For what it's worth, I use a mixture of DV and analogue footage from DV and Hi8 camcorders. I capture all my footage as DVD NTSC MPEG2 into the ADS DX2 Cap Wiz and then import the files into VS10. I think that both John and Steve make valid points regarding MPEG2 here. In terms of my own use, I have only experienced audio sync problems when using the 'Smart Render' feature in Video Studio and I therefore stopped using the feature and now I have no sync problems at all, even using transitions etc. John's posting pointed out something I didn't realize; If the Mpeg2 video is converted to SQUARE pixels and the bitrate is 8Mbps, it can be recursively edited and re-rendered with no discernible degradation. Restoring proper 4:3 aspect ratio by converting the final video file back to NON-SQUARE pixels after all the re-editing preserves the original resolution.
Thanks to John and Steve for their continued input in this forum.
Getting the Best Picture Quality from Video Studio 10+
Moderator: Ken Berry
-
jchunter
Terry,
I leave Smart-Render enabled all the time because it preserves the original encoded picture quality, wherever is has not been edited. Only the cut junctions, or video that has had a global filter applied are altered. If you Smart-Render a short clip and watch the progress bar, you will see definite pauses wherever you cut/modified the video.
I don't know why you are having A/V synch problems but if I were you, I would try some experiments to isolate the cause. First, try separating the different types of video in different projects. Note that your analog video has is Upper Field First and your DV clips are Lower Field first. Don't mix them in the same project. I wrote another paper that showed the bad effects of cross-renderding video files - I'll see if I can find it.
See if the homogeneous projects are OOS. If you find an OOS project, try the binary search method of isolating the source clip that gives the problem. If the encoding is corrupt, it can sometimes be repaired by creating a video file of just that clip with smart render turned off. Then, the whole project can be repaired by replacing one clip. etc .etc.
Some users have had problems around transitions. Some may be buggy. Maddrummer was on the track of some tx problems last year. I stick to Fade to Black, CrossFade, and Flying Flip transitions and haven't had problems.
Methodical trouble shooting is, by far, the best technique to locate the source of these problems.
I leave Smart-Render enabled all the time because it preserves the original encoded picture quality, wherever is has not been edited. Only the cut junctions, or video that has had a global filter applied are altered. If you Smart-Render a short clip and watch the progress bar, you will see definite pauses wherever you cut/modified the video.
I don't know why you are having A/V synch problems but if I were you, I would try some experiments to isolate the cause. First, try separating the different types of video in different projects. Note that your analog video has is Upper Field First and your DV clips are Lower Field first. Don't mix them in the same project. I wrote another paper that showed the bad effects of cross-renderding video files - I'll see if I can find it.
See if the homogeneous projects are OOS. If you find an OOS project, try the binary search method of isolating the source clip that gives the problem. If the encoding is corrupt, it can sometimes be repaired by creating a video file of just that clip with smart render turned off. Then, the whole project can be repaired by replacing one clip. etc .etc.
Some users have had problems around transitions. Some may be buggy. Maddrummer was on the track of some tx problems last year. I stick to Fade to Black, CrossFade, and Flying Flip transitions and haven't had problems.
Methodical trouble shooting is, by far, the best technique to locate the source of these problems.
-
jchunter
In the process of rechecking my earlier measurements of re-rendered Mpeg2 video, I was unable to reproduce my earlier results that showed loss of resolution when a 4:3 aspect ratio image with Non-Square Pixel property was re-Smart-Rendered. In fact, my latest tests showed no visible loss of resolution when edits were followed by re-smart-rendering, five times.
After studying the test file that I used in the earlier test, I came to the conclusion that I must have accidentally cross rendered it, changing the field order property of another test file, rather then creating the test file from scratch. (See next post on cross rendering).
So I have to retract my statement, earlier in this thread, that Video Studio 10+ has a bug when rendering 4:3 mpeg2 video files with the Non-Square Pixel property.
I have to give Ulead Video Studio 10+ an A+ for its ability to recursively Edit then Render Mpeg2 up to five times with no measurable* loss of resolution.
* I used Virtual Dub Mpeg v. 1.6.15 at 3X magnification to measure horizontal and vertical resolution on an ISO 12233 resolution chart. These measured 6.0 in the horizontal axis and 5.25 in the vertical axis, before and after five sequences of edit followed by smart render.
After studying the test file that I used in the earlier test, I came to the conclusion that I must have accidentally cross rendered it, changing the field order property of another test file, rather then creating the test file from scratch. (See next post on cross rendering).
So I have to retract my statement, earlier in this thread, that Video Studio 10+ has a bug when rendering 4:3 mpeg2 video files with the Non-Square Pixel property.
I have to give Ulead Video Studio 10+ an A+ for its ability to recursively Edit then Render Mpeg2 up to five times with no measurable* loss of resolution.
* I used Virtual Dub Mpeg v. 1.6.15 at 3X magnification to measure horizontal and vertical resolution on an ISO 12233 resolution chart. These measured 6.0 in the horizontal axis and 5.25 in the vertical axis, before and after five sequences of edit followed by smart render.
Last edited by jchunter on Sat Aug 26, 2006 1:10 am, edited 1 time in total.
-
jchunter
Property Cross Rendering Distortions
As promised, here are the results of the tests that demonstrate some of the bad things that happen to video when a user accidentally or deliberately changes the Field Order property in Video Studio 10+.
The new tests showed that
1) The worst distortions occurred when UFF (Upper Field First) video was re-rendered to LFF (Lower Field First) and vice versa. Not only was there significant loss of resolution but also the image was “doubled” in the vertical axis. This has got to be a serious bug because the fields in the test file were identical (subject unmoving).
2) Converting from UFF or LFF to FB (Frame-Based) reduced the vertical resolution by one step from 5.25 to 4.25. Horizontal resolution was unchanged at 6.0, although vertical edges were softer.
3) Converting from Frame Based to UFF or LFF showed no loss of resolution.
4) High definition video (1440x1080) that is cross rendered from Upper Field First to Frame-Based, as when deinterlacing, showed a loss of vertical resolution from 9.0 to 7.0.
Comment
Result 4) is problematic for me because I see smother motion on moving subjects after deinterlacing. High definition interlaced video shows distinct comb-effect distortion on moving vertical edges, which are removed by deinterlacing. The length of the comb “teeth” is proportional to the speed of motion and can be many pixels in length. All in all, I prefer to see smooth motion and accept the loss of vertical resolution as a necessary evil.
Mixing UFF and LFF video in the same project might be better done by converting both to Frame-Based because the resulting resolution would be only slightly diminished.
The new tests showed that
1) The worst distortions occurred when UFF (Upper Field First) video was re-rendered to LFF (Lower Field First) and vice versa. Not only was there significant loss of resolution but also the image was “doubled” in the vertical axis. This has got to be a serious bug because the fields in the test file were identical (subject unmoving).
2) Converting from UFF or LFF to FB (Frame-Based) reduced the vertical resolution by one step from 5.25 to 4.25. Horizontal resolution was unchanged at 6.0, although vertical edges were softer.
3) Converting from Frame Based to UFF or LFF showed no loss of resolution.
4) High definition video (1440x1080) that is cross rendered from Upper Field First to Frame-Based, as when deinterlacing, showed a loss of vertical resolution from 9.0 to 7.0.
Comment
Result 4) is problematic for me because I see smother motion on moving subjects after deinterlacing. High definition interlaced video shows distinct comb-effect distortion on moving vertical edges, which are removed by deinterlacing. The length of the comb “teeth” is proportional to the speed of motion and can be many pixels in length. All in all, I prefer to see smooth motion and accept the loss of vertical resolution as a necessary evil.
Mixing UFF and LFF video in the same project might be better done by converting both to Frame-Based because the resulting resolution would be only slightly diminished.
-
sjj1805
- Posts: 14383
- Joined: Wed Jan 26, 2005 7:20 am
- System_Drive: C
- 32bit or 64bit: 32 Bit
- motherboard: Equium P200-178
- processor: Intel Pentium Dual-Core Processor T2080
- ram: 2 GB
- Video Card: Intel 945 Express
- sound_card: Intel GMA 950
- Hard_Drive_Capacity: 1160 GB
- Location: Birmingham UK
John
Thank you for sharing the results of your extensive tests. I am aware that you mainly work with High Definition Video. Could I ask how you obtained your source material. I have a suspicion that the success you enjoy might possibly be something to do with the equipment that you use producing a crisper and cleaner initial file. If so this might also explain why you are able to edit MPEG files without suffering other issues experienced by other forum members. In particular the audio/video synch issue.
You wouldn't strongly support editing MPEG files if you suffered these problems which clearly you do not. Conversely many other experienced forum members have stated that they do suffer these issues.
This leads me to conclude that it is possibly the superior quality of your High Definition equipment.
Thank you for sharing the results of your extensive tests. I am aware that you mainly work with High Definition Video. Could I ask how you obtained your source material. I have a suspicion that the success you enjoy might possibly be something to do with the equipment that you use producing a crisper and cleaner initial file. If so this might also explain why you are able to edit MPEG files without suffering other issues experienced by other forum members. In particular the audio/video synch issue.
You wouldn't strongly support editing MPEG files if you suffered these problems which clearly you do not. Conversely many other experienced forum members have stated that they do suffer these issues.
This leads me to conclude that it is possibly the superior quality of your High Definition equipment.
-
heinz-oz
And don't come back with your VS8 versus VS9 differences. It might well be that there were problems fixed with VS9 (sad if they weren't) but I have had similar problems with editing mpeg files using MFP7.3
I would love to trust your recommendations but, unless someone else can confirm your findings, I will not even waste a minute on any tests. My approach of working with DV-AVI works perfectly well and I do not see a good enough reason to change that. I will also not advise people to do mpeg editing/rendering if they have the possibility to work in DV-AVI.
I would love to trust your recommendations but, unless someone else can confirm your findings, I will not even waste a minute on any tests. My approach of working with DV-AVI works perfectly well and I do not see a good enough reason to change that. I will also not advise people to do mpeg editing/rendering if they have the possibility to work in DV-AVI.
-
jchunter
Steve, Edited for clarification in italics.
All tests (except 4, which is high def) were done with standard definition, DVD compliant video: Mpeg2, 720x480pixels, bitrate = variable 8Mbps, Field Order: primary test variable: where Test files, created with one field order were converted to a different field order.
The Cross rendering Test files were made separately for each test (LFF -> UFF, LFF->FB, etc.) with the following sequence:
1) New Project, preset Preferences and Project Properties
2) Insert ISO12233 jpeg into timeline, with 10 sec duration
3) Share/ Create Video file/ Custom/ set properties to produce mpeg2 file with desired characteristics - Field Order is primary TEST variable.
4) Setup Non-Square Pixels and Smart Render properties (secondary TEST variables) according to test currently being run.
5) Produce TEST video file.
6) Test file was placed in Edit timeline, then steps 3, 4, and 5 were repeated to convert the Test file to a different field order.
7) Resolution was measured with Virtual Dub Mpeg at 3X magnification on 1280x1024 pixel monitor.
All in all, I ran about 90 separate tests on different standard definition configurations, including some with Motion1 video. I ran 8 more on the effects of deinterlacing HD video.
All tests (except 4, which is high def) were done with standard definition, DVD compliant video: Mpeg2, 720x480pixels, bitrate = variable 8Mbps, Field Order: primary test variable: where Test files, created with one field order were converted to a different field order.
The Cross rendering Test files were made separately for each test (LFF -> UFF, LFF->FB, etc.) with the following sequence:
1) New Project, preset Preferences and Project Properties
2) Insert ISO12233 jpeg into timeline, with 10 sec duration
3) Share/ Create Video file/ Custom/ set properties to produce mpeg2 file with desired characteristics - Field Order is primary TEST variable.
4) Setup Non-Square Pixels and Smart Render properties (secondary TEST variables) according to test currently being run.
5) Produce TEST video file.
6) Test file was placed in Edit timeline, then steps 3, 4, and 5 were repeated to convert the Test file to a different field order.
7) Resolution was measured with Virtual Dub Mpeg at 3X magnification on 1280x1024 pixel monitor.
All in all, I ran about 90 separate tests on different standard definition configurations, including some with Motion1 video. I ran 8 more on the effects of deinterlacing HD video.
Last edited by jchunter on Sat Aug 26, 2006 6:37 pm, edited 2 times in total.
-
sjj1805
- Posts: 14383
- Joined: Wed Jan 26, 2005 7:20 am
- System_Drive: C
- 32bit or 64bit: 32 Bit
- motherboard: Equium P200-178
- processor: Intel Pentium Dual-Core Processor T2080
- ram: 2 GB
- Video Card: Intel 945 Express
- sound_card: Intel GMA 950
- Hard_Drive_Capacity: 1160 GB
- Location: Birmingham UK
John
What hardware was used?
Method of connection to the computer?
"Field Order: primary test variable." most people refer to Upper Field First / Lower Field First : A/B : 1/2 :
by "primary" I take it you mean Upper Field First
by "test variable" I take it you mean a variable bit rate and not a constant bit rate.
What hardware was used?
Method of connection to the computer?
"Field Order: primary test variable." most people refer to Upper Field First / Lower Field First : A/B : 1/2 :
by "primary" I take it you mean Upper Field First
by "test variable" I take it you mean a variable bit rate and not a constant bit rate.
-
jchunter