The only way I can ever see Jpeg being "replaced" is by Jewgle&frens shoving AVIF down the users' throat.
It might see some degree of wider adoption if Jewdroid 9 adopts it as the forced default image format for onboard photography and Jewsweek/Salon/Washington Post/CNN/Hareetz/Computer Bild/Moronix/Newsplus start to simultaneously shit out articles with high res AVIF images and ant res Jpeg fallbacks while informing the users how they should upgrade their webbrowsers for next-gen 4K HDR image support, otherwise AVIF won't go much further than becoming the Android analog of HEIC.
Image encoding
Liam White
Anthony Young
I think it's seeing its last years. The successor will be either HEIF, AVIF or JPEG XL.
Sebastian Collins
Better question: When are people going to ever stop using bitmaps for everything? Will a vector format with some serious backing as a GIF/PNG replacement better than the bloated XML trash that is SVG ever come along? Will the fansub/scanlation weeb community ever cobble together some combination of a vector format and automatic tracing software using UHD BD 4K raws to make incredibly tiny, yet flawless, releases?
>JPEG isn't that bad
Daniel Evans
If JPEG XL beats AVIF and is royalty-free, I'd prefer JPEG XL to make it since AV1s primary goal is video encoding where interframe compression is more important.
Noah Collins
I share your wish.
Now that's some utopia thinking. Anime is full of textures. Reproducing all those tiny 2x3 pixel shapes would take more file size than raster graphics and it's totally impossible to trace all those subpixel things. Remember Flash? It was pretty limited in what you could do. A mixed approach might work but the animation would have to be released in your mixed raster vector anime dream format.
It's more likely those vector things will get added to raster formats than the raster features to vector formats.
Pure vector graphics will mostly stay an icon or image thing. Not video. That already died long ago with Flash (which also had raster features; all content apart from basic shapes was raster graphics, so even through it was a vector format, it's contents aged badly see vid and because Flash is such a piece of shit gradients don't look that good either).
Jackson Kelly
I for one would appreciate some kind of real-time 3DCG rendering "codec", but with certain effects mostly reflections hardcoded in the video file and passed through a standardized texture mapping "decoder" so the user's GPU doesn't melt itself.
In order to avoid bloat the input's camera perspective would be raytraced during encoding with the encoder only storing visible textures+model data in the video file, with variable levels of compression.
The decoder/encoder could also automatically truncate non-discernible triangles+texture detail at smaller playback resolutions/lower encoding "bitrates".
Is this doable with CY+4 tech and if so would it be excessively bloated?
Daniel Morris
That vidrel.
Tfw a Polish museum makes Hatsune Miku sing a Polish folk song.
Nathan Turner
im sure that google and everything owned by facebook and similar will change when its possible. then they can put that storage and bandwidth money to profits instead
Oliver Russell
One of Miku's most popular songs is a Finnish folk song.
Lucas Campbell
Aside from the typical "Bézier curves and fills" vector geometry, there are other more exotic types, such as the shader-like skeletal strokes, warping, and dynamic filters, introduced in vector software like Expression and Xara.
Even in the case of Flash, the problem with the animation you posted isn't Flash, but the fact that it's "limited animation", e.g.: Rigid paper dolls, panning/scaling objects, simple 4-point stretching, lazy tweening. It's perfectly possible to do "full animation" in Flash, complete with rich shading through careful use of gradients and transparencies, vid related.
That would be ideal (especially since most anime is already done using vectors after the inking phase, so colorists can do their job easier), but hardly necessary. An approach already common in OCR and tracing software is to "punch out" sharp objects such as lineart and text for vectorization, leaving softer backgrounds to be bitmap compressed more aggressively. Combined with interframe motion detection already used in modern video codecs and framedoubler software like SVP, this would allow some measure of tweening to be reintroduced as well, further increasing both efficiency and quality.