High Resolution Video Processing

This page gives instructions for processing video that you captured in high resolution.  If you captured in low resolution, you should go to my Low Resolution Video Processing section.

In video, background noise is the unwanted movement of pixels in your movie, even when there should be no motion.  Most analog video sources tend to have a lot of background noise in them, especially if you aren't using professional video equipment.  Fortunately, by using some of VirtualDub's plug-ins, we can significantly reduce the background noise.  This improves the image quality and results in a better video after it's been encoded to its final format.

You have the choice to either save your processed video to an AVI compressed with Huffyuv or frameserve directly to your video encoder.  To learn about frameserving, read my Frameserving article.  If you choose to save to an AVI, you'll need a lot of free disk space.  For example, if your final resolution will be 640 x 480, you'll need about 4.5-7.0 GB of free space to store each 10 minutes of your processed video with Huffyuv.  If your final resolution is going to be 320 x 240, you will need around 1.5-2.75 GB of free space to store each 10 minutes of your processed video with Huffyuv.

Make sure that you have installed all the plug-ins that I mention in the VirtualDub Plug-ins section.

A)  Classify Your Video

There are many possible formats for high resolution video.  In order to optimally process your high resolution video, you must determine the kind of video you are working with.  For example, your video can be field swapped, NTSC telecined, field shifted, or from a truly interlaced source.
If you do not know the format of your video, please go through my article entitled Classifying Your High Resolution Video.  If you have any questions, you can post them on my Message Boards.

B)  Configuring the Audio Options

If you did not use Cool Edit 2000 to edit your audio, follow these steps:

1)  Make sure that Audio -> AVI Audio is selected.

2)  Make sure that Audio -> Direct stream copy is selected.

If you did use Cool Edit 2000 to edit your audio, follow these steps:

1)  Click on Audio -> WAV audio.

2)  Select the .wav file you edited in Cool Edit 2000 (perhaps capture.wav) and hit OK.

3)  Make sure that Audio -> Direct stream copy is selected.

C)  Trimming the Video

1)  The first thing you will do is trim off the unwanted portions at the beginning and end of your captured video file.  To begin, run VirtualDub, go to File - > Open Video File, make sure the "Automatically detect and load additional segments" box is checked, and open the file you captured in AVI_IO.  (The file name you want to open is usually "capture.00.avi".)

2)  Using the video slider and the Play button with the little letter "I" next to it, find the spot where your program begins and press the stop button.  Usually, the screen will fade in from black a second before a program begins.  Using the video slider on the bottom of the VirtualDub window, move to a position approximately 1/2 to 1 second before the blackness disappears and your program starts.  When you are in a position 1/2 to 1 second before the program starts, press the END key on your keyboard.

3)  Using the video slider, rewind back to the very beginning of the video.  When the slider is in the left-most position, press the HOME key.  Then press the DELETE key to remove the unwanted portion of the video before your program starts.

4)  Now use the video slider to move to the very end of your program.  When the slider is at a point approximately 1/2 to 1 second past the end of your program, press the HOME key.

5)  Move the video slider all the way to the right and press the END key.  Finally, press the DELETE key to delete the extraneous video.

6)  Listen to both the very beginning and very end of your video.  You can do this by moving the position slider to either point and pressing the play button.  If you did a good job of trimming the video, there will not be any audio that has been cut off at the beginning or end of the video.  If there is some introductory or ending audio in your program that has been cut off, start over from Step #1.  This time, include a longer stretch of video before or after your program ends.  (Note:  if you have a slower computer, playback of the audio and video may be choppy.  This will not affect your final encoded output.)

7)  If there are any commercials in the middle of your video, you can also remove them.  Just select the unwanted portion of the video as described in the previous steps with the HOME and END keys.  Then press DELETE to remove the commercials.

8)  Finally, play the video through the portion where you removed the commercials to ensure that you did not cut out any desired audio.  If you did, you will have to restart from Step #1.  This time, try to make your selection around the commercials not quite as close to the desired video.

D)  Configuring the Video Options

Once your video has been properly trimmed, the next step is to configure the video and plug-in (filter) options.  For the sake of simplicity, I've tried to avoid a technical description of the filters.  However, you are free to ask questions in the Message Boards.  With that being said, let's begin:

1)  If your video is field swapped, click on Video -> Filters -> Add, choose "field swap", and hit OK.

2)  If your video was NTSC telecined, then you should inverse telecine your video to restore the original ~24 fps.
(Note:  If you are making a VCD or SVCD, be sure to read the important information in my page called VCDs/SVCDs at 23.976 fps.)

To inverse telecine your video, click on Video -> Frame Rate, go to the bottom section, and choose the option labeled "Reconstruct from fields - adaptive".  The window should look like this:

Click OK.

3)  If you video is field shifted (usually found in PAL/SECAM movies), click on Video -> Filters -> Add and double-click 'telecide' in the list.  You can leave the settings at their defaults, which should look like this:

Click OK twice to return to the main VirtualDub window.

Since the Telecide filter introduces a one-frame delay in your video, the audio will be desynchronized by one frame.  To fix this, first click on Audio -> Interleaving.  Then change the "Delay audio track by" option to 40 if your video is PAL/SECAM (25 fps) or 33 if your video is NTSC (29.97 fps).  You do not need to alter any other options, so just click OK when you are finished.  The settings should look similar to this:

4)  The next step is to insert a deinterlacing filter, if necessary.  You should only deinterlace your video if any of the following is true:

A)  Your video is from a truly interlaced source, and you are not making a SVCD.  (SVCDs don't need deinterlacing because they support interlaced video.)

B)  You are using inverse telecine or the Telecide filter, you are very concerned about any interlacing artifacts (horizontal lines) being left in your video, and you don't mind a slight loss in detail.  (The inverse telecine function and Telecide filter are not 100% accurate, so interlacing artifacts can be left in your video.  The deinterlacing filter will remove these artifacts, but the filter may also deinterlace inappropriate portions of your video, leading to a loss in detail.  Personally, I don't think the artifacts left by inverse telecine and Telecine are that noticeable, so I prefer not to deinterlace and risk a loss in detail.)

Click on Video -> Filters -> Add and double-click "deinterlace - area based".  If you check the box for "Blend instead of interpolate", your output video may appear to play more smoothly, but motion will be blurred and there will be ghosting.  If you uncheck the box for "Blend instead of interpolate", your output video will be sharper during motion and not have any ghosting effects, but motion might not appear to be as smooth.  I think most people prefer to leave the box unchecked, but the choice is up to you.
Next, change the Threshold to 75 and the Edge detect to 0.  Also, the option to "Show deinterlaced areas only" should not be checked.  The window should now look similar to this:

Click OK twice to return to the main VirtualDub window, and then begin scrolling through your video.  Look for portions in your video that have interlacing artifacts (horizontal lines).  In the output pane (the video on the right), there should not be any noticeable interlacing artifacts left in your video.  If you do see interlacing artifacts, click on Video -> Filters, double-click "deinterlace - area based", and lower the threshold by 5.  Now hit OK twice and start scrolling through your video again.  If you can still see interlacing artifacts, lower the threshold by another 5.  Keep repeating this procedure until there are no visible interlacing artifacts in your video.

Note #1:  The higher the threshold you use, the more detail will be left in your video.  Therefore, it's best to pick the highest threshold that removes the visible artifacts from your video.

Note #2:  If you are deinterlacing your video to remove interlacing artifacts that were missed by inverse telecine or Telecide, try to keep the threshold between 70 and 80.

5)  If you captured your video in a resolution that differs from your desired output resolution, you need to resize the video.  If your video is already in your desired resolution, then you may skip this step.
Click on Video -> Filters -> Add, select the "resize" filter from the list, and hit OK.  You should now see a window like this:

In the "New width" and "New height" boxes, enter your desired output resolution.  Then select a Filter mode of "Precise bicubic" or "Precise bilinear".  The "Precise bicubic" option is more commonly used because it produces a sharper image.  Using 'A=-0.60' may provide a more faithful reproduction of the image, whereas using larger negative values can sometimes provide an even sharper image.  You can choose the "Precise bilinear" option if your video source if very noisy or you want a slightly smaller video after you encode it to its final format.  However, the "Precise bilinear" option will blur the video a bit, so "Precise bicubic" usually looks better.  When you are done configuring the resize filter, hit the OK button.

Note:  When entering your desired output resolution, it is important to choose a width/height combination that preserves the correct aspect ratio.  All of the output video resolutions I have listed on the Video Capture Overview page maintain the proper aspect ratio, so you should use one of them.  In general, the correct aspect ratio of width to height is 4:3 (1.333) for video that is intended to be viewed on a computer monitor.  Some formats, such as VCD, have different aspect ratios because they are not meant to be watched on a monitor.

6)  The next step is to crop off unwanted potions of the image.  Click again on Add, select the "null transform" filter from the list, and hit OK.  You should see text similar to "640x480 640x480 null transform" listed in the Filters window.  Highlight this filter and select Cropping.  You should see a window like this:

If you are capturing from an analog video source, look closely at the top and bottom of your video.  It's very likely that you will notice what appears to be lines of video "garbage".  These lines will appear to move from left to right as you scroll through your video, and they are sometimes a different color from adjacent portions of the image.  The lines make the video look bad and increase the file size, so we are going to crop (cut) them off.  You can do this by changing the Y1 and Y2 offsets.  Bringing the Y1 offset down will cut off the top of the image, and bringing the Y2 offset up will cut off the bottom of the video.  Your goal is to cut off all of the video garbage.  In addition, if your video has black borders on the top and bottom, you should crop off those too.  Furthermore, if you have any unwanted video on the left or right sides of the image, you can use the X1 and X2 offset to crop the sides.  However, if you do not have any video garbage, do not crop off anything.

Note #1:  Most video encoding programs have requirements about the resolution of the video you want to encode.  For example, most MPEG-1 encoders require that width and height of your video are divisible by 16.  Because of these kinds of requirements, you may be limited in the number of lines that you can crop off.  For instance, let's say you crop off 3 lines of video garbage, but the encoder needs the resolution to be a multiple of 4.  In this case, you would have to crop off 1 more line, either from the top or the bottom.  In other words, the sum of the Y1 and Y2 offsets would have to be 4 (or a multiple of 4).  If your video had 5-8 lines of video garbage, you would have to crop off 8 lines, which is the closest multiple of 4.  The same would be true if you were cropping the left and right sides of the video with the X1 and X2 offsets.  Here are some requirements for common video encoders:

• If you plan on encoding your video to the RealMedia format,  you are required to crop off a multiple of 4 lines.

• If you are going to encode your video to an MPEG-4 format (such as MS MPEG-4 V2), you should try to crop off a multiple of 8 lines (acceptable) or 16 lines (preferable).  However, a multiple of 4 will also work.  You may want to be careful though because many video cards disable video acceleration for playback if the resolution isn't a multiple of 16.

• If you want to compress your video to an MPEG-1 or MPEG-2 format, you should crop off a multiple of 16 lines.  However, if you are making a VCD or SVCD, you can crop off as many lines as you like.  Having a multiple of 16 is still preferred, but it's not necessary since we will be letterboxing the video if you are making a VCD or SVCD.

When you are satisfied that you have cropped off all the unwanted video, hit OK.

7)  Note:  This step should only be followed if you are making a VCD or SVCD and you cropped off part of your video in the previous step.
To begin, look at the text for the null transform filter in Filters window.  The first part of the text will contain the resolution for your cropped video.  (For example, you might see "352x236".)  Remember this resolution, and then hit the Add button and choose the resize filter.  For the New width and New height boxes, enter the resolution that you just memorized from the null transform filter.  The Filter mode is unimportant because we won't be modifying the actual size of your video.  Next, check the box labeled "Expand frame and letterbox image".  For the Frame width and Frame height, enter the resolution required for the type of VCD or SVCD that you are making.  (NTSC VCD = 352 x 240, PAL/SECAM VCD = 352 x 288, NTSC SVCD = 480 x 480, PAL/SECAM SVCD = 480 x 576.)  The default Fill color of black is fine.  If you want to, you can click on the preview button to ensure that your video is the proper size and has black borders on one or more of its edges.  The settings for the resize filter should look something like this, although your numbers may be different:

Click OK when you are finished.

8)  You should be back at the filters window.  Now we will insert a temporal noise reduction filter, which will eliminate much of the random static (noise) in your video.  Hit Add to add another filter.  From the list, select "temporal smoother" and hit OK.  Then a dialog box with a slider will pop up.  Each notch on the slider represents a level of strength from 0-10.  If your video is rather noisy, set the strength to 6 (the 7^th notch from the left).  If your video has little noise, set the strength to 5 or possibly lower.  Here is a picture of the Temporal Smoother filter configured to a strength of 6:

Click OK when you have selected the proper level of strength.

Note:  If you plan on frameserving your video, you won't be able to use the temporal smoother filter.  As a replacement, I suggest using the "Dynamic noise reduction (original)" filter, which can be obtained from the VirtualDub Plug-ins page.  If your video has little noise, try setting the threshold to about 16 or lower.  If your video has a moderate to high amount of noise, trying using a threshold of 25 or lower.  For extremely noisy video, you can use a threshold up to 31, but ghosted images are likely to start appearing during motion as you increase the threshold.

9)  Next we are going to add a spatial noise reduction filter, which will further smooth out the noise in your video.  Click the Add button, select "smart smoother" from the list, and press OK.  The window that appears should look like this:

I suggest configuring Smart Smoother to a Diameter of 7 and a Threshold of 45.  If your source video isn't too noisy, you might want to lower the Threshold a little bit.  

Note:  If you are making an interlaced SVCD, you should check the box labeled "Interlaced source".

When you are done configuring Smart Smoother, hit OK.

10)  Note:  In steps #10 and #11, we will be adding another pair of noise reduction filters that should eliminate any remaining static noise in your video.  If your source video doesn't have a lot of random static motion, you may wish to skip steps #6 and #7.

Click the Add button again, chose "dynamic noise reduction (original)" from the list, and press OK.  Then set the Noise threshold to 6, which should be sufficient to freeze any remaining static motion in your video.  The configured window should look like this:

Click the OK button after you have configured the filter.

11)  Note:  Only follow this step if you have also followed step #10.

The final noise reduction filter will be another instance of Smart Smoother.  Therefore, click on Add, select "smart smoother", and hit OK.  This time, our goal is merely to smooth out any pixels locked in place by the Dynamic Noise Reduction filter.  Therefore, I suggest configuring the filter to a Diameter of 5 and a Strength of 8.  Here is a picture of the configured filter window:

Note:  If you are making an interlaced SVCD, you should check the box labeled "Interlaced source".

When you have properly adjusted the settings, click the OK button.

12)  With all your filters configured, the Filters window should look something like this:

Of course, your list of filters could look quite different from this sample list.  You are probably using different settings, and certain filters may not be present in your list.  You could also have some additional filters in your list.  As long as you followed the previous steps on this page, you should be fine.

13)  With all of your video filters set up, click OK to exit the Filters window.  Now click on Video -> Compression.  Select Huffyuv 2.1.1 from the list and hit configure.  (If you don't see Huffyuv in the list, visit my Software page.)  Adjust the settings to match those found below:

When you've configured Huffyuv, hit OK to exit the Huffyuv dialog box and hit OK again to exit the Video compression menu.

14)  Make sure Video -> Full processing mode is selected.


E)  Begin the Processing

You now have the choice of saving your processed video to a file or Frameserving it directly to your desired video encoder.  The advantage of frameserving is that it produces no intermediate files, which saves disk space.  The disadvantage is that your video will have to be re-processed every time you encode it, which can take a long time.  If you have enough free hard drive space, I usually recommend saving your processed video to a file instead of frameserving it.

If you choose to frameserve your video, follow only Step #1 below and then go through the instructions for VirtualDub frameserving found in my Frameserving page.  If you want to save the video to a file, continue following all of the steps below.

1)  If you are using the Telecide filter, you must click on the rewind button before starting to process your video.  (The rewind button [|<] is the fourth button from the left at the bottom of the VirtualDub window.)  Once you have done this, do not attempt to play or scroll through your video.  Immediately move on to the next step and do exactly what it says.

2)  Depending on your operating system and file system, you can save your processed video in one of two ways:

A)  If you are running Windows 9x/ME, Windows NT with the FAT16 file system, or Windows 2000/XP with the FAT32 file system and your captured file(s) are 2 GB or larger (in total), click on File -> Save segmented AVI.  Type in a file name, choose a file segment size of 2000 MB, and hit the Save button.  (You have to use the segmented option because the FAT file systems do not support files bigger than 2 GB or 4 GB).  This method of saving your processed video is the most reliable, so you should use it if you are unsure about the file size of your output video or capabilities of your operating system.

B)  If your captured file(s) are less than 2 GB (in total) or you are running Windows NT/2000/XP with the NTFS file system, click on File -> Save AVI, type in a file name, and hit Save.  (Windows NT/2000/XP with the NTFS file system can handle files of virtually unlimited size.)

A progress indicator should appear, telling you how long it will take until the video finishes processing.  It will also tell you the estimated file size of the processed file.  Make sure you have enough free hard disk space to accommodate the processed file.  You can uncheck the options labeled "Show input video" and "Show output video" to increase performance.  While your video processes, feel free to do whatever you like.  You can go outside and enjoy the fresh air.  Or, if you're like me, just set up your video to process before you go to bed.  When you wake up in the morning it should be done.

3)  Finally, do not forget to open and test your new video file after the processing.  Make sure that the video looks and sounds good.  (Note:  if you have a slower computer, playback of the audio and video may be choppy.  This will not affect your final encoded output.)

You're almost there!  You can now move on to the Encoding Overview.