2443235 Sorry for the very late reply - Real Life has been busy.
It depends on what you mean by "built-in." JPEG doesn't have some probabilistic image model stored in it's guts, but the way the algorithm works (Chroma subsampling into DCT into non-uniform quantization into entropy coding) it only works with data that has certain mathematical properties assuming certain error models that natural images perceived by humans happen to possess. You may want to consider looking into perceptual measures of image reconstruction error.
As for the style thing, Google's search likely uses a neural net where one or more "neurons" triggers on things that have very high contrast and are primarily one color, which line drawings would be.
Twilight is the BEST
mlkj
Mitch H
JustSomeRandomUsernameBad Horse
AJtheRaven
BrightIdea
homicidal doktor
Whitewolf_Stormrunner
Dream Seeker
NightFuryStriker
Cross Lament
Naduran
djazz
Spiral_Donut
Pearple Prose
Bookish Delight
azore24
Skywriter
RavensDagger
CoastalSarv
Benman
defender2222
Cyanide
xTSGx
paleowriter
Andrew Joshua Talon
AuroraDawn
HellRyden
Lucres
Conner Cogwork
Varanus
Ponydora Prancypants
Pen Stroke
Eakin
kuromi
AugieDog
Bad_Seed_72
Pineta
Twinkletail
geldon
TheWraithWriter
Cola_Bubble_Gum
Einhander
The Equestrian Gentlecolt
Capn_Chryssalid
Immanuel
iisaw
TheLastBrunnenG
dlazerous
horizon
Spike120812
AestheticB
BronyWriter
Obselescence
Steel Resolve
CyborgSamurai
The Rarispy
Saphroneth
Anonymous Pegasus
Cloudy Skies
soulpillar
Aragon
Happy date of birth if that is true.
2459300
I'll check those out. Thanks!
2459273
JwildFire definitely, and Mandlebulb3d if you want anything more to do with 3D.
2459241
Nice. I've only used Apophysis for my fractals. Are there any other good user-friendly ones you'd recommend?
2443235
Sorry for the very late reply - Real Life has been busy.
It depends on what you mean by "built-in." JPEG doesn't have some probabilistic image model stored in it's guts, but the way the algorithm works (Chroma subsampling into DCT into non-uniform quantization into entropy coding) it only works with data that has certain mathematical properties assuming certain error models that natural images perceived by humans happen to possess. You may want to consider looking into perceptual measures of image reconstruction error.
As for the style thing, Google's search likely uses a neural net where one or more "neurons" triggers on things that have very high contrast and are primarily one color, which line drawings would be.