Allegedly, the storage of high res video via vector graphics has been cracked. Vector graphics are good for posters, cartoons etc, they are infinitely scaleable, but they have always had problems with true video or photographs, because they don't handle colour/shading transitions well. Supposedly this has now been sorted, and they're saying pixels could be history within 5 years: http://www.gizmag.com/vector-video-codec/25481/
Someday, we'll be able remove hydrogen from water, so that finally, water will no longer be explosive and unsafe to drink.
Ya that's what I was thinking. Even if you could make everything scale-able to any size you would still be limited by the display itself.
Wouldn't the file sizes required to have the video capable of being displayed at ultra-high resolutions be massive? A Blu-ray movie is 40-50GB now, and the display on those is 1920x1080. So is this somehow compressing the video in a way that you can keep 8K or more in less space?
I'm not an expert on this, but the idea behind vectoring is that the image itself is not stored in a file, only information necessary to generate the image. That is, 10x10cm of red in the image is not stored as 10x10 worth of red pixels but as the mathematical formula that generates the same thing. Thus saving space.
We have rasterized screens, and they're not going away anytime soon. It doesn't matter how you store images, as long as you can produce a rasterized representation to display the image on a screen. The more detail an image or video should contain, the more data is necessary to represent the visual information. And because the information in a video can be extremely expensive to store on a per-pixel basis, you have to employ data reduction techniques. So when someone invents an algorithm to reduce data by storing some kind of vectors, then that's a perfectly valid way to do it. However, this doesn't mean that pixels are gone, because a vector-based compression or reduction algorithm doesn't free you of the need to produce a rasterized image for viewing. And it also doesn't mean that you can zoom into a vector-encoded video to see more detail, because that detail would have to be recorded by the video camera in the first place (which records rasterized images). Using vectors to store information in video files is actually employed by current video compression technologies. Most often, not every single video frame is compressed as a whole. That's because the new information in a video frame is small if you know the previous frame. So video compressors only encode the difference to known key frames. To accurately reproduce the missing pieces, compressors often also encode velocity vectors for pixel groups, which are created by motion-detection algorithms.
I would think that breaking a typical HD video still image down into vectors and regions would require a ton of processing power and that it would take a boatload of bandwidth to transfer.
To be fair, current compression technologies like H.264 also require a ton of processing power, which is why real-time compression can only be done by specialized hardware. Real-time decompression is easier, but it also requires a high-end desktop processor to do it without specialized hardware.
Wow, I work in both vector and raster images, and this is fantastic news! I really didn't think this was possible, as even converting a still image from raster to vector is spotty at best. (Literally, spotty.) This is pretty much like a cave man inventing a rock golem and then someone else saying they can possibly invent BSG's Six or Boomer model. That's how big of a leap this would be in terms of graphic design. Quite correct. Vectors are mathematical calculations (which is why this thread isn't in the Red Room.) not constrained by pixel information. A Mickey Mouse logo would be [radius of circle=q at position x,y] + [radius of circle=q at position x+2,y+2] + [radius of circle=v at position x+4,y+4] + [fill with RGB color 0,0,0]. This is in contrast to pixel information, which is [black pixel at x,y]+[black pixel at x,y]+[black pixel at x,y]+[black pixel at x,y]+[black pixel at x,y] x 10,000 pixels for each individual pixel. JPG compression does remember chunks of pixels in the same color, which is why, taking a picture of a clear blue sky at 1920x1080 pixels will produce a smaller file size than say, a 1920x1080 picture of a gay pride parade, where people are wearing fabulous rainbow patterns. It would take a boatload of processing power to convert the pixels into vectors, but as demonstrated in my above Mickey Mouse example, vector information is much less once it has been converted. So we could potentially see HD video at greatly reduced sizes, and scalable to whatever screen resolution you want. It would be easily transferable over the current internet infrastructure. Personally, I'm still very skeptical whether it can be done without a noticeable loss of quality. I've been working with both types of images half my life, and in my industry, taking anything from a highly detailed raster image to vector looks like shit. No one has really figured out a way to do this properly with even still images, and now they are claiming they can do it with high def video? Like I said, it sounds like going from rock golem to TNG's Data overnight.
If it can be done to a still image, or say a single frame of a video, then why couldn't it just be done over and over again? If you solve it for one picture you then have the solution for multiple pictures. The bottle neck is processing power.
I was impressed when I downloaded my first H.264 video file. I couldn't believe that the image I was looking at had a file size so small. From that point on, my [-]porn[/-] movie collection improved dramatically!
Yup. The unbelievable part is I've not seen a GOOD example of a highly detailed raster image being turned to vector and still retain its quality. Most of the ones I've seen make it look like the picture is painted, or has some sort of shitty Instagram filter on it. The easy ones would be any cartoon films. Those are gimmes, as they are lines anyways. The difficult ones would be, say, the battle scenes in LOTR. Or the texture of skin in Taylor Rain's asshole as she gets double penetrated.
http://www.youtube.com/watch?v=5AXyjWcj2Ds [YT="Video"]5AXyjWcj2Ds[/YT] Here is a video regarding what they are trying to accomplish, albeit at a much much higher level.
