vReveal Improves the Image Quality of Your Video
We've all captured precious, one-of-a-kind moments on video, only to find that the recorded image didn't look that great. Perhaps the video was too shaky, dark, blurry, or grainy. Or perhaps the device used to capture the video wasn't ideally suited for capturing video, such as you might see with video captured by a digital camera or mobile phone. But what if you had an inexpensive application that could rescue your video quality and make it look significantly better--much like the expensive video enhancement tools available to video professionals? And what if this software could harness the additional processing power of a GPU you might already have installed in your system in order to perform up to five times as faster than powered by a CPU alone?
As of today, such a tool is now available to consumers in the form of MotionDSP's vReveal video enhancement software for Windows. The technology behind vReveal is actually based on MotionDSP's video-enhancement, Ikena forensic software that is used by law enforcement and intelligence agencies. If you watch NCIS, CSI, or any one of the many forensic investigation TV shows, then you have a pretty good idea of what the technology is capable of--mind you, you might not be able to eke out complete license plate numbers from a seemingly obscured image, but it will be able to draw out more detail from videos that are too dark or washed out. The software uses what MotionDSP calls "super-resolution algorithms":
"Super-resolution is based on multi-frame analysis, which has the unique ability to increase resolution and to provide cleaner, better detail. As you know, a video is a series of many, rapid snapshots of a scene. Multiple neighboring frames of video each provide a slightly different view of the subjects in the scene, collectively providing a rich amount of information about scene details. Our powerful super-resolution algorithms analyze and combine the information from multiple frames of video to produce output frames with the best possible details. In addition, sophisticated motion-tracking across multiple frames enables very effective stabilization."
By analyzing multiple related frames, the software is able to interpolate missing information and re-apply that missing information back into the image. vReveal can potentially "stabilize, brighten, and sharpen problematic videos and is uniquely able to increase resolution and to remove noise artifacts such as graininess and pixelation." Of course, not every video that is enhanced will see the same level of treatment. MotionDSP advises that the software works best on original videos or videos that have not gone through too much re-encoding (a file that has gone through several format conversations would be a poor candidate). Also, the maximum vertical resolution video that vReveal's video enhancement features currently support is 576-pixels (PAL resolution); you can still import HD videos and use vReveal to at least rotate and trim them, but the actual super-resolution enhancements currently only work on SD video. MotionDSP has stated, however, that it does plan on releasing a higher-end version of vReveal for enhancing HD video.
Users who already have CUDA-enabled Nvidia GeForce GPUs in their systems will benefit from vReveal's built-in CUDA support, which takes advantage of the parallel-processing capabilities of the GPU. Last September, Nvidia and MotionDSP announced a strategic partnership to develop a video-enhancement application codenamed "Carmel;" vReveal is the fruit of their labors. MotionDSP claims that vReveal can run up to five-times faster on a CUDA-enabled system than on systems where the software is powered by the CPU alone. We ran a number of tests on a 3.2GHz Intel Core i7 965-based system with an Nvidia GeForce GTX 280 GPU. With CUDA-acceleration enabled, we saw about a 40 to 50-percent performance improvement over when the software was only using the CPU. We were also impressed with the software's ability to live up to its advertising--it did a rather good job enhancing the quality of some less-than-ideal-looking videos.
Enhancing a video can be as simple as pressing the "One Click Fix" button--in fact, you can even "fix" multiple videos at once. Users who want more control can use the software's "Fine Tuning" features. While faster systems with CUDA-enabled GPUs will certainly see speedier performance, vReveal actually has rather modest minimum system requirements: a 1.6GHz Intel or AMD CPU, 512MB RAM, Windows XP (32- or 64-bit) or Windows Vista (32- or 64-bit). vReveal will work fine without a CUDA-enabled GPU--the end results will be the same--CUDA only noticeably speeds up the process.
You can download a (14MB) free trial version of vReveal from MotionDSP's site here or from Nvidia's nZone online store here. The trial is fully functional for 30 days, but it places a watermark on all exported videos. After 30 days you can no longer export videos. You can also purchase vReveal from Nvidia's nZone Online store for $49.99.
--Additional contributions to the article by Marco Chiappetta
 |
| Credit: MotionDSP |
As of today, such a tool is now available to consumers in the form of MotionDSP's vReveal video enhancement software for Windows. The technology behind vReveal is actually based on MotionDSP's video-enhancement, Ikena forensic software that is used by law enforcement and intelligence agencies. If you watch NCIS, CSI, or any one of the many forensic investigation TV shows, then you have a pretty good idea of what the technology is capable of--mind you, you might not be able to eke out complete license plate numbers from a seemingly obscured image, but it will be able to draw out more detail from videos that are too dark or washed out. The software uses what MotionDSP calls "super-resolution algorithms":
"Super-resolution is based on multi-frame analysis, which has the unique ability to increase resolution and to provide cleaner, better detail. As you know, a video is a series of many, rapid snapshots of a scene. Multiple neighboring frames of video each provide a slightly different view of the subjects in the scene, collectively providing a rich amount of information about scene details. Our powerful super-resolution algorithms analyze and combine the information from multiple frames of video to produce output frames with the best possible details. In addition, sophisticated motion-tracking across multiple frames enables very effective stabilization."
 |
| Credit: MotionDSP |
By analyzing multiple related frames, the software is able to interpolate missing information and re-apply that missing information back into the image. vReveal can potentially "stabilize, brighten, and sharpen problematic videos and is uniquely able to increase resolution and to remove noise artifacts such as graininess and pixelation." Of course, not every video that is enhanced will see the same level of treatment. MotionDSP advises that the software works best on original videos or videos that have not gone through too much re-encoding (a file that has gone through several format conversations would be a poor candidate). Also, the maximum vertical resolution video that vReveal's video enhancement features currently support is 576-pixels (PAL resolution); you can still import HD videos and use vReveal to at least rotate and trim them, but the actual super-resolution enhancements currently only work on SD video. MotionDSP has stated, however, that it does plan on releasing a higher-end version of vReveal for enhancing HD video.
|
|
| Original video |
Users who already have CUDA-enabled Nvidia GeForce GPUs in their systems will benefit from vReveal's built-in CUDA support, which takes advantage of the parallel-processing capabilities of the GPU. Last September, Nvidia and MotionDSP announced a strategic partnership to develop a video-enhancement application codenamed "Carmel;" vReveal is the fruit of their labors. MotionDSP claims that vReveal can run up to five-times faster on a CUDA-enabled system than on systems where the software is powered by the CPU alone. We ran a number of tests on a 3.2GHz Intel Core i7 965-based system with an Nvidia GeForce GTX 280 GPU. With CUDA-acceleration enabled, we saw about a 40 to 50-percent performance improvement over when the software was only using the CPU. We were also impressed with the software's ability to live up to its advertising--it did a rather good job enhancing the quality of some less-than-ideal-looking videos.
|
|
| Enhanced video |
Enhancing a video can be as simple as pressing the "One Click Fix" button--in fact, you can even "fix" multiple videos at once. Users who want more control can use the software's "Fine Tuning" features. While faster systems with CUDA-enabled GPUs will certainly see speedier performance, vReveal actually has rather modest minimum system requirements: a 1.6GHz Intel or AMD CPU, 512MB RAM, Windows XP (32- or 64-bit) or Windows Vista (32- or 64-bit). vReveal will work fine without a CUDA-enabled GPU--the end results will be the same--CUDA only noticeably speeds up the process.
You can download a (14MB) free trial version of vReveal from MotionDSP's site here or from Nvidia's nZone online store here. The trial is fully functional for 30 days, but it places a watermark on all exported videos. After 30 days you can no longer export videos. You can also purchase vReveal from Nvidia's nZone Online store for $49.99.
--Additional contributions to the article by Marco Chiappetta
