|Digital Compositing for NTSC Broadcast Segments with Sonic Foundry Vegas Video 3.0|
Los Angeles California USA
CreativeCOW.net. All rights reserved.
Late last year, Bryan Wilhite had the opportunity to do some digital compositing work on a feature-length documentary about the life of the great theatre, television and screen actress Beah Richards. Vegas Video was the indespensable tool for his role in the production. In this article, Bryan shares some tips that he learned.
As the company name suggests, Sonic Foundry is known for their "sound" products, namely their well-respected Sound Forge digital audio editors and their innovative ACID digital audio sequencers. My personal investment in Sonic Foundry products rightfully started with these professional sound design tools. When I first heard of Vegas Video, I immediately thought of what the word "Vegas" meant to me and quickly associated my feelings about this word with the product and overlooked the software for several months.
Then R/Kain Blaze, a Mac-based friend of mine, began pointing me to articles--like the one by Charlie White (digialvideoediting.com)--showing an unusual amount of respect for this Wintel-based product. When Charlie writes, "And, acing out Final Cut Pro 3 is Vegas Video 3's ability to see what you're really doing, previewing all your work through 1394 on an NTSC or PAL monitor in real time, something that's not yet possible with Final Cut Pro 3 (more on that later) without hardware assistance..." I began to sit up and take notice and eventually invested in Vegas Video.
I've recently had the opportunity to do digital compositing work on a feature-length documentary about the life of the great theatre, television and screen actress Beah Richards. Vegas Video was the indispensable tool for my role in this production.
Many professional filmmakers are discovering Sonic Foundry products and solutions. For example, the recent Sony Classics documentary film "Dogtown and Z Boys" depended heavily on Sonic Foundry frame-rate conversion and digital image transfer technologies. However, I found myself, like many nascent Vegas Video users, contributing to a project already in production using more "mature" technology. In this case, I was supporting a Windows NT, non-linear editor--hard-wired to a Targa 2000 video card--that appeared to be related to Discreet Logic's edit*, descending from the legendary Intergraph TDZ rigs from the late 1990s.
First challenge. The editor, Kate Amend, and the director of this project gave me written instructions (called "shot lists") detailing the duration of the shot, identifying the media assets used in the shot, specifying transitions among these assets and inquiring about creating variations of a particular shot. We will see that Vegas video is able to translate these written words into its user interface, making a "living record" of their instructions saved in Vegas Project files (with the extension VEG).
Second challenge. The media assets required in the shots are mostly digitized photographs or scanned pages from books and magazines. This material comes in a variety of formats, like JPEG, TIFF and even GIF. All of this source material will be consolidated under one, still-image file format and issues of pixel aspect ratio--moving from the computer screen to the television screen (discussed below)--will be adequately addressed.
Third challenge. The Targa 2000 video card demands that all footage in the online system be rendered in its proprietary codec. Vegas Video with the help of the French company Morgan Multimedia will make meeting this challenge easy going. Additionally, this production has been focused on domestic broadcast television under the NTSC standard here in the United States of America. This implies that all of my contributions to this project must comply with this standard. As we will see, this will not be a serious obstacle.
Sonic Foundry provides great introductory tutorials about the Vegas Video interface (tutorials about previous versions are still relevant), so with that in mind let's drill right down to the essentials for setting up a shot for NTSC broadcast. The following scenario is identical to what has happened while making the Beah Richards documentary: we have an item on our shot list specifying that we need a five-and-a-half second push in on an old portrait of a man in a suit. We know that every shot should have a one-second leader and trailer in addition to the specified duration. We also know the editor and director are looking for at least two variations of the shot.
The first step (after opening Vegas) is to edit the Project Properties--specifically the properties of the Video and Summary tabs. The Video tab is shown below. The Width:, Height:, Field order: and Pixel aspect ratio: properties have a direct effect on the Video Preview window:
The dialog shows that the built-in template (or Preset),
NTSC Standard (720x486, 29.970 fps), has its Field order: property modified to
None (progressive scan). I am doing this because I will be previewing shots on an LCD screen instead of an external analog monitor as I'll be compositing this shot with my Dell Inspiron 8100, a portable computer. (Recall, by the way, that progressive scanning refers to the top-to-bottom, left-to-right, line-by-line path of the electron gun of CRT computer monitors--opposed to the every-other-line scanning of interlaced television displays.) I found editing with a portable computer more convenient for the editor and director as I could bring my work to their facilities, having a near-zero impact on the established workflow.
Clicking on the Apply button will change the Video Preview window. When undocked, it should look like picture below:
The dimensions of the Project (shown on the bottom of the Video Preview window) and those of the Preview are different. I have deliberately resized the Video Preview window to roughly half the size of the actual rendered media making previewing faster. Please make note of the buttons at the top of the Video Preview window--namely the Video Output FX... and Overlays buttons. These will come in handy later on (so understanding what they do will be a big help).
Returning to Project Properties we quickly note that this shot will not need any audio (this is to be added later by other team members) so we don't need to see the Audio tab. The Project Properties Ruler tab is set to
Time & Frames by default--we'll go with this format and move on to the Summary tab:
The Vegas Video manual states that Project Properties are mostly for "informational purposes." We can see that the Summary tab stores most such information. The Comments: field saves the instructions from the editor and director directly in the file for future reference.
Based on the comments stored in the project, we know we need a shot with a five-and-a-half second duration (this is about 00:00:05.15 seconds and frames, based on 30 frames/second)--plus one second leading and one second trailing. By entering
Ctrl+g or by double-clicking on the Cursor Position field at the lower right of the Timeline, we can make our first move to the one-second position on the Timeline and insert a Marker. The animations below show this two-part process:
Our second and third cursor position markers denote the actual shot length and the one-second trailer respectively. The picture below summarizes the duration specified in the shot list:
Note that the second marker specifying shot duration is actually at position 00:00:06.15 seconds and frames on the timeline. It is possible to use the Set Time at Cursor command while the cursor is at zero and enter negative one second (-00:00:01.00) to offset the timeline "correctly" but this not necessary for our shot.
Researchers for documentary projects often receive source material from people who have very little interest in professionally archiving photographs. In our scenario, we will use the sample image from the Adobe Photoshop 7.0 distribution called
Old Image.jpg (shown at left, at half the size of the image actually used).
This image is a perfect example of how documentary source material is not exactly in ideal condition. Additionally, this image is a portrait (more tall than wide), while our final NTSC format is landscape (more wide than tall). We will need to use the Vegas Event Pan/Crop tool to pan and scan this image (the Vegas Video online help does a great job explaining how this tool works).
The Vegas Video documentation showed a marked preference for the relatively new Portable Network Graphics (PNG) format. This is definitely the wise choice if alpha-channel transparency is desired in the source images for advanced compositing. However, I don't have much call for 16-bit alpha channels--or even lossless compression. My need for Photoshop's Adjustment Layers outweighs any further consideration of image formats other than its native format (type PSD). Fortunately, Vegas Video is able to detect and import Photoshop files (but Vegas does not, as yet, recognize Photoshop layers in ways similar to Adobe After Effects).
With Photoshop's Adjustment Layers I can apply Curves to the source image to adjust the brightness and contrast--nondestructively. By taking advantage of nondestructive editing of the source image I can make changes on the fly if the shot looks too light or too dark on the monitors of the editor's NLE system. Simultaneously, some "destructive" editing can be done in Photoshop at my discretion: the Rubber Stamp Tool and/or the Healing Brush can be used to reduce the dust and scratches on the original digitized image.
After a lively combination of nondestructive and destructive editing techniques--and with my trusty Wacom tablet and stylus--we can compare the original source image to the final Photoshop still:
The Vegas Video Explorer has our Photoshop file,
Old Image.psd, selected. We can see that the Photoshop, PSD file type is registered as IrfanView (for "obvious" reasons for the Wintel-based lovers of IrfanView). Pay attention to the file sizes: the
Old Image.jpg file has ballooned to 1.42 MB after adding a layer and being converted to Photoshop native format. I am willing to pay the price of cheap disk space for the sake of convenience!
By clicking and dragging the file to the Timeline window, we add this file to our current project creating a Video Event. Let's snap our event in place to the third marker by dragging the right edge of the event. With the cursor over the event, we can see that the image looks stretched horizontally in the Video Preview window. Vegas Video imports still images using the aspect ratio of 1.0, the aspect ratio of the source media instead of the aspect ratio of the project. The image appears stretched wide on the progressive scan computer screen, but would look right on an external NTSC monitor. To compensate for this, right-click on the Video Preview Window and select Display Square Pixels (thanks goes to Software Design Engineer, Dennis Adams of Sonic Foundry for pointing this out to me):
By snapping the event to the third marker (at 00:00:07.15), we have a Video Event on the Timeline like the following:
Notice that the Event track has a Track Name, "Version 1." I entered this name knowing that we can use the Mute button to hide or show this track. Later we will see how we can use these features to store multiple versions of the same shot in one project file.
I have decided that track
Version 1 will push in on the portrait in full frame. This means using the Event Pan/Crop... tool which is available by clicking Event Pan/Crop... button directly on the Video Event, right-clicking the event and selecting Video Event Pan/Crop... from the context menu or by selecting the event and choosing Tools > Video > Video Event Pan/Crop.... I prefer setting the cursor to the "home" position on the Timeline (usually at the zero-second mark) and clicking on the Event Pan/Crop... button. This prevents the creation of unnecessary keyframes in the Event Pan/Crop window.
Making sure that the Sync Cursor toggle button (at the lower left) is down in the Event Pan/Crop window, press
Alt+0 and use the
Ctrl and left/right arrow keys to move to the first marker. If Sync Cursor is active then the position of the cursor in the Event Pan/Crop window will also move to the location of the first marker. Drag the default keyframe to this location. We should now have one keyframe placed in the Keyframe Controller (at the bottom of the Event Pan/Crop window) at position 00:00: 01.00 like the following:
Right-click on the image in the Event Pan/Crop window and select Match Output Aspect. This command crops the image according to the frame width and height specified in Project Properties:
The Video Preview window with the Safe Areas overlay activated shows what will appear on an NTSC-compatible monitor. Recall that the inner dashed rectangle is the title safe area while the outer rectangle is the action safe area:
I'm pretty confident that this is not the shot the editor and director are looking for! We'll need to work harder! Eventually two keyframes are set. The animation below, jumping between the first and second keyframes, summarizes:
The rate of tweening between the keyframes is Linear by default. To simulate more "natural" camera movement the interpolation of the first keyframe is set to Smooth. Interpolation is set by right-clicking on a keyframe and selecting an option from the context menu. The table below (from the Vegas Video manual) summarizes the options:
With the Pan/Crop keyframes set, we can see the corresponding frames in the Video Preview window:
Many a web programmer was lectured about "web-safe" colors. We were encouraged to confine our indexed-color palette within a range that's compatible with most browsers on most computer operating systems. A similar principle is behind "broadcast-safe" colors for NTSC television. On a computer-generated color scale of 0 to 255, where black is 0 and 255 is white, "NTSC black" would be 16 and "NTSC white" would be 235. In fact, computer monitors can generate up to 16 million colors while NTSC television displays about two million.
It follows that footage generated by computer may have literally millions of colors that are "out of gamut" for broadcast television and should be filtered out before the final render. Vegas Video 3.0 ships with a plug-in called Sonic Foundry Broadcast Colors (or Broadcast Colors Filter). From the Plug-Ins window, I drag and drop this filter on the Video Output FX... button and take the most conservative route by selecting the
Maximum Bounds Preset:
By activating the Histogram -- Luminance overlay in the Video Preview window, we can alternately enable and disable the Broadcast Colors Filter producing an effect like what is seen in the animation below:
Clearly we are "chopping off" the edges of the gamut of colors. We won't see the effects of this unless we are looking through an NTSC monitor (which, I remind you, is possible in Vegas).
It is temping to Render track
Version 1 to see exactly what's going on. Yes, we can build a Dynamic RAM Preview or simply press Play to get something to look at in the Video Preview window but I have done hundreds of shots like this one so I'm confident enough to set up some "version 2" tracks.
The version 1 track is very tight on the face--which is great for dramatic effect--but we lose the suit. We can't really tell if this guy is really duded up or not! To up the ante, I want to see the face and the suit at the same time (no panning and scanning--only pushing). One way to achieve this effect is by taking advantage of Vegas Video's Track Motion tool (among other great features). So let's duplicate our first track by right-clicking its item in the Track List (shown at left) and name it
Version 2. We also need to disable the first track by pressing its Mute button in the Track List. The picture below summarizes:
In turn, duplicate track
Version 2. Vegas Video automatically names this track
Copy of Version 2. Rename this track to
Version 2 (Track Motion) and open its Event Pan/Crop window. Remove its second keyframe at 00:00:07.15. At the first keyframe, right-click on the image in the Event Pan/Crop window and select Restore. The Event Pan/Crop window for track
Version 2 (Track Motion) should look like the following:
Close this window. Drag the left edge of track
Version 2 to somewhere around 00:00:03.20 seconds and frames. Dragging the edges of this event will cause the Event Pan/Crop keyframes to change position so make sure that the first keyframe is positioned at 00:00:00.00 (in the Event Pan/Crop window Keyframe Controller--not the project Timeline) and the second keyframe rests at the third marker. The Event Pan/Crop window for track
Version 2 should look like the following:
The last thing we need to do for track
Version 2 is to set a fade-in using the default Fade Type. This is very easy to do in Vegas Video (by dragging the little blue triangle in upper left corner of the Event on the Timeline) and this feature really sets the product apart from its competitors. Finally, view the Text/Backgrounds window (type
Alt+9), drag and drop the Gray solid color to an empty area at the bottom of the Timeline creating a new track. Name it
Version 2 (Solid). Now there should be a total of four tracks, where three tracks will be used in the second version of the shot:
Now we are ready to add Track Motion to the third track. Let's move to the first marker in the Timeline and click on the Track Motion... button for the track
Version 2 (Track Motion). The Keyframe Controller for Track Motion has three levels for Position, Shadow and Glow. Check the Shadow level in the Keyframe Controller and move the Position keyframe to 00:00:01.00. Making sure that Position level has focus in the Keyframe Controller, rotate the frame about negative five degrees (
-5.0) and resize such that the image fits in the title safe area of the Video Preview window. We should have a Track Motion window like the picture below:
The effect in the Video Preview window is shown below:
Let's add one more Track Motion position keyframe at the second marker in the Timeline where we resize the frame much larger than the dimensions at the first keyframe. This creates the effect of pushing in on the subject. This is the Track Motion window:
We have completed the second version our shot. It's far more complicated than the first version but here we are. I hope we didn't lose anyone. Now we are ready to render!
The Targa 2000 line is being discontinued by Pinnacle Systems and is widely considered venerable "legacy" technology. This certainly should have been a showstopper for me--literally. But a key member of the Beah production team, Bill Russell (who has, by the way, recently finalized production of his labor of love The Lost Skeleton of Cadavra), led me in the right direction and I found myself at morgan-multimedia.com, purchasing their LSI M-JPEG Codec which allowed me to software-emulate the rendering functionality of the Targa 1000/2000 series hardware. The image below shows the properties/settings of the Morgan LSI Codec. This dialog is available from the Windows 2000 Video Codecs Properties dialog (discussed below) or from a convenient System Tray icon installed by the Morgan product itself.
After installing the Morgan LSI codec, I verified that my copy of Windows 2000 Professional recognized this new codec. In the Windows 2000 Device Manager, I selected Sound, video and game controllers > Video Codecs and I saw the dialog below:
And, of course, we can see the properties/settings of the Morgan LSI Codec by right-clicking on Morgan Multimedia LSI M-JPEG Codec (shown in the list above) and selecting Properties > General > Settings.... Seeing this, I realized that Vegas Video (with the help of the Morgan Multimedia Codec) is effectively a video hardware emulator! Without this solution (or some other software-based solution) I would have had to pass on the project or find a used Targa 2000 card and build an entire system around it.
Press the Solo button for track
Version 1 and select File > Render As.... The dialog should be set like the settings shown:
Let's review my Template for Targa 2000 emulation. Click on the Custom button and move to the Video tab:
Bill Russell was a great help "reverse engineering" these settings. He discovered that disabling the Interleave every (seconds): and Create an OpenDML (AVI version 2.0) compatible file fields and setting Data rate (Kbytes/second):
1000, was most compatible with the Targa2000. Disabling the Audio tab helped as well. Of course, the Project tab will show Video rendering quality:
Best (which is the appropriate setting for down-sampling still images) and Motion blur type:
None. Clicking OK in the Custom Settings dialog takes us back to the Render As dialog. Press the Save button and
VegasNTSC (Version 1).avi will be written to disk.
Press the Mute button for track
Version 1 and select File > Render As.... Enter
VegasNTSC (Version 2) for the file name and press Save. That's it.
The techniques discussed in this article are direct descendants of the inspiration given to me by the classic American hymn The Civil War by Ken Burns. Almost all of The Civil War is camera moves over still images. Now with Vegas Video we don't need a several-thousand-dollar camera and an extremely steady hand. The "special effects" we have explored can adequately replicate what a "real" camera can do, and "go beyond."
|VegasNTSC.zip||Vegas Video project files and Media Pool assets.|
|VegasNTSC(Version1).mov||Version one of the shot. This file requires Apple QuickTime.|
|VegasNTSC(Version2).mov||Version two of the shot. This file requires Apple QuickTime.|
Visit the forums at Creativecow.net if you found this page from a direct link.