During the years with Final Cut, I’ve accumulated a number of workflow steps to eliminate some of the “gotchas” and bottlenecks. I’ll share a few in this post and hope you find them to be helpful.
1. Gamma setting
A couple of versions ago, FCP added a gamma settings preference (User Preferences / Editing). This lets you compensate for gamma differences in graphics created on other platforms. Ever since FCP7, I have found that the best setting to use is 2.2 (the native gamma of PCs). “Source” and other options don’t seem to yield the best results. As such, I now set all systems I use to 2.2 gamma.
This gamma preference applies to imported QuickTime movies as well, if they were created with an RGB codec, like Animation. Bring them in as “source” instead of 2.2 gamma and the level will be wrong. Like all other such settings, changes made to this setting only affect files imported after the change was made.
While we are looking at this tab, note the Still/Freeze Duration time. Crank this baby up. You aren’t creating any more media by doing so. If you have it set to 4 minutes, then that means you can have a still or a freeze last up to 4 minutes long as a single clip on the timeline.
2. Graphic Converter
I started a big project at the launch of FCP7 and ran into all sorts of new level problems with customer-supplied PNG files. This format is a lossless compressed format that stores images at high quality with a small file size. Unfortunately, the color syncing system used by the Mac OS looks for various color profile flags to get the levels right, which also affects what happens to such files in QuickTime. In my case, this caused unacceptable gamma shifts in FCP7. Most likely the culprit was either an incorrect color profile or an incorrect “assumption” by QuickTime. The solution for me was Lemke Soft’s Graphic Converter. This app has been a graphics conversion staple that I’ve used for two decades. It used to be bundled by Apple and in fact, the version I’m using under Snow Leopard was still the one I migrated over from a PowerBook G4!
I use Graphic Converter mainly for its batch conversion functions. In this example, I “washed” all the PNG files through Graphic Converter and turned them into uncompressed BMP files. Viola – no gamma shift – and proper levels! I have also run into cases where Photoshop-generated JPEGs had issues on a Mac. Here again Graphic Converter saved the day, by stripping out the Photoshop color profile info and rewriting a clean JPEG file.
3. Gaussian blur on stills
FCP is frequently slammed for the quality of moves on high-res photos using the DVE functions of the motion tab. Typically the offense is that detail in the image tends to scintillate or that diagonal lines look aliased. This really isn’t the fault of FCP, but rather the fact that these stills have a lot more detail and resolution than can be properly displayed when reduced to that size. As far as I know, FCP has no subpixel filtering, so the texture or detail in an image falls on one scan line or another without any smoothing in between. Other apps don’t necessarily create better results, but are actually softening, blurring or filtering the image in ways that look better in a video display format.
The secret for FCP users is to do the same thing. In other words, add blur to the image yourself. This can either be achieved in a graphics program like Photoshop or within FCP by adding a filter. If you use Photoshop, then prep the image by adding a slight Gaussian blur. Experiment with the right setting, but typically a value that looks a bit soft in Photoshop will ultimately look good on a TV screen. If you decide to just add an FCP filter, then use either the Gaussian blur or the Flicker filter. Again, play around with settings to taste, but I find that a Gaussian blur value of between .5 and 2 is generally right.
4. Audio as 48kHz
Repeat after me: Always work with uncompressed audio at a sample rate of 48kHz inside Final Cut. Yes, you can import MP3s and yes, you can work with 44.1kHz audio on the timeline, but in the end, it will bite you. Typically these files seem to be OK if left unrendered – leaving FCP to do all conversions on-the-fly. However, once they are rendered, forcing a sample rate conversion, you run into nastiness. For example, I’ve made music edits and found that the music at the edit points shifted. Or that levels didn’t react properly when I was trying to mix the audio.
When I work with audio in FCP, I will almost always convert the files to 48kHz, 16-bit AIFF files first. This can be done in two easy ways. I use QuickTime Pro (make sure to use QuickTime 7 if you are on Snow Leopard) to convert the files before importing. If you’d rather have this be automatic, simply purchase and install Digital Heaven’s Loader. It runs resident with FCP. Audio files that are imported into FCP by dragging them to the Loader tab will automatically be copied and converted during the import process.
5. Audio files running at the right speed
One of the quirks with FCP is handling audio synchronization. This is mainly a factor in film projects using double-system sound or music videos shot to a playback track. FCP determines speed based on audio samples and not timecode. Unfortunately it’s not as simple as making sure your audio setting matches the project. That’s because FCP apparently does some hidden things under the hood.
For example, I’ve seen issues where a project was started with one frame rate setting, but later changed. I hit this on a PAL music video job. When I imported the final mixed track and matched it against the client’s rough cut, nothing I could do would eliminate the drift between the final mix and the temp track. Even though everything appeared to be correct, imported audio would not line up as anticipated. The only culprit appeared to be the settings used when the project was initially created. Sometimes you have no control over this, because you’ve inherited the project already in progress.
It appears that this is an issue that affects audio files – AIFF and WAVE. One workaround is to convert the audio to QuickTime movies, which forces timecode onto the clip. This isn’t a silver bullet, though, and I have found it to work at times and not at others. If it doesn’t work, you may be forced to alter the audio speed in order to maintain sync. This becomes a big issue if you happen to record double-system sound when shooting with a Canon EOS 5D Mark II at 30fps. You can post in a true 30fps project and maintain sync or you can post at 29.97 and be video-friendly. In the latter case, it’s easy to change the video 30fps frame rate to 29.97fps using Cinema Tools, but the audio will have to be speed-adjusted by .999 to maintain sync.
6. Compressor and interlaced video
Compressor 3.5 (part of the “new” Final Cut Studio) changed the way the Inspector tab works. Now it includes an A/V Attributes tab that will actually “read” the format of the source file. This will also affect what an encoding preset does to that file. If you click on the source file in the Job panel, the Inspector will identify among other things the field order of the file. Unfortunately, it randomly gets it wrong. Files that are top field first are identified as progressive or bottom field first and the other way around.
If you are using Compressor to convert an interlaced camera file or a stock footage file into ProRes for use in FCP, for example, it is important that the field order be correct. If the field order is misidentified, the file will either be improperly converted or the field order will be incorrectly flagged. For example, HD files are supposed to be top field first. If this is identified as bottom first and converted to top, the field order will be swapped and the result on an FCP HD timeline will look wrong.
In other cases, the files won’t necessarily be converted incorrectly, but will only be flagged incorrectly. The file will actually be top field first, but it will show up inside FCP as bottom field first. When this happens a Field Shift filter is automatically applied by FCP onto the clip in the timeline. The file is actually the correct field order, but the filter makes it wrong, which again, yields the wrong results. Unfortunately, you won’t see this issue unless you are monitoring the video though a hardware i/o card or device to a broadcast display or CRT.
When you work in Compressor 3.5, it is important to check each source file for field order. If the file is misidentified, you must change the setting manually in the Inspector pane. This will generally fix the problem. In addition, check the Field Dominance column in the FCP Browser. If a file is incorrectly identified, this can be changed in that column. Once it is properly set, this will prevent a Field Shift filter from being automatically applied when it isn’t needed.
7. Better slomos
FCP7 has improved the variable speed workflow, but not the actual quality of the video. Video fields and/or frames are blended, which doesn’t result in the smoothest rendering of motion. This is especially true of “24p-over-30i” content. Other NLEs, like Avid Media Composer, employ more advanced technology, offering several options for motion rendering. The best looking Avid choice is FluidMotion Timewarp. Like other retiming technologies, this creates new “in-between” frames derived from actual surrounding frames when motion is slowed or sped up. A similar quality improvement can be achieved if you opt to use either Compressor or Motion instead of FCP.
In Compressor, apply a conversion preset and use Frame Controls to achieve a speed change. Select Best for Rate Conversion. Click Duration and set a percentage or a time value consistent with the speed you are trying to achieve. Inverse telecine can also be applied, but remember that this method only works for constant speeds and not speed ramps.
Motion is your other option. Speed changes can be achieved by either changing Properties for the clip or adding a Retiming behavior. To alter the image quality, reveal the Timing controls in the Properties tab. Set the speed and then select the type of Frame Blending in the drop-down menu. Optical flow yields the best result, although it will also on occasion introduce unwanted motion artifacts.
8. Controlling filter selections in Final Cut Pro
Ever wonder why some filter choices are listed twice in the FCP effects palette? That’s usually because you are seeing both the native FCP and the native Motion filters in the same pulldown menu. This is controlled by the Effects / Effects Availability setting. For example, when you select All Effects, you’ll see two Gaussian Blur and two Zoom Blur filters in the Blur category. Look in the Effects Class column of the FCP Browser and you’ll see that one of these is listed as an FxPlug filter. That’s the native Motion version of this filter. Change the setting to Only Recommended Effects and the FxPlug version of those two disappears, leaving you with only the native Final Cut version (not FxPlug). In most cases, either version can be used without issue.
Lastly, if you’ve installed a lot of plug-ins and would like to reduce the clutter of effects palette, then use the selection, Only My Preferred Effects. When this is selected, filters will only show up if you have placed a check in the Preferred column of the Browser.
9. Preparing stills for Final Cut
Final Cut Pro is resolution independent, but this doesn’t mean “resolution infinite”! Working with large, high-resolution files is almost always an issue when incorporating digital still photos into an FCP timeline. Although you can throw a lot at the software, some things work better than others. The exact point that you’ll choke FCP depends on factors like the software version, processing power, amount of RAM and the installed graphics card.
Final Cut Pro performs best with stills that are RGB-mode JPEG, TIFF or BMP files under 4,000 x 4,000 pixels. Many print-resolution files or digital still photos from high-megapixel cameras like the Canon 5D will exceed these sizes. Merely clicking on a 6,000 x 6,000 pixel TIFF file that’s CMYK and not RGB in the FCP Browser will often crash the app. Poof! Straight to Jail – do not pass GO, do not collect $200! My recommendation is always to review and prepare your graphics and stills prior to using them inside FCP.
The best applications for organizing, preparing and adjusting stills include Photoshop, Graphic Converter, Aperture, iPhoto and Lightroom. Files should always be 8-bit/channel RGB. Remove any alpha channels unless you intend to use the file for keying. Some folks will argue that FCP does better with uncompressed files, like TIFFs, as this reduces the decoding overhead of other files. That may be true, but I’ve always used JPEGs with good results. I’ll export the files in the JPEG format at the highest quality setting (12 in Photoshop) and reduce the frame size if it’s an extremely large file. Typically I’ll resize the frame to a maximum horizontal or vertical dimension of 2,500 pixels for SD and 3,500 for HD. This will usually give me plenty of space for a camera-style move on the image and still stay below 100% of the actual frame size.
10. Changing alpha setting
An ongoing dilemma with NLEs is how alpha channels are treated. Once imported into FCP you have the option of changing the alpha setting – None/Ignore, Straight, Black or White. The default is typically fine, but in some cases, a glow or a soft shadow will not have the proper transparency. For example, a white glow against a white background may result in a darker halo at the edge of the glow instead of seamlessly blending into the background. This can be corrected in FCP by opening the Format setting for that clip and changing the alpha setting to one of the other options. Change the master clip in the FCP Browser so it’s correct every time it is used, or correct it only in the timeline if it just needs to be fixed for one instance.
11. Positioning text and graphics
When you move a graphic or line of generated text using the FCP motion tab’s positioning controls, the result will often be soft or blurry. I believe this is related to FCP’s lack of subpixel filtering. Two rules I generally apply in these situations. First, whenever a generator (like Boris 3D or the Text generator) has a Control tab, I use the X,Y positioner in that tab and not the motion tab to reposition the text. This aligns the graphic without adding an additional modifying layer.
Second, any time I change scale or position parameters, I try to stay with even and whole number values. In order words, a scale value of 32, not 33; or a Y position value of 202, not 201.33. Try this for yourself and I think you’ll quickly see that the quality of anything with definite lines and detail, like text or a logo, will look much sharper at these even values. On rare occasions the opposite will be true. For example, it might look crisper if you nudge the text up to an odd number setting. In either case, experiment and see what looks best to you. Remember to only judge this when your Canvas is at a view setting of exactly 100%.
© 2009 Oliver Peters