Homepage
search
ProductsShopNewsSupportCompany
English
Products

Contact
SheerVideo
PhotoJazz
Synchromy

SheerVideo

Testimonials
Speed
Power
Fidelity
Versatility
FAQ
Test Data

FAQ

Introduction
Uses
Speed
Power
Fidelity
Formats
Availability

Power

Introduction
Other Perfect-Fidelity Codecs
High-Quality Approximating Codecs


BitJazz
SheerVideo FAQ
Power: High-Quality Approximating Codecs

Is SheerVideo as powerful as DV or DV-25?

Is SheerVideo as powerful as DV-50?

Is SheerVideo as powerful as DigiBeta?

Is SheerVideo as powerful as MPEG-2?

Is SheerVideo as powerful as MPEG-4 or DivX?

Is SheerVideo as powerful as Motion JPEG?

Is SheerVideo as powerful as Photo JPEG?

Is SheerVideo as powerful as Sorenson?

SheerVideoIs SheerVideo as powerful as Photo JPEG?

This isn't a fair question, because SheerVideo is a perfect-fidelity ("lossless") codec that compresses the data by packing all of it more efficiently, while Photo JPEG is a DCT-based approximating ("lossy") codec that "compresses" the data by throwing any amount of hopefully less-important information away. So, in principle, Photo JPEG can achieve infinite compression by discarding all the image information.

Nevertheless, for real-world Y'CbCr 8bv 4:2:2 video, the answer is Yes, SheerVideo (2.43) is on average 4% more powerful than Apple's superb implementation of Photo JPEG (2.11) at its Best quality setting. This is a remarkable achievement considering that SheerVideo is also 4 times as fast as Best-quality Photo JPEG, and considering that SheerVideo maintains perfect fidelity, while Best-quality Photo JPEG degrades the image by an average of 0.38 bits per sample.

Similarly, for real-world Y'CbCr 8bv 4:4:4 footage, Yes, SheerVideo (2.84) is on average 15% more powerful than Best-quality Photo JPEG (2.46), despite the fact that SheerVideo maintains perfect fidelity compared to Photo JPEG's average of 0.38 bits of error per sample, and despite the fact that SheerVideo is 4 times as fast as Photo JPEG.

However, for RGB 8b imagery, No, SheerVideo is on average 13% less powerful (2.20) than Best-quality Photo JPEG, although SheerVideo compensates by being 4 times as fast as Photo JPEG and maintaining perfect fidelity, whereas Best-quality Photo JPEG degrades the image by an average of 0.70 b/s.

Note also that, unlike Photo JPEG, SheerVideo supports alpha.

Of course, in situations where image quality is less critical, you can get Photo JPEG to throw away a lot more information, thereby shrinking the video data far more than any perfect-fidelity codec can.

top | next

SheerVideoIs SheerVideo as powerful as Motion JPEG?

This is an unfair question, because SheerVideo is a perfect-fidelity ("lossless") codec, compressing all the data by encoding it more efficiently without losing a single bit of information, while Motion JPEG is a DCT-based approximating ("lossy") codec which "compresses" the data by throwing away an arbitrary amount of (hopefully) less-important information. So depending on how much information you're willing to throw away, you can theoretically get infinite compression with Motion JPEG.

Even so, for real-world Y'CbCr 8bv 4:2:2 video, the answer is Yes, SheerVideo (2.43) is on average 26% more powerful than Best-quality Motion JPEG A and B (1.92), quite an achievement considering that SheerVideo is also 4 times as fast as Best-quality Motion JPEG, and considering that SheerVideo maintains perfect fidelity, while Best-quality Motion JPEG degrades the image by an average of 0.38 bits per sample.

However, for Y'CbCr 8bv 4:4:4 video, the answer is No, SheerVideo is an average of about 1% less powerful than Best-quality M-JPEG A (2.88) and M-JPEG B (2.89). And for RGB 8b, SheerVideo averages 25% less powerful than Best-quality M-JPEG A (2.93) and M-JPEG B (2.94). However, SheerVideo more than makes up for its lower compression power by being 3 times as fast as Motion JPEG for these formats, and by maintaining perfect fidelity, in contrast to Motion JPEG's mean error rate of 0.80 bits per sample for Y'CbCr 8bv 4:4:4, and 1.21 b/s for RGB 8b.

And with SheerVideo, you can also encode an alpha channel, which Motion JPEG doesn't support.

Of course, as an approximating codec, Motion JPEG can compress video much further than any perfect-fidelity codec if image quality isn't a concern.

previous | top | next

SheerVideoIs SheerVideo as powerful as DigiBeta?

Yes, SheerVideo is on average 21% more powerful (2.43) than DigiBeta (2.00), even though DigiBeta "compresses" the video by throwing away information, while SheerVideo compresses it with perfect fidelity. Sony's legendary-quality Digital Betacam uses a DCT-based approximating codec to compress the Y'CbCr 8bv 4:2:2 video by a factor of 2.00, yielding a fixed video data rate of 83 Mb/s (9.9 MiB/s). SheerVideo compresses real-world Y'CbCr 8bv 4:2:2 video by an average factor of 2.43, yielding an average data rate of 69 Mb/s (8.2 MiB/s). Note that SheerVideo also supports 4:4:4 video and RGB, as well as alpha-channel encoding, none of which DigiBeta supports.

previous | top | next

SheerVideoIs SheerVideo as powerful as DV-50?

DV-50, also known as D9, Digital-S, and DVCPRO50, is a DCT-based approximating codec that "compresses" the data by a factor of 3.3 by discarding as much information as it needs to in order to bring the Y'CbCr 8bv 4:2:2 video data rate down to 50 Mb/s (megabits per second), or 6 MiB/s (mebibytes per second).

SheerVideo, on the other hand, is a perfect-fidelity codec that never discards any information, compressing the data only as far as it can by packing it more tightly, averaging 2.43 for real-world Y'CbCr 8bv 4:2:2 footage, which yields an average data rate of 69 Mb/s (8.2 MiB/s) at NTSC or PAL DV resolution and frame rate.

So this isn't a very meaningful comparison. But the answer is no. Nevertheless, for real-world Y'CbCr 8b 4:2:2 footage, SheerVideo's compression power averages only 38% lower than DV-50's. Not a bad tradeoff, considering SheerVideo's speed and perfect fidelity. Besides, SheerVideo also supports 4:4:4 video, RGB, alpha-channel compression, and arbitrary image resolution, none of which DV supports.

previous | top | next

SheerVideoIs SheerVideo as powerful as DV or DV-25?

No. DV-25 is a DCT-based approximating ("lossy") codec that "compresses" the data by a factor of 5 by throwing away as much information as it needs to in order to bring the Y'CbCr 8bv 4:2:0 (PAL) or 4:1:1 (NTSC) video data rate down to 25 Mb/s (megabits per second), or 3 MiB/s (mebibytes per second). In the case of Apple's DV codec, this results in an average error of 3.75 b/S (bits per sample) for Y'CbCr 8bv 4:2:2, and 2.48 b/S for RGB 8b.

SheerVideo, in contrast, retains every single bit of image information, compressing the data only as far as it can with perfect fidelity by encoding it more tightly. SheerVideo doesn't yet support the Y'CbCr 8bv 4:2:0 or 4:1:1 formats, so we can't directly compare it to DV. However, SheerVideo Y'CbCr 8bv 4:2:2, which contains 4/3 as much data as 4:2:0 or 4:1:1, averages 69 Mb/s (8.2 MiB/s) at NTSC or PAL DV resolution and frame rate.

On the other hand, SheerVideo compresses 5 times as fast as DV, always maintains perfect fidelity, and supports 4:4:4 video, RGB, alpha-channel encoding, and arbitrary image resolution, none of which DV supports. But of course, for the low-end standard-definition video work for which it's tailored, DV is generally the most cost-effective way to go.

previous | top | next

SheerVideoIs SheerVideo as powerful as MPEG-2?

No. MPEG-2 is a highly developed DCT-based approximating codec that "compresses" video by approximating each component with a combination of horizontal and vertical discrete cosine waves, discarding as much detail as you're willing to lose in order to shrink the video file size. Since MPEG-2 is used as a distribution format, where the size of the video file is of far greater concern than the quality, the compression power is typically in the tens or hundreds. In theory, you could even discard all the information to attain infinite compression power.

SheerVideo, in contrast, always compresses and decompresses video with perfect fidelity, without ever introducing the slightest error, because SheerVideo is used in production, interchange, and archival, where quality and speed are of far greater concern than file size. In other words, SheerVideo compresses the video by packing the information more efficiently, rather than throwing it away. So SheerVideo's compression power is limited by the amount of information in the source video.

previous | top | next

SheerVideoIs SheerVideo as powerful as MPEG-4 or DivX?

No. MPEG-4 (one brand of which is DivX) is a complex standard that includes both DCT-based and wavelet-based approximating ("lossy") video codecs (among many other things), but no perfect-fidelity ("lossless") video codecs. MPEG-4 is meant for very-low-bandwidth distribution, where movie file size, as opposed to quality, is extremely important. As such, MPEG-4 is usually used with a compression power in the tens or hundreds, at considerable cost in quality.

SheerVideo is meant for production, interchange, and archival, where quality and speed are paramount. As such, SheerVideo is an extremely fast perfect-fidelity codec. Its compression power of 2.20 for real-world RGB 8b imagery is the highest attainable perfect-fidelity compression power at that speed, and close to the highest attainable at any speed.

previous | top | next

SheerVideoIs SheerVideo as powerful as Sorenson?

No. Sorenson Video is an excellent wavelet-based approximating codec that "compresses" video by approximating each component with a combination of horizontal and vertical wavelets, throwing away as much detail information as you feel you can afford to lose in order to reduce the video file size. At its maximum quality setting, Sorenson 3 has an average compression power of 5.48 for real-world RGB 8b footage. But ordinarily, since Sorenson is used for distribution, where file size is far more important than quality, Sorenson is used at way higher compression power, typically in the tens or hundreds.

In contrast, SheerVideo is used in production, interchange, and archival, where speed and quality are overriding concerns. Accordingly, SheerVideo is a perfect-fidelity codec that compresses video by encoding it more efficiently, without giving up a single bit of information, and on decompression perfectly restores the video without any error. So it isn't possible to meaningfully compare Sorenson's approximating "compression" power with SheerVideo's true compression power. But since you asked, the answer is that, for real-world RGB 8b footage, SheerVideo's compression power averages 2.20, well below Sorenson's minimum compression power of 5.48.

However, Sorenson 3's minimum error rate averages 1.7 bits per sample, compared to SheerVideo's 0.0. Moreover, SheerVideo compresses 20 times as fast (100 MiB/s on a 1-CPU 1 GHz Mac G4) as Sorenson 3 (5.5 MiB/s), and decompresses more than 4 times as fast (100 MiB/s) as Sorenson 3 (23 MiB/s) at maximum quality. And SheerVideo also supports alpha, unlike Sorenson.

previous | top


previous pagenext page

Copyright © 2003..2011 BitJazz Inc. All rights reserved.
Site design by BitJazz Inc.