Doom9's Forum - View Single Post

rickyshamilton · 9th April 2021, 23:06

We’re still at crossed purposes here. I clearly have not explained myself well.

A full HD BlueRay 3D movie is encoded as a compressed left eye image with a differences file for the right eye. The compression is “lossy” but set to a high subjective quality. When unpacked this gives two full HD frames. The stereoscopic information I refer to is implicit within the pair of images by virtue of their differences from one another. This element of the “information” is not present in a 2D movie, but is decodable by the human brain.

Loss of 50% of the pixel data does not necessarily equate to a change in information entropy of 50%. The notion that cutting out 50% of the horizontal pixels translates simply into a 50% change in information entropy is simply not a correct assumption. Yes, the amount of “data” has halved, but the “information” reconstructible may not have done.

I absolutely agree that using the original BlueRay stereoscopic images at 2 x HD will be better. Of course it will - just as using lossless compression instead of lossy compression would make it better. But in a similar way as the result of a BlueRay 3D playback is, to my eye, impossible to distinguish from a losslessly compressed movie, so the 1/2 horizontal SBS is close enough in subjective quality to my eye as to be almost impossible to identify in a comparison on my equipment. It is about trading off the small difference for the convenience.

Uncompressed stereoscopic movies would be 311MB/s at 24-bit colour. A 3D BlueRay might be encoded in 35Gb of data which would amount to 4.9MB/s. The two would be very hard to tell apart subjectively yet there is 63.5 times less “data” in the BlueRay. That is what I mean by information, entropy and data being not equivalent.

9th April 2021, 23:06	#1040 \| Link
rickyshamilton Registered User Join Date: Oct 2015 Posts: 5	We’re still at crossed purposes here. I clearly have not explained myself well. A full HD BlueRay 3D movie is encoded as a compressed left eye image with a differences file for the right eye. The compression is “lossy” but set to a high subjective quality. When unpacked this gives two full HD frames. The stereoscopic information I refer to is implicit within the pair of images by virtue of their differences from one another. This element of the “information” is not present in a 2D movie, but is decodable by the human brain. Loss of 50% of the pixel data does not necessarily equate to a change in information entropy of 50%. The notion that cutting out 50% of the horizontal pixels translates simply into a 50% change in information entropy is simply not a correct assumption. Yes, the amount of “data” has halved, but the “information” reconstructible may not have done. I absolutely agree that using the original BlueRay stereoscopic images at 2 x HD will be better. Of course it will - just as using lossless compression instead of lossy compression would make it better. But in a similar way as the result of a BlueRay 3D playback is, to my eye, impossible to distinguish from a losslessly compressed movie, so the 1/2 horizontal SBS is close enough in subjective quality to my eye as to be almost impossible to identify in a comparison on my equipment. It is about trading off the small difference for the convenience. Uncompressed stereoscopic movies would be 311MB/s at 24-bit colour. A 3D BlueRay might be encoded in 35Gb of data which would amount to 4.9MB/s. The two would be very hard to tell apart subjectively yet there is 63.5 times less “data” in the BlueRay. That is what I mean by information, entropy and data being not equivalent.