced video is always harder to encode than progressive scan material, so avoid it if at all possible.

There are others on this board much better qualified than I to advise on the process of getting the video off the DVD in the first place so I will not go into that here. However, the next version of the On2Compi encoder (available as part of the paid for “On2 Personal Video Solution”) will include direct support for VOBs and AVS scripts.

Note that the personal use video solution also includes the On2 Java video decoder that allows you to publish VP6 clips on a personal web site.

I hope this is helpful.

Regards On2Tech.

Sharktooth · 29th July 2005, 16:35

On2Tech what is VP7VFW_beta7-0-9.exe in your public ftp?

On2Tech · 29th July 2005, 17:09

Quote:

Originally Posted by Sharktooth

On2Tech what is VP7VFW_beta7-0-9.exe in your public ftp?

We gave a pre version of the codec out prior to this release to a customer who had trouble getting to our ftp site. Its been removed.

Thanks for pointing it out to us.

Sharktooth · 29th July 2005, 17:26

np. just another question about it. it was 7.0.9 and dated 07/07/2005 (IIRC). is it different from the 7.0.8 personal version?and since i downloaded it i wish to know if it is a commercial version. In that case i'll delete it from my HD.

On2Tech · 29th July 2005, 18:51

Quote:

Originally Posted by Sharktooth

np. just another question about it. it was 7.0.9 and dated 07/07/2005 (IIRC). is it different from the 7.0.8 personal version?and since i downloaded it i wish to know if it is a commercial version. In that case i'll delete it from my HD.

Its a version of a codec that was sitting on my hard drive that had some of the improvements that got into the released version of 7.0.8 but maybe not all. There is nothing more commercial about it than 7.0.8 and its probably worse.

It was named 7.0.9 because at the time I couldn't remember what version number we had released, not because its better.

On2Tech

Sharktooth · 30th July 2005, 10:37

so i can delete it.
thanks for the reply

Manao · 30th July 2005, 10:40

Peter1234 :

> The X264 encode was the worst since it would skip frames on fast action scenes and was jerky when played back (on 3GHz P4).

With what did you played it back ? Because 3 GHz is far more than enough to play back FULL@900kbps in h264.

Sharktooth · 30th July 2005, 11:41

however vp7 decoding without PP and film grain approximation is lighter than h.264...

Josef.Bilek · 30th July 2005, 22:53

VP7 seems really nice. I did couple of tests and VP7 seems really fully comparable with H.264. Sometimes better, sometimes worse.
I am curious what is the intended usage of this codec. I am using MDA device with Mobile Windows 2003 and TCMP player (by my opinion the best player in PDA world). Sometimes I tried convert some trailers to this device and the most difficult was select the proper audio and video codec due to the fact, that computing power of these devices is simply low. I am using standard H.263 Xvid with Ogg Vorbis in Matroska containter. Typical bitrate 150 kbps and resolution 320 x 240. I am looking for any option to this codec, but H.264 (X264) has the problem with computing power. Is there any plan, for example, to develop a VP7 plugin for TCMP optimized for StrongARM processors to use this codec on these devices?
It would be fine, because it seems that with bitrate 100 kbps is there the same level of details as in case of Xvid with 150 kbps. And SD cards have limited size.

Peter1234 · 31st July 2005, 00:10

Mano,
I encoded with X264 281A VFW codec using default Advanced settings except Partition Decision Quality was 5(High Quality) which may not have been the default. All tests played back with Windows Media Player 8 and WinDVD 4 (same results). Gspot shows that the Nero DVD filter was used for decoding X264 whereas Xvid was decoded with Xvid mpeg4 video decoder. Gspot shows that VP7 plays back using the VP7 video decoder filter. Source was 4:3 720x480 at 29.97fps mpeg2 9000 kbps originally from DV camera and encoded for DVD. Source was not de-intrlaced and I think I read somewhere that X264 can not use interlaced video, so that may have been the problem. This is the first time I have used X264, it was used to provide a comparison for VP7. I plan to do some more testing with it when I get some more time. Is there an X264 decoder for Windows that I should have installed? Is interlaced video a problem for X264?

EDIT:
I de-interlaced the source with AviSynth before feeding it to the X264 encoder and the results were still not good.

EDIT 2:
X264 playback problems were due to players. I installed VLC 0.8.2 player and the X264 video plays OK in it.

Irwin · 31st July 2005, 11:56

Strange. On my cpu (pIII 500mhz) vp7 required about twice more power then h264 (with cabac). 640x272 700kbps h264 plays smooth, this same file in vp7 -> jerky like hell (even 480x192).

Peter1234 · 31st July 2005, 14:04

TEST UPDATE (720x480 900 kbps):
X264 playback problems were due to lack of H264 decoder. When H264 decoder was installed VP7 and X264 quality was about the same (X264 seemed slightly better). X264 did have some minor problems if source video was interlaced and thus needs an external deinterlacer whereas VP7 has built-in deinterlacer.

Sharktooth · 31st July 2005, 15:02

Quote:

Originally Posted by Irwin

Strange. On my cpu (pIII 500mhz) vp7 required about twice more power then h264 (with cabac). 640x272 700kbps h264 plays smooth, this same file in vp7 -> jerky like hell (even 480x192).

Disable post processing and noise grain reproduction options in the decoder.
however a p3 500 is not enaugh for those new generation codecs.
just "forget about it".

jrb · 2nd August 2005, 11:14

i've followed the vp* codecs for a while, usually they're pretty interesting, and i've had some good results in the past - i look at codecs from encoding gaming movies, which pushes codecs in a totally different way to the usual film or cartoon type material. my last movie had to have a vbr rate 3400 in xvid to attain the quality i wanted - 21 minutes, plus 5 audio streams meant the resulting file was getting towards cd sized.

so, i was quite looking forward to trying out vp7, especially after reading the comments by other posters here. The issues about slow encode speeds, especially on the second pass of a 2 pass encode didn't really bother me, i'm personally in favour of quality for the viewer over encoding time for me.

i got bored of waiting for the movie to encode after 3-4 hours, and it had only completed around 50% of the job. I was also upset to find that i had a couple of unexpected results
- parts of the far left of the image were appearing on the far right
- the chroma layer appeared to be mal-aligned
- quality, in parts, not as good as my zoned xvid encode. (although i was using 3000, not 3400 for vp7).

I'll post some screenies when i get home, for comparison - along with the actual settings i used.

I've not tested encoding the same source with x264 yet, so i can't compare with that codec.

On2Tech · 2nd August 2005, 14:22

Quote:

Originally Posted by jrb

i've followed the vp* codecs for a while, usually they're pretty interesting, and i've had some good results in the past - i look at codecs from encoding gaming movies, which pushes codecs in a totally different way to the usual film or cartoon type material. my last movie had to have a vbr rate 3400 in xvid to attain the quality i wanted - 21 minutes, plus 5 audio streams meant the resulting file was getting towards cd sized.

so, i was quite looking forward to trying out vp7, especially after reading the comments by other posters here. The issues about slow encode speeds, especially on the second pass of a 2 pass encode didn't really bother me, i'm personally in favour of quality for the viewer over encoding time for me.

i got bored of waiting for the movie to encode after 3-4 hours, and it had only completed around 50% of the job. I was also upset to find that i had a couple of unexpected results
- parts of the far left of the image were appearing on the far right
- the chroma layer appeared to be mal-aligned
- quality, in parts, not as good as my zoned xvid encode. (although i was using 3000, not 3400 for vp7).

I'll post some screenies when i get home, for comparison - along with the actual settings i used.

I've not tested encoding the same source with x264 yet, so i can't compare with that codec.

This seems very odd. Can you give us details of exactly what you did plus if possible a short section of source material that exhibits this problem.

Thanks

On2Tech

On2Tech · 2nd August 2005, 14:33

Quote:

Originally Posted by Peter1234

TEST UPDATE (720x480 900 kbps):
X264 playback problems were due to lack of H264 decoder. When H264 decoder was installed VP7 and X264 quality was about the same (X264 seemed slightly better). X264 did have some minor problems if source video was interlaced and thus needs an external deinterlacer whereas VP7 has built-in deinterlacer.

VP7 has a mode switch that is designed to improve the quality when coding interlaced material but unless you want your output to be interlaced I would still reccommend that you de-interlace the source first.

On2Tech

jrb · 2nd August 2005, 22:10

okay, as promised, here's some screenshots. Just a couple of things to note
- the frames were chosen based on keyframes in the vp7 encode (to save scanning time in vdub), and therefore should paint vp7 in a better light than xvid
- the xvid shots are taken from a version encoded over a year ago, with, as far as my memory serves me right, version 1.0
- the source material really pushes most codecs, and as yet have been running my own tests with every stable codec worth investing time in (i need to add x264 to this list) - as yet xvid comes out on tops for consistent quality
(if you're into this kind of thing and want to check out the xvid encode, it can be found here - (sorry for the self pimping))
- the input lossless file is a huffyuv v2.1.1 encoded file, in 24bit rgb format

http://shaolinproductions.org/images/13796-lossless.png
http://shaolinproductions.org/images/13796-xvid.png
http://shaolinproductions.org/images/13796-vp7.png

http://shaolinproductions.org/images/14296-lossless.png
http://shaolinproductions.org/images/14296-xvid.png
http://shaolinproductions.org/images/14296-vp7.png

http://shaolinproductions.org/images/5852-lossless.png
http://shaolinproductions.org/images/5852-xvid.png
http://shaolinproductions.org/images/5852-vp7.png

http://shaolinproductions.org/images...ettings-01.jpg
http://shaolinproductions.org/images...ettings-02.jpg

29th July 2005, 02:09	#21 \| Link
Peter1234 Registered User Join Date: Aug 2004 Posts: 130	I tested VP7, fmeg (xvid), and X264 at 720x480 30fps using 900 kbps video encoding rate with the same test clip. VP7 was a clear winner. It had the fewest macro blocks and the sharpest image. The X264 encode was the worst since it would skip frames on fast action scenes and was jerky when played back (on 3GHz P4). EDIT: ONE PASS ENCODE TIMES on 3 GHz P4 (30 second video + mp3 audio) Xvid = 30.3 sec X264 = 1 min 32 sec VP7 = 3 min 30 sec Last edited by Peter1234; 29th July 2005 at 13:11.

29th July 2005, 16:35	#25 \| Link
Sharktooth Mr. Sandman Join Date: Sep 2003 Location: Haddonfield, IL Posts: 11,768	On2Tech what is VP7VFW_beta7-0-9.exe in your public ftp? __________________ MPEG-4 ASP Custom Matrices: EQM V1(old), EQM AutoGK Sharpmatrix (aka EQM V2), EQM V3HR (updated 01/10/2004), EQM V3LR, EQM V3ULR (updated 04/02/2005), EQM V3UHR (updated 17/12/2004) and EQM V3EHR (updated 05/10/2004) Info about my ASP matrices. MPEG-4 AVC Custom Matrices: EQM AVC-HR Info about my AVC matrices My x264 builds. Mooo!!!

29th July 2005, 17:26	#27 \| Link
Sharktooth Mr. Sandman Join Date: Sep 2003 Location: Haddonfield, IL Posts: 11,768	np. just another question about it. it was 7.0.9 and dated 07/07/2005 (IIRC). is it different from the 7.0.8 personal version?and since i downloaded it i wish to know if it is a commercial version. In that case i'll delete it from my HD. __________________ MPEG-4 ASP Custom Matrices: EQM V1(old), EQM AutoGK Sharpmatrix (aka EQM V2), EQM V3HR (updated 01/10/2004), EQM V3LR, EQM V3ULR (updated 04/02/2005), EQM V3UHR (updated 17/12/2004) and EQM V3EHR (updated 05/10/2004) Info about my ASP matrices. MPEG-4 AVC Custom Matrices: EQM AVC-HR Info about my AVC matrices My x264 builds. Mooo!!!

30th July 2005, 10:37	#29 \| Link
Sharktooth Mr. Sandman Join Date: Sep 2003 Location: Haddonfield, IL Posts: 11,768	so i can delete it. thanks for the reply __________________ MPEG-4 ASP Custom Matrices: EQM V1(old), EQM AutoGK Sharpmatrix (aka EQM V2), EQM V3HR (updated 01/10/2004), EQM V3LR, EQM V3ULR (updated 04/02/2005), EQM V3UHR (updated 17/12/2004) and EQM V3EHR (updated 05/10/2004) Info about my ASP matrices. MPEG-4 AVC Custom Matrices: EQM AVC-HR Info about my AVC matrices My x264 builds. Mooo!!!

30th July 2005, 11:41	#31 \| Link
Sharktooth Mr. Sandman Join Date: Sep 2003 Location: Haddonfield, IL Posts: 11,768	however vp7 decoding without PP and film grain approximation is lighter than h.264... __________________ MPEG-4 ASP Custom Matrices: EQM V1(old), EQM AutoGK Sharpmatrix (aka EQM V2), EQM V3HR (updated 01/10/2004), EQM V3LR, EQM V3ULR (updated 04/02/2005), EQM V3UHR (updated 17/12/2004) and EQM V3EHR (updated 05/10/2004) Info about my ASP matrices. MPEG-4 AVC Custom Matrices: EQM AVC-HR Info about my AVC matrices My x264 builds. Mooo!!!

29th July 2005, 04:40	#23 \| Link
neo_anderson Guest Posts: n/a	pls, can anyone tell me what should be the estimated bitrates, sharpness settings, etc. for a 2-hr. dvd movie to vp7 encode, please on2 dude, pls give some help!

29th July 2005, 13:18	#24 \| Link
On2Tech Registered User Join Date: Nov 2003 Posts: 163	Regarding you first question, I am sorry if this sounds obvious but the rate you choose depends on how big you want the final file to be. If you want a 2 hour DVD to fit on a single CD then about 750Kbits/sec is a good starting point and should give you something eminently watch-able, as most real movies are overall easier to encode than movie trailers, which tend to have 50% of the action crammed into 2-3 minutes plus tons of scene cuts, fades etc. It is always possible to get some improvement in metrics or psycho visual characteristics by fiddling with the various VBR settings, sharpness settings and preprocessing level, especially on short clips, but we have set what we believe to be reasonable defaults that should give you reliably good results. Things like sharpness are a bit of a matter of personal taste. The default value of 0 will usually give the best metrics response but some people are willing to accept some increase in the level of some artifacts in return for more sharpness. I would not recommend that you try and tune this on a clip by clip basis, particularly if you are going to encode at best quality (note that good quality speed 0 is almost as good on most clips and about twice as fast), because you will waste a lot of time. Find a value that suits your preferences and stick with it. I would keep the pre-processing set to 0 unless you have a very noisy source. Pre-processing / noise reduction techniques can help improve the visual quality if the source is noisy but will always tend to hurt metrics numbers. For coding a movie at around 750K I would avoid temporal or spatial re-sampling. These really only come into their own at much lower rates. CBR (or “Stream from Server”) is designed for case where you want to stream your video over a fixed bandwidth link and it enforces strict buffering constraints. If you are going to play back off something like a CD or hard drive then VBR or “Local File Playback” is your best option. In general I would suggest that you de-interlace you video before you encode it rather than encoding the raw interlaced video using “interlaced mode”. If you have to encode interlaced video then the interlaced mode optimization switch will help, but interlaced video is always harder to encode than progressive scan material, so avoid it if at all possible. There are others on this board much better qualified than I to advise on the process of getting the video off the DVD in the first place so I will not go into that here. However, the next version of the On2Compi encoder (available as part of the paid for “On2 Personal Video Solution”) will include direct support for VOBs and AVS scripts. Note that the personal use video solution also includes the On2 Java video decoder that allows you to publish VP6 clips on a personal web site. I hope this is helpful. Regards On2Tech.

30th July 2005, 10:40	#30 \| Link
Manao Registered User Join Date: Jan 2002 Location: France Posts: 2,856	Peter1234 : > The X264 encode was the worst since it would skip frames on fast action scenes and was jerky when played back (on 3GHz P4). With what did you played it back ? Because 3 GHz is far more than enough to play back FULL@900kbps in h264.

30th July 2005, 22:53	#32 \| Link
Josef.Bilek Registered User Join Date: May 2005 Posts: 6	VP7 seems really nice. I did couple of tests and VP7 seems really fully comparable with H.264. Sometimes better, sometimes worse. I am curious what is the intended usage of this codec. I am using MDA device with Mobile Windows 2003 and TCMP player (by my opinion the best player in PDA world). Sometimes I tried convert some trailers to this device and the most difficult was select the proper audio and video codec due to the fact, that computing power of these devices is simply low. I am using standard H.263 Xvid with Ogg Vorbis in Matroska containter. Typical bitrate 150 kbps and resolution 320 x 240. I am looking for any option to this codec, but H.264 (X264) has the problem with computing power. Is there any plan, for example, to develop a VP7 plugin for TCMP optimized for StrongARM processors to use this codec on these devices? It would be fine, because it seems that with bitrate 100 kbps is there the same level of details as in case of Xvid with 150 kbps. And SD cards have limited size.

31st July 2005, 00:10	#33 \| Link
Peter1234 Registered User Join Date: Aug 2004 Posts: 130	Mano, I encoded with X264 281A VFW codec using default Advanced settings except Partition Decision Quality was 5(High Quality) which may not have been the default. All tests played back with Windows Media Player 8 and WinDVD 4 (same results). Gspot shows that the Nero DVD filter was used for decoding X264 whereas Xvid was decoded with Xvid mpeg4 video decoder. Gspot shows that VP7 plays back using the VP7 video decoder filter. Source was 4:3 720x480 at 29.97fps mpeg2 9000 kbps originally from DV camera and encoded for DVD. Source was not de-intrlaced and I think I read somewhere that X264 can not use interlaced video, so that may have been the problem. This is the first time I have used X264, it was used to provide a comparison for VP7. I plan to do some more testing with it when I get some more time. Is there an X264 decoder for Windows that I should have installed? Is interlaced video a problem for X264? EDIT: I de-interlaced the source with AviSynth before feeding it to the X264 encoder and the results were still not good. EDIT 2: X264 playback problems were due to players. I installed VLC 0.8.2 player and the X264 video plays OK in it. Last edited by Peter1234; 31st July 2005 at 06:49.

31st July 2005, 11:56	#34 \| Link
Irwin Registered User Join Date: Nov 2004 Posts: 49	Strange. On my cpu (pIII 500mhz) vp7 required about twice more power then h264 (with cabac). 640x272 700kbps h264 plays smooth, this same file in vp7 -> jerky like hell (even 480x192).

31st July 2005, 14:04	#35 \| Link
Peter1234 Registered User Join Date: Aug 2004 Posts: 130	TEST UPDATE (720x480 900 kbps): X264 playback problems were due to lack of H264 decoder. When H264 decoder was installed VP7 and X264 quality was about the same (X264 seemed slightly better). X264 did have some minor problems if source video was interlaced and thus needs an external deinterlacer whereas VP7 has built-in deinterlacer.

2nd August 2005, 11:14	#37 \| Link
jrb Registered User Join Date: May 2004 Posts: 5	i've followed the vp* codecs for a while, usually they're pretty interesting, and i've had some good results in the past - i look at codecs from encoding gaming movies, which pushes codecs in a totally different way to the usual film or cartoon type material. my last movie had to have a vbr rate 3400 in xvid to attain the quality i wanted - 21 minutes, plus 5 audio streams meant the resulting file was getting towards cd sized. so, i was quite looking forward to trying out vp7, especially after reading the comments by other posters here. The issues about slow encode speeds, especially on the second pass of a 2 pass encode didn't really bother me, i'm personally in favour of quality for the viewer over encoding time for me. i got bored of waiting for the movie to encode after 3-4 hours, and it had only completed around 50% of the job. I was also upset to find that i had a couple of unexpected results - parts of the far left of the image were appearing on the far right - the chroma layer appeared to be mal-aligned - quality, in parts, not as good as my zoned xvid encode. (although i was using 3000, not 3400 for vp7). I'll post some screenies when i get home, for comparison - along with the actual settings i used. I've not tested encoding the same source with x264 yet, so i can't compare with that codec. __________________ http://shaolinproductions.org

2nd August 2005, 22:10	#40 \| Link
jrb Registered User Join Date: May 2004 Posts: 5	okay, as promised, here's some screenshots. Just a couple of things to note - the frames were chosen based on keyframes in the vp7 encode (to save scanning time in vdub), and therefore should paint vp7 in a better light than xvid - the xvid shots are taken from a version encoded over a year ago, with, as far as my memory serves me right, version 1.0 - the source material really pushes most codecs, and as yet have been running my own tests with every stable codec worth investing time in (i need to add x264 to this list) - as yet xvid comes out on tops for consistent quality (if you're into this kind of thing and want to check out the xvid encode, it can be found here - (sorry for the self pimping)) - the input lossless file is a huffyuv v2.1.1 encoded file, in 24bit rgb format http://shaolinproductions.org/images/13796-lossless.png http://shaolinproductions.org/images/13796-xvid.png http://shaolinproductions.org/images/13796-vp7.png http://shaolinproductions.org/images/14296-lossless.png http://shaolinproductions.org/images/14296-xvid.png http://shaolinproductions.org/images/14296-vp7.png http://shaolinproductions.org/images/5852-lossless.png http://shaolinproductions.org/images/5852-xvid.png http://shaolinproductions.org/images/5852-vp7.png http://shaolinproductions.org/images...ettings-01.jpg http://shaolinproductions.org/images...ettings-02.jpg __________________ http://shaolinproductions.org Last edited by jrb; 2nd August 2005 at 22:12.

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode