Avant-Garfield - Creating New Comics With Neural Networks

preview_player
Показать описание
Using neural networks I generate new, never before seen Garfield comics and explain how it works.

Chapters:
0:00 Intro
0:54 Ideas
2:15 Network Architecture
3:16 First Test
4:12 Bigger Test
4:50 GANs
5:40 Embedding Encoder
7:16 Hybrid Approach
8:28 Final Results

=== Garfield Resources ===

=== Music ===
? - Aivi & Surasshu
Charmaine's Jam - Cody Wright
Рекомендации по теме
Комментарии
Автор

This looks like one of those "what English comics look like to non-English-speakers" demonstrations.

medexamtoolscom
Автор

I feel like this whole video is just an excuse to have cat pictures in those transition slides

grivar
Автор

the 'omega omega' in the last panel of the comic at 9:30 is the Garfield equivalent to 'oh time thy labyrinths' in the Borges library

flaetsbnort
Автор

Never seen this channel, but subscribed because this is awesome

TheMagicat
Автор

PCA 1: Controls if Jon appears or not
PCA 2: Size of characters
PCA 3: Height of characters
PCA 4: Lighting
PCA 5: Controls if Garfield is on the left or right
PCA 6: Controls if Garfield is sitting or not
PCA 7: Controls if Garfield and Jon are close together or not
PCA 8: Controls if Garfield appears or not
PCA 9: Controls if Garfield is in the middle or not
PCA 10 & 11: ?
PCA 12: Background color
PCA 13: Controls if Garfield is sleeping or not
PCA 14: How much green is on the comic
PCA 15-17: ?
PCA 18: Controls if Jon appears in the third panel or not
PCA 19 & 20: ?
PCA 21: Floor color (Yellow/Blue)
PCA 22-26: ?
PCA 27: Floor color (Green/Red)
PCA 28-38: ?

firstnameiskowitz
Автор

Those titles with the cats are so cute

Kiriixer
Автор

9:49 what on earth is garfield trying to tell us i don't like it

kackers
Автор

Great video, I love it. It inspired me to try and combine the markov-chain dialogue with the created images, and make fully legible abominations. I love your channel!

MultiPunci
Автор

i didnt understand anything he said, but this video was entertaining af, good work boi

cose
Автор

Garfield entered the *_v o i d_* and came back.
He shouldn't exist.

Magnus_Deus
Автор

Not even kidding here... *why is my high school yearbook being pictured at **2:13**?*

i-never-look-at-replies-lol
Автор

It would be interesting if you combined this with the markov chain comic generator to make a generator that makes virtually infinite new garfield comics. If you could teach the program to detect and remove the garbled text (or alternatively feed it a set of garfield comics with empty speech bubbles and have it detect those) and then replace it with markov-generated comic scripts, that'd be really interesting.

xxHantamachine
Автор

Maybe it needs lasagna to produce better results. And certainly it doesn't work on Mondays.

digitalunity
Автор

09:02 Flipping' the bird on Jon's face. The AI had gotten the idea

razornetout
Автор

4:50 is important
8:28 is the results

tnaoro
Автор

Transcript:
Jon: TAKE A XUERE R AVEROGI ΩeXD I GRDOΛEGA

Jon: OHNY ГIANL:PUSI I×LFFN PIECA n

Jon: PIO T KB中I ZAOGR MAY
Garfield: A AVRYOEIC IEGA GAET!

want-diversecontent
Автор

It almost feels like they’re speaking the enchanting table language lol

caprisunaddict
Автор

I dare you do it again with stable diffusion fine-tuning :D

andreasreich
Автор

What inspired you to choose garfield? Are you a fan of Bill Murray? Lorenzo Music?

hahathatsgood
Автор

Garfield after he destroyed the fabric of reality

cyankoopa