SDXL 0.9を試してみた。baseのみとbase+refinerで生成してみる。Midjourney クラスだわ。

12 47

おはようございます〜

SDXL 0.9が申請した人に続々許可が出てるらしい?
アカデミックでも研究者でもないから特になにもしてないけどWDのチューニング楽しみだね〜

といいつつこれはニジジャーニー

9 51

Current state of my finetuning of
Please welcome AnyLoRA XL (with different substyles).
Using DreamShaper as refiner here

10 37

First tries on SDXL.
from the paper it's using penultimate layers so it's probably better to add clip skip2 on comfy? Didn't check the exact impl here. The last sd1 i2i is to remove the refiner "artstation-ish" bias. seems to work pretty well for hand fixes

0 6

テキストから画像を生成する新しい画像生成AI「SDXL 0.9」が発表された。Stable Diffusionの様々なバージョンの中で最も高性能とのこと。まずは研究用の公開で、7月中旬には改良を加えてオープンリリースとのこと(左の画像は以前のSDXL Beta、右の画像が最新のSDXL 0.9)
https://t.co/FJI50ULfRm

25 70

3. Color painting

Color painting in SDXL gives a smooth touch.
Change the color by putting prompt like complementary, split complementary,triadic or tetradic.

0 0

Wow! It's actually really good!
The 2 first ones were instant (with SDXL), while if you use DALL-E you may spend over 100 credits (the 3rd); it's worth it tho

0 1

Very cool prompt! Here's our results using SDXL. These are in the order of your prompt, and the /describe prompt. 🐵

0 2

Growing up with multiple blends of religions and races, I found these AI Arts to be liberating.

1 1

Testing with this prompt, but also with "image input" weight. https://t.co/gV2miQw5xl

0 2

Loving the cute doggy Twitter logo 🤩

made with

0 7

Beautyfull girl un art Nouveau style "Beautyfull girl un art Nouveau style - made with

https://t.co/TRasP82SS9

https://t.co/TRasP82SS9

0 0