|
--- |
|
license: gpl-3.0 |
|
datasets: |
|
- pawkanarek/spraix_1024_best_96 |
|
library_name: diffusers |
|
pipeline_tag: text-to-image |
|
tags: |
|
- art |
|
--- |
|
Intented to use with FLAX |
|
|
|
Model will generate ugly sprite animations, out of shape, deformed, pixelated and uneven. |
|
Prompt ideas |
|
``` |
|
"12-frame sprite animation of: dinosaur with backpack, that: is running, facing: East" |
|
"4-frame sprite animation of: girl with big sword, that: is idle, facing: West" |
|
"8-frame sprite animation of: a devil, that: swings his sword, facing: West" |
|
"6-frame sprite animation of: a wizzard, that: is casting a fireball, facing: East" |
|
``` |
|
Trained for 48 hours, on Google TPU v3-8 with this this script: https://github.com/PawKanarek/spraix/blob/48d8c209a359622e6db56e6d555667ac466dc952/train_text_to_image_flax_sdxl.py |
|
|
|
Images sampled from 16 - 32 training epochs (90MB image): |
|
https://github.com/PawKanarek/spraix/blob/48d8c209a359622e6db56e6d555667ac466dc952/output/sdxl-base-flax/sdxl_best_96_32.png |