Dear All,
I took VOB files which are 29.97 fps and converted them using HuffyUV AVI in VirtualDub.
I edited this video in Premier and exported using the Huffyuv AVI at 29.97 fps. The result is jerky video.
Any suggestions?
I used the exact same method and did all of the above for 23.976 fps video and it worked fine with the 2:3 pulldown.
Does Premier not like the codec I am using with 29.97 fps or does it only like Microsoft DV AVI when it comes to 29.97 fps. I was reading that Panasonic is more friendly with the hard telecine frame rate.
Any help would be appreciated.
Thanks,
Wasim Ismail.
+ Reply to Thread
Results 1 to 28 of 28
-
-
Are the files 29.97 with pulldown?If you exported them as 29.97 then it will look that way.If they are hard telecine then you have to do a Inverse telecine which premiere cant do,avisynth can do what you want.Theres all kinds of script that you can search in the forums.
I think,therefore i am a hamster. -
Really fast jerky (30 back and forth per second) when there is movement? That would be caused by setting the wrong field order.
-
Originally Posted by johns0
I will mention something here which relates to my problem and this will definitely narrow it down.
I was making a video in Premiere which basically had a picture fade in and then an object move across the screen. Now with an object moving across the screen you will definitely know if the video is jerky or not.
Anyway, I tried making this video using the Huffyuv compression and no matter how I did it, the result was jerky. Whether I exported it as 23.976 fps or 29.97 fps it made no difference.
I did the exact same thing using the Microsoft DV AVI (Premiere's default compression other than the Video for Windows compressions) and it was perfect.
So the problem is with Premiere exporting any video at 29.97 fps when using Video for Windows compressions. My example above wasn't even taking VOB files, it was simply a BMP file that I took into Premiere, gave it a fade in effect, and then took a Photoshop file and had the object in it fly across the screen with sound effects. The fade in was fine both ways, the object moving across the screen was the problem.
Does that explain it better? Perhaps I am doing something wrong with the VFW standards? -
Originally Posted by wasimismail
Originally Posted by wasimismail -
Originally Posted by jagabo
Also, why didn't it matter when using the Microsoft DV AVI or is that field order in that one already preset? -
DV is bottom field first. VOB/MPG can be either top or bottom field first. More often than not it's TFF. If Premiere misidentified the source files field order you would expect jerky results.
-
Originally Posted by jagabo
So I should make sure that I enable BFF when exporting anything from Premiere.
OK why are the VOBs TFF if they are supposed to be BFF in the first place (in initial production) ?
BFF is the standard then correct ?
I am new in this particular area so thanks for your past explanations and ones coming up. -
My other question is this. If the field order was the problem, then why didn't I encounter that when exporting 23.976 fps video from Premiere ?
-
If the field order was the problem, then why didn't I encounter that when exporting 23.976 fps video from Premiere ?
For the other VOB's , if they were telecined, I would remove telecine so you have progressive content before editing. You might do this in avisynth and/or vdub
Don't enable TFF or BFF automatically in the settings, this must be determined manually. But if you get the order wrong, it should be clear in the preview window, just try to swap the order then by interpreting the footage (right click the asset in the clip bin) -
Each frame of interlaced video contains two half pictures called fields. One field consists of all the even scanlines, the other all the odd scanlines. These two fields are meant to be viewed one at a time, in the correct sequence. If the video is TFF the top field (all the even scanelines) is to be viewed first, then the bottom. If you view them the other way around you will see very fast jerks as all motions will move two steps forward, one step back, two steps forward, one step back...
Your source video has a particular field order. The output has to have the same field order. Or, if you want to encode with the opposite field order, the fields have to be rearranged (some editors can do this, some can't).
With progressive video both fields are to be viewed at the same time. But even on progressive displays (like old CRT TVs), where you see one field at a time, it doesn't matter which field you see first. -
Originally Posted by poisondeathray
I am in VirtualDubMod under Video --> Frame Rate, under Inverse telecine (3:2 pulldown removal), do I choose the "Reconstruct from fields - manual" option to bring 29.97 fps back to 23.976 fps then ? I guess after this I won't have to worry about this field reversal thing do I ? -
I've never used vdub's built in inverse telecine filter - I don't know how good it is or if it even works correctly. There is an issue with cadence and where it first starts e.g. if it's 2-3 or 3-2, and there is a possiblity some bad fields slip though.
Personally I would do as manono suggested above and do it through avisynth, and feed the avs script in vdub to encode to a lossless intermediate. I would use an adaptive field matcher, which should take care of possible cadence breaks or pattern issues.
e.g use dgindex with honor pulldown flags (shouldn't matter if it's hard telecined, there would be no "flags")
MPEG2Source("video.d2v")
TFM()
TDecimate()
But to answer your other question, if the IVTC is correctly done, you should have 23.976 fps progressive (or really 24000/1001) , and the "field order" would no longer apply , because you have progressive content.
Note, Premiere has an .avs import plugin. So you can import the .avs directly without the intermediate. To Premiere, the .avs file looks like an IVTC'ed video. -
Originally Posted by jagabo
-
VirtualDubMod's IVTC filter in manual mode can work perfectly -- if your video is perfectly telecined with no breaks in the pulldown pattern. Unfortunately, that describes very few sources. The automated "adaptive" method can work fairly well with clean sources. I haven't used it in a long time because AviSynth's various IVTC algorithms work better -- fewer mistakes.
-
Not sure if you saw my edited post above - one option is the .avs importer for Premiere.
This means no intermediate files, no time wasted encoding to the intermediate or huge amounts of HDD space taken up.
Each .avs file is like a proxy of the actual video (with IVTC in the script), and avisynth frame serves it into Premiere. So to Premiere, it looks like a 23.976 fps video
Unfortunately there is no .avs EXporter or export frameserver for CS4, but for CS3 and below, you can use debugmode frameserver to export out
You never mentioned what your final goal was? -
Originally Posted by poisondeathray
-
Originally Posted by jagabo
Thanks very much, you've always been helpful. How would a mistake look like if I did it this way, this is animation so would I have the problem of jerky video where the misakes would be for instance ? -
Originally Posted by wasimismail
-
Originally Posted by jagabo
-
OK it's all figured out, thanks to all !
I used GSpot to figure out that the VOB files in questions were TFF so I did the same thing when exporting from Premiere and it worked perfectly.
From what I got, it is better to leave videos in their original frame rates to avoid any potential errors. I tried inverse telecine by using the adaptive setting in VirtualDubMod and I think it is better just to leave the video in its original frame rate.
Thanks again, this is really a great forum. -
Originally Posted by wasimismail
You can try and rationalize not learning the superior methods afforded by the use of AviSynth all you want, but the fact remains that what you did is inferior to what you could have done. -
Originally Posted by wasimismailI think,therefore i am a hamster.
-
Originally Posted by manono
-
Just to beat a dead horse - another issue with "editing" telecined material if your editor doesn't know it's telecined - is that when you make cuts, and they don't happen exactly at the end of the 2-3 (or 3-2) field mark, you break the cadence. Also when effects or overlays or titles are applied, they can be applied to the wrong field.
It's only the case when your editor has interpreted the footage as telecined (with the correct cadence pattern) that it is "safe" to edit it in that format. (i.e it will IVTC it internally and display the progressive 23.976 frames for you to edit) - this is not the case for you.
Here is the procedure in avisynth if you choose to go that route:
1) install avisynth
2) download tivtc.dll, unzip it and put the .dll your avisynth/plugins folder (.dll's in the plugins folder will "autoload")
http://bengal.missouri.edu/~kes25c/TIVTCv105.zip
3) index your vob with dgindex, set the video=> field operation to "honor pulldown flags". If you need to "join" some vobs you can add them in the first file list dialog when you start dgindex. Also copy the dgdecode.dll into your avisynth plugins folder
4) create your .avs script in notepad in the same directory as your vob files. i.e. open notepad and type the following, changing the filenames to match yours, save the file, then change the extension from .txt to .avs
MPEG2Source("vob.d2v")
TFM()
TDecimate()
5) you can open that .avs in vdub to encode to your lossless intermediate to import into premiere later (to vdub, the .avs script looks like a "video" that has been IVTCed) , or open it directly in Premiere as described above with the avs import plugin
All the avisynth plugins and install have documentation for instructions if you need more help. If you still are lost, post again
Cheers -
Originally Posted by wasimismail
For interlace video in Premiere (or other similar NLE) there are three issues you need to coordinate. I'll ignore telecine, progressive or high definition source for now to keep it simple. Let's keep it to SD 720x480i/29.97 or SD 720x576i/25.
1. Source Format: First you need to know your various source formats before you set a project format. Interlace, source can be top field first (TFF) or lower field first (LFF) regardless of codec (e.g. uncompressed, DV, MPeg2, Cineform, ProRes 4:2:2, etc.). Most of these formats have header flags that ID them as TFF or LFF. In some cases the flag is missing or non-standard. Premiere will assume missing flags mean project format. You can manually correct field order flag errors in clip properties.
Typical LFF source includes DV, DVCPro and DVCAM. Most other is TFF. Problem is Premiere and most others default to DV LFF as the project format assuming a DV camcorder (with device control enabled) as source.
2. Project Format: First step when opening a new project is choice of project format. Premiere allows the following project formats but defaults to DV LFF the first time. DV has been a popular consumer and pro camcorder format and requires only an IEEE-1394 port in hardware for capture.
http://www.adobe.com/products/premiere/pdfs/prempro_cs4_2_formatguide.pdf
The project format is the default for the timeline. If you drag a source clip from a bin to the timeline it will play directly if it matches the project format. If not, it will require "rendering" to the project format before it can preview from the timeline. Editor generated objects such as titles or effects moves will be rendered to project format. If the project format is DV and the source is TFF, "rendering" will attempt to convert TFF to LFF for preview playback. Preview render files go to the "scratch" or "tmp" disk.
If the source field order header is missing, Premiere will default it to project format. Preview should reveal a field order problem but only if you preview on an interlace monitor. If the low res progressive computer preview window is used, most users won't notice a field reversal. If an external LCD-TV is used, field order reversal may get blurred out in progressive conversion. This is one reason experienced users use IEEE-1394 out to an interlace monitor (via a DV camcorder or ADVC type device) for preview where field order problems are obvious.
3. Export Format Regardless of project or source format, one can export to project or a different format. Typical export from a LFF DV project is MPeg2 for DVD. Normally one would set the MPeg2 encode to LFF to match the project format but it is possible to convert to TFF (field order reversal processing). You can also specify conversion to progressive on export.
Field order reversal problems that aren't detected when previewing the timeline are passed to the encoding stage. It's important to preview the rendered export file on an interlace monitor before passing it on to DVD authoring or other process. It's a good idea to burn a test DVDR and preview from a DVD player to an interlace monitor as a quality check.Recommends: Kiva.org - Loans that change lives.
http://www.kiva.org/about
Similar Threads
-
Video jerky on MPC-HC
By bapalaps in forum Software PlayingReplies: 6Last Post: 13th Jun 2011, 17:02 -
Jerky video playback
By realmfighter in forum Camcorders (DV/HDV/AVCHD/HD)Replies: 5Last Post: 12th Nov 2009, 01:41 -
jerky video with quicktime 7.1
By clk1946 in forum Newbie / General discussionsReplies: 2Last Post: 6th Aug 2009, 23:54 -
video still jerky
By milway in forum Video ConversionReplies: 15Last Post: 3rd May 2008, 16:01 -
Video is jerky
By arnie22112000 in forum Newbie / General discussionsReplies: 5Last Post: 22nd Apr 2008, 00:49