Content guidelines apply to producing or assessing as well as acquiring and prepping video content for FrameGlide systems. Please also see our Custom Content page for general information on working with us on FrameGlide video production.
- If You Plan To Do Your Own Filming
- Pixel Resolution
- Square Pixels
- Color Depth
- Separate Movies
- File Wrapper & Codec
- Still Frames
- Photo JPEG / MJPEG Compression / Data Rate
- Progressive Scan
- Test Movie
- Head-To-Tail Transition
- Avoid Repeated Fades to Black
- Time, Frames & Perceptions
- Progress Indicator
- Attract Mode
- Audio Production
- Additional Resources
If You Plan To Do Your Own Filming
There are considerations, unique to the FrameGlide software with the Spin Browser dial environment , that are quite different from broadcast. For example, it is desirable to shoot very long and uncut sequences (think from minutes to hours) and to set short exposure time to eliminate any within-frame motion blur. Note that we rent out live capture systems as one means of acquiring video content for what will eventually be pre-produced installations.
In terms of processing power, the FrameGlide application running on a 3GHz quad-core PC can readily handle up to two 1920x1080 (1080p) images or up to four 1280x720 (720p) images. It is typically best to match the FrameGlide application graphics resolution to the native pixel resolution of your screen. Video production typically also matches this same resolution, unless there is the requirement for reduced processing power and/or disk storage space, side-by-side live capture and pre-produced videos, etc., in which case smaller-than-screen resolutions may be desirable.
The pixel aspect ratio must be square.
Note that this is not the native format for either Mini DV or HDV, so you must change it in post-production. For example, HDV records at 1440x1080 with rectangular pixels. To play back the movie on a square-pixel 1920x1080 monitor, the content must be stretched to account for the greater horizontal resolution. In other words, export a 1440x1080 rectangular pixel clip at 1920x1080 square pixels.
The file needs to be 24 bit color, RGB (CMYK normally is not even an option).
Note: Because we require JPEG compression for the FrameGlide | SpinBrowser environment, a 32-bit color is not an option because the JPEG format does not support an alpha channel.
For each section that might possibly benefit from different dial speed parameters (for example adjacent scenes of vastly different lengths and transitions versus main content, etc.), export the section as a separate movie. This is because the FrameGlide software's ability to set unique parameters is on a file-by-file basis. We also recommend that no video segment be larger than 10-30 Gigs in order to avoid the situation where fixing one small error requires re-rendering the entire production. Also, if you want to put the content on a Fat32 Flash drive, then each segment should not exceed 4 Gigs. Note that once imported into the FrameGlide environment, all the separate movies will appear as one large seamless experience to the user.
File Wrapper and Codec
Wrapper: Quicktime .mov file.
Codec: PhotoJPEG (or MJPEG, although this is deprecated.)
If you would like to include static images within your video production (which differ from Frameglide runtime-applied static bitmap overlays), note that on rendering, you must uncheck any "optimize stills" option. This will ensure that identical frames are not skipped at render time. (Optimization is appropriate for linear, constant velocity playback, but unsuited to the Spin Browser environment, where the time between frames is never predictable.)
Photo JPEG / MJPEG Compression / Data Rate
Compress as much as you can without noticeable visual degradation. What this means in practice is that you should compress until you notice some difference and then back off a bit until you can't. You rarely need to go over about 75% quality (at least in the Adobe CS suite - the "% value" has no industry standard and varies significantly between algorithms.).
We are sometimes asked for a target data rate for pre-produced media. However, virtually all the video editors with which we are familiar provide a "percentage" setting, which dynamically adjusts the compression based on image complexity. We therefore recommend using the procedure above, rather than the overall/average data rate.
(And while you should typically not need to worry about this, if interested, check out our post on color space subsampling: Photo-JPEG / MJPEG Video Compression.)
Movies need to be either shot in progressive scan mode (which is ideal provided the frame rate is high enough) or deinterlaced properly after the fact and exported in progressive/single-field mode.
(Deinterlacing refers to recombining every two sequential frames - each composed of either the odd or the even horizontal scan lines - into one complete image. Improper deinterlacing can leave horizontal "comb-like" tear line artifacts in the final product, which dramatically reduces the image quality. There are several programs that offer deinterlacing functionality, such as Adobe After Effects, Adobe Premier, etc.
Most consumer-grade cameras only allow progressive scan at relatively low frame rates that are unsuitable for use in the FrameGlide | SpinBrowser environment. And note that even if video is shot in progressive scan mode, it is possible to accidentally convert it to an interlaced form.)
Unless you are prototyping on a local FrameGlide system, we strongly recommend emailing us a short clip (for example a 4 frame test movie), so we can confirm proper format, before you render the entire production.
With fixed content, the FrameGlide | Spin Browser system is typically (though not always) configured to loop seamlessly "head-to-tail". In that case, remember to add whatever you want to happen between the end and start of the video.
In other words, you might want to cross-dissolve from the end frame to the start frame, or fade to black, or to an intro/credit screen (this is separate from any potential Attract Mode movie),... or you might just want a simple head-tail connection with no transition at all. .
Avoid Repeated Fades To Black
Note that you typically will NOT want to fade to black between each little segment, but rather do a cross-dissolve or some other fun transition. The reason for this is that if you fade to black, and the user is scrolling fast through the content, it will flash light/dark/light/dark in a very distracting strobe-like manner.
Time, Frames, & Perceptions
A) Remember during design and production that the visitor is just as likely to move backwards through the video as forwards.
B) In terms of the length of transitions between segments, it is better to think about scene change per frame (which you want to be very small) rather than the duration in time (which is visitor-controlled). If you don't render the transitions as separate movies which can be individually speed-adjusted, then you need to think about both motion per frame and duration relative to the adjoining clips.
C) Panning is difficult to implement in a pleasing manner due to the human perceptual system more than anything else. If you do need to pan, often combining panning with zooming or placing the panning window within a visually-stationary framework are good solutions. Definitely test within the FrameGlide | Spin Browser environment with actual dial before going too far.
It is generally a good idea for the FrameGlide content to include a progress indicator that provides the visitor feedback as to where they are, what is ahead and what is behind. When video content is only a few minutes long, this may not be needed, but when the content is many hours in length (as is the case with many installations) the progress indicator is very helpful.
Sometimes this graphic element is simply a moving pointer against a semi-transparent background, sometimes it has content designators that identify a current section (for example, "Animals", "Weather", etc.), and sometimes both. See diverse examples from previous projects.
The FrameGlide | Spin Browser system has the ability to dynamically overlay such graphics at runtime (as well as enable/disable a text time-code overlay with the "T" key). If you want the overlay behavior to be content dwell-time-dependent (e.g. a legend that pops up only when you first enter a region, only if you have stayed more than 10 seconds, etc.), then you must use our software's dynamic capability (including alpha channel transparency). Otherwise, for fixed content systems it is often better to add the progress overlays into the video itself during production.
Advantages of Embedding Progress Indicator Overlays Within Video
- You will have more flexibility. For example, you might want the progress indicator to be an elegant photo-realistic analog clock.
- It will reduce the amount of on-the-fly computation needed at runtime. Whether this makes any difference will depend on the details of your installation - such as whether you are trying to drive three HD screens off one PC etc.
Advantages of Dynamically Overlaying Progress Indicators at Run Time
- You may wish to update the video over time, in which case dynamic overlaying will save a great deal of time because you won't have to re-render the entire movie when you add a new segment.
- You may wish to adjust the progress bar's appearance without re-rendering the entire underlying video.
A combination of the two can be appropriate when you want to add designators such as "Fish", "Flowers", etc. within the video itself, but also want a dynamically overlaid moving pointer to automatically adjust to segments being added, rearranged, etc.
The attract mode is what the system does when the application has been idle for some (configurable) amount of time - usually in the neighborhood of 20 seconds. It's intended to catch a visitor’s eye and entice them to play. You can choose one of three options:
- Play a pre-produced QuickTime movie - The content of this movie is completely unrestricted, but it should conform to the standards listed above regarding codec, compression, etc. Upon exiting the attract mode, the system can deposit the visitor anywhere within the content (at the beginning is most common).
- Use a 'temporal path' - You can have the system loop through a fun temporal path through the main content. Typically, this looks as if some "invisible visitor" were using the dial. With this option, when a visitor grabs the dial, he/she gains control of the video at the exact point where the attract mode left off.
- Omit an attract mode entirely, simply showing a static frame. (This frame could be where the previous visitor left off, or the "first" frame of the movie sequence.)
We generally strongly recommend the second option above. The 'temporal path' attract mode is specifically intended not to present conventional uniform playback, but rather to mimic a user’s interactive experience of the exhibit and by doing so:
- Make immediately clear to the visitor that the system is not simply a video loop to be watched. (If the system just "plays" at normal rate, there is a significant risk of visitors thinking it is something to just passively watch.)
- Show visitors exactly how the exhibit enables them to interactively experience cool behaviors and how they can explore by moving backward and forward, fast and slow. In this “ghost user” mode, they are much more apt to immediately approach and start using the exhibit interactively.
Choice of specific content ‘moments’ to highlight and movement creation within those is a bit of an art and an effort - but the result is very worth while!
If audio accompanies the video, it should be saved as a separate .wav file (no sound track in the QuickTime movie itself), uncompressed (integer big endian), at one of the standard sample frequencies in the 32-48Khz range, typically 16 bit stereo.
For additional information see: