Re: Human Encoding
"Dustin Stuart" <[email protected]> Sat, 20 Jul 2002 20:13:03 +0800
| Newsgroups | gmane.comp.multimedia.ogg.tarkin.devel |
|---|---|
| Message-ID | <[email protected]> |
Hey, I know its a crazy idea, but my thoughts for compression are rather than exactly describe the tree, 'quantize' the information. You would use a decoder to randomly generate a bunch of leaves perhaps only describing one. If the tree is not central to the video, pretty much any tree will do (random type, size etc.), especially in a blurred background. Of course there are some cases where this would be inappropriate and the trees would have to be described more exactly (in an avenue of poplars [how spelt?] for instance) and an encoder would have to be very clever to discern the importance of the trees. But for smaller things (walls, doors, roads, grasses etc.) even the tiniest memory of a smart encoder may make a difference. Perhaps not for Tarkin though :). BTW, Im not convinced about the human mind remembering individual pixels though ;). Dust >From: Michael Kreitzer <[email protected]> >Reply-To: [email protected] >To: [email protected] >Subject: Re: [tarkin-dev] Human Encoding >Date: Sat, 20 Jul 2002 03:30:30 -0500 > >If you only knew the magnitude of what you are suggesting... You would >still need to describe the shape, location, size, shape, texture, etc. of >every leaf on that tree. Also consider how you would have the encoder >differentiate between objects and form a descriptive outline of each frame. > >Also... nobody knows for sure if the human brain is even efficient at all >in storing visual memory. For all we know it could just be raw storage >with no special tricks. > >Using associative memory to build a scene is an interesting idea. It'd >tailor a video to something the viewers imagination would create if you >take into account that imagination is based largly on assembling an >image of things you've experienced and/or creating a unique image >based on a combination or distorting if something you have seen before. > >Unfortunately you'd have to find a way to force the brain to see only >what is in the minds eye, and then you have to somehow tell the brain >what to put there. That technology is a good long ways away =). > >Descriptive languages as we know them today create what we call 3d >rendered images. We aren't exactly at photorealism in that arena yet. >It's an interesting idea that we just don't have the technology to create >today and won't have to create for a long long while. > >On Friday 19 July 2002 08:22 pm, Dustin Stuart wrote: > > Hey everyone, > > > > It seems to me that the most effective video compression algorithm that > > will every be created must focus on what the human mind percieves rather > > than what a camera lens captures. Although video compressors claim to > > already do this using Psycovisual Enhancements, the codec I am talking > > about would represent a tree as a memory of a tree (30m Camphor Laurel >with > > gangly bark etc...). Then a hypothetical decoder would generate a tree >from > > the description. It sounds crazy but perhaps the final solution in this > > problem of video compression. > > > > Maybe one day Tarkin will get to 'Movie on a Zip Disk' but what about > > 'Movie on a Floppy'? > > > > Look forward to your comments, > > Dust > > > > _________________________________________________________________ > > Send and receive Hotmail on your mobile device: http://mobile.msn.com > > > > > > --- >8 ---- > > List archives: http://www.xiph.org/archives/ > > Ogg project homepage: http://www.xiph.org/ogg/ > > To unsubscribe from this list, send a message to > > '[email protected]' containing only the word 'unsubscribe' in >the > > body. No subject is needed. Unsubscribe messages sent to the list will >be > > ignored/filtered. > > >--- >8 ---- >List archives: http://www.xiph.org/archives/ >Ogg project homepage: http://www.xiph.org/ogg/ >To unsubscribe from this list, send a message to >'[email protected]' >containing only the word 'unsubscribe' in the body. No subject is needed. >Unsubscribe messages sent to the list will be ignored/filtered. _________________________________________________________________ MSN Photos is the easiest way to share and print your photos: http://photos.msn.com/support/worldwide.aspx --- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to '[email protected]' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.