ImageのTwitterイラスト検索結果。 272,762 件中 2ページ目

@EphHuit here's the full image of that one if you're curious 😂

[ ADOPTABLE ] NOW ON SALE!

It's $250 *payment via PayPal

You can get 11 PNG images.

Please contact on DM or Telegram
Telegram: https://t.co/XCdCIRBjR3

“Pumu when are you gonna stop using broken chain imagery it’s getting repetitive”

NEVER ⛓️‍💥⛓️‍💥⛓️‍💥

i uploaded the 2nd image in low res my b everyone

minha imagem favorita

Coming back to try out image2video model. The generated audio is a bit wonky (since I don't provide audio inputs for testing). Also takes a while to get decent vid output, especially w/ more complex scenes. Anyways, I'll just chime in occasionally cuz I got office job to do~ 😊

it came from these 2 images, left one was made by @boy_illu

Fathia the Qalin Magus, with the combination of Calligraphy & Magic she turns words and imagery into reality.

image #oc #illustration

💚💚💚
#nimagery https://t.co/8kKU0Qm0PW

All of the dolls in this image have skins. Klukai have 3 already. VOTE FOR MECHTY!!! And keep in mind that if she gets a new skin, I may make a second doujin about her!!! VOTE FOR MECHTY!!!! https://t.co/8nb2f0lL6y

Z-Image-Turboの人種の理解度
日本人わかるのすごい

https://t.co/uZQVZbcuBp

The character who appears this time is the Red Wolf Girl! She officially appears in the comic version now, and she has already shown up several times in the supporter-only images. Due to unfortunate circumstances, she ends up being isolated in the place where the fox girl lives!

z-image lora追加でテスト
癖つよスタイル多少崩れてるやつもあるけど
トレーニング次第なのかな
プロンプト、強度、設定、派生モデル、組み合わせ、CNでまた変わりそうだけど https://t.co/WeLs7hhhM8

Z-Image turbo アニメ系LORAいくつかテスト
style変換でsdxl系今まで使ってたけど
z-image充実したら完全に移行できそう

あとは崩れずに好みの癖つよ絵柄にできるかどうか
anytestがほしい

😠 #nimagery

changed this image feature Claire instead heh

z-image-turboお試しポン出し！
恐ろしく軽くて、自環境だと1枚5秒かからずに生成できる！
ただ絵柄は安定しない！（気がする）
まだいろいろ整っていない状況でbanana proも使えるとなると、自分の制作スタイル的にはしばらくは寝かせることになりそうかなぁ、、、

Okay I forced myself to calm down through mental image training.

I'm normal again. https://t.co/hF6sT5FFqJ

最近の画像生成モデルたち
ちょっと複雑なプロンプトで
Z-Imageとnanobanana proとflux2が難なくクリア

Z-Imageが
5090で9step,1024×で3.78s
ランダム性なくて爆速
もう少し複雑でも行ける。

派生系が楽しみなモデル。