FCP Tips & Tricks
This page is, and is intended to be a "Work in Progress". Below are tips or methods that I find particularly useful or interesting. It is not a comprehensive guide.
Making Broadcast Quality Quicktimes with H.264 or Mpeg-4
The following suggested procedure will produce a broadcast quality movie with a small file size; typically 100-160 Megabytes for a :30 spot. I get the best results with H.264 compression. MP-4 doesn't handle the color quite as well, and file size is somewhat larger, but it's performance can be improved some as described below. These methods can be adapted to any Non-Linear application.
The first step is to export your sequence as a Quicktime movie. In FCP this is a "Quicktime Conversion". It's under the "file" menu and "export". FCP (or Quicktime Pro) will bring up a window that asks you where you want to save the file. At the bottom of that window are more options. Be sure that you're saving as a "Quicktime movie", then hit the "Options" button. That will bring up a window that looks like this although the words in the video and sound boxes will probably say something else (at first):
Now click the Video "Settings..." button. That will bring up a window that looks like this:
Now you'll be back to the "Movie Settings" window ... the first one I put up. Your video settings should match that portion. Make sure the size is correct ... typically 720x486. If you use DV settings in FCP it would be 720x480.
To change Audio settings click the Sound "Settings" button. Use whatever you want, but I usually use "None" to avoid compatibility problems with other platforms.
I uncheck "Prepare for Internet Streaming", unless that's sepcifically what I want the file to do.
If "H.264" is unavailable either to you or to the system to which you are delivering the video, MP-4 compression is a decent alternative. You set up the "Standard Video Compression Settings" window (the lower one of the two above) the same way, except, of course, the "Compression Type" is MP-4 rather than H.264. The most important things to remember are:
Key Frames: 6
Frame Reordering; off
Quality: Best
Data Rate: Automatic
One last thing to do if you use MP-4 (don't do this for H.264). In the "Movie Settings" window select "Filter..." and, under "Adjustments", turn the brightness down to between -3 to -7, and the contrast up to between +3 to +7.
back to top
"Filmlook"
I don't know of any way to make video look just like film, and I've tried. However, it is possible to achieve a look that's between the two. There are various "filters" or "Plug-ins" that attempt to accomplish the goal. I use a different method, and I think it gets better results. I explain both the method, and why I think it works better than "filters" below. If you are familiar with "filmlook", "frame rates", "video fields", and "interlacing" you might want to skip down to the part called My Method.
There are 2 differences between film and video pertaining to the "look" that's displayed. First ... film is displayed at 24 images per second, video at 60 images per second. Second ... there is the quality of the individual frame ... film captures color bettter than video, and video tends to look harsher. To make video look like film one has to minimize these differences.
-Image Rate
I'm going to use the term "images per second" rather than "frames per second" because although video supposedly runs at 30 frames per second, each frame is composed of 2 fields, each different from the other. Video really presents something to the eye at a rate of 60 images per second.
Here's a little further explanation. Under the NTSC system TV sets present a "frame" of video by drawing on a screen 484 lines (top to bottom) 30 times a second ... BUT ... it takes 2 passes to make those lines. It first paints 242 (half of the 484) lines in one 60th of a second, and then paints the remaining 242 lines in-between the first ones in the next 60th of a second. Then it starts over again for the next frame. This is called interlacing. So really a new image is being presented to the eye each 60th of a second. That's what gives video its "live" look. Here is a close-up of a screen showing the first field and the second field. It's difficult to see, but the scan lines on the second image are exactly in-between the scan lines on the first image. These two images comprise one frame:
To make video look more like film the video image rate (60) needs to be reduced to as close to the film image rate (24) as possible. It's not possible to make NTSC video display at 24 images per second, but it is possible to make it appear to display 30 images per second (rather than 60), and that's close. This alone can make video look more like film.
This is accomplished by processing the 2 field images in such a way that, even though the screen draws 2 images per frame, both of those images, or fields, contain identical content. In other words, had this process been applied to the above images, both would look the same (no perception of movement), and if both fields appear identical to the eye then the eye perceives 30 images per second rather than 60.
-Image Quality
Film captures color better than video, handles light better than video, and has higher resolution. Film has a more natural look. Video looks harsher. There is no way you can make video have more colors (or shades of color). There are, however, several things that one can do to change the video image to make it look more like film. One is to increase the saturation (carefully, and not too much), and the other is to slightly soften the image (again carefully, and not too much). One can also tweak the highlight and shadows levels.
-Other Methods
Often "filmlook" methods perform what is called "de-interlacing". Fields (those lines that are on the image above) are referred to as "upper" and "lower", or "odd" and "even". Typical "filmlook" methods simply remove or ignore one of those fields, and replace it by duplicating the other. Now the video screen displays exactly the same thing for the first field as it does for the second. So, even though the screen paints on the image in two fields, each taking a 60th of a second, as described earlier, it paints the same image both times. The 2 sets of lines look just the same. There is no sense of movement between the two fields because there is nothing different about them. This reduces the perceived image rate to 30 images per second (instead of 60).
-My Method
My Method doesn't require a plug-in or filter, but it does require PhotoShop, although there may be other graphics applications that can do the same thing.
I first export my clip as an "image sequence". Image sequences make individual graphic files of each frame. You need to create a folder in which to export this sequence. If your clip is 5 seconds long an image sequence export will produce 150 individual graphic files (5 seconds x 30 frames = 150). FCP supports export of many graphic formats. I usually use jpeg or pict. The export process automatically numbers the files so that they can be read and processed in the right order.
Once my clip has been exported to a folder (containing all the frames converted to graphic files) I turn to PhotoShop for "Batch Processing". Batch Processing will perform a set of PhotoShop actions to multiple files. First one has to create this set of actions, and save that set with a name of your choice. This is done by performing these actions on one individual file, recording those actions, and saving them as an "Action". Once this set of actions is saved it can be applied to a group of files. Under the "Files" menu you'll find "Automate", and then "Batch...". Batch Processing will perform your set of actions on each file in a folder and output the results as new graphic files in a new folder (you have to first create the designated output folder). The actions that I perform are listed below:
Image Size - Height: 300% ("Constrain Proportions" is off, no width change) - Interpolation: bicubic
Motion Blur - Angle: 90 - Distance: 5 pixels
Image Size - Height: 486 pixels (with DV images this would be 480 pixels) - Interpolation: bicubic
Save (as whatever graphic format you want the output to be. I use jpeg or pict)
Here's what those actions do. First the image Height is tripled (stretched). Then vertical blur is applied to the stretched image. Next the image height is brought back to it's original size. The interlaced scan lines will no longer be apparent. This has the effect of de-interlacing by blending the two fields. The usual methods throw away an entire field (half the image). This process preserves the information from both fields. The blur, though vertical only, softens the image slightly. Both fields will be apparent on the final file (may look like ghost images), but not separated by scan lines.
Next I import the image sequence back into FCP and color correct as needed, usually a little saturation boost.
Notes:
1.) You may not see the difference between a normal video clip and a "filmlook" processed version on your computer. You should be able to see the difference on a video monitor.
2.) Film is transferred to video using a process known as "3:2 Pulldown". The result is 24 fps film is converted to 30 fps video, but that video does not display at 60 images per second. It approximates a display of 30 images per second.
3.) Some people like to add various kinds of noise to the "filmlook" video.
back to top
Square and Non-Square Pixel Issues
Why do some graphic files look different on TV or in the editing application than they do on the computer on which they were made? Usually it's because the two systems treat pixels differently. It's a fixable problem.
Computers, graphics applications, and computer monitors typically display graphics in "square pixels". In other words the pixel is as wide as it is high. Video displays (TV sets, video monitors, etc.) squish those images from the sides. They display the pixels as "non-square", a little less width than height. The result can be that a perfectly fine logo or graphic file created on a computer can look wrong on the TV screen. On some graphics the difference isn't very noticeable, but on others it is very noticeable. Circles, as an example, can turn into vertical ovals when displayed on video.
The problem can be solved by doing a few things when the graphic is created. Rather than explain I'll refer to Apple's document on the subject. It's pretty thorough, but easy to understand.
If you're stuck with a graphic that doesn't look right when editing you can adjust the aspect ratio in FCP under the "Motion" tab and "Distort".
Newer versions of PhotoShop support non-square pixels, or so I've read.
backto top
Applying Alpha to a Video Track in FCP
FCP makes it easy to apply an alpha to a video track. Assuming that one has a base layer (V1) of video, and wants to super something over that layer here's what you do. Place the alpha layer on the next track up (V2). This alpha can be a still image or video. Best results will occur with black and white alphas. Place the video or material (fill) that you want to super (using the alpha) on the track above the alpha track (V3). Next select the track above the alpha (V3). Now go to the "Modify" menu, go down to "Composite Mode", and select "Travel Matte - Luma". What you will see is the base layer with a super of video track 3 that corresponds to the white portions of the alpha on track 2.
A circumstance might come up where you want to use a graphic file that has an alpha channel, but you want to fill with something different than what that graphic offers. You can use the alpha channel from that graphic. Follow the same procedure as above, placing your graphic with alpha on V2. Use "Travel Matte - Alpha". FCP will read the alpha from your graphic and apply it to the material on V3.
The alpha track can be re-sized. For instance, you can make it larger. More of the video layer above will show through as a result, but that layer (V3) will not be re-sized. Motion can also be applied to the alpha track.
The composite mode offers lots of possibilities.
back to top
Adjusting Audio Levels on Multiple Segments
Final Cut Pro offers an easy way to affect multiple audio segments. This is handy under circumstances like the following:
Let's say that the narration consists of on-screen talent segments, off-screen vo, and the material comes from different locations so the incoming audio levels are all over the place (not to mention eq). So after balancing out many segments of voice track so that their levels match each other you discover that you need to boost all of those tracks by 6 db. Just do this. Select the audio segments you want to affect (could be all of them), and hit "Command-Option-L". A small window will appear called "Gain Adjust". You can increase or decrease the level of everything selected all at once. The "Gain Adjust" windows offers a choice of "relative" or "absolute". My choice is usually "relative".
back to top
Applying an Effect to an Entire Nested Sequence
Hold down the "Option" key when selecting the nested segment. Now any effect added or motion change will apply to everything within the nested segment.
back to top
Links
Video Shooting Tips for Non-Professionals
Final Cut Pro Tips & Tricks for Experienced Editors
I encourage questions and/or comments?
CreativeOasis Production Services