西乔 XiQiaoさんのイラストまとめ

Experiment with Color Palette, accurate and artistic outcomes.
blue & red | Yellow & blue | black & white | teal & burnt sienna
The anatomical coherence of human face is good, I picked below 4 from less than 20 generations.

#stabledifffusion2 5/n

For prompt enginering of SD2, either too short or too long prompt doesn't work well. I found SD2 can offer higher controllability with more precision than 1.5 So highly targeted prompt design is possible, not stacking of keywords or artist refer as blindfold alchemy anymore. 4/n

DFserver just added support for Stable Diffusion 2.0.

Add "-M sd2" after your prompt .

SD 2.0 has a big change that you may need to redo your prompt engineering completely

Default size in DF server: 768*769, CGS:9 https://t.co/v7kdGQJWas

two groups of extremely magic short prompt from MJ community trending.

"cute ivysaur in vines, manga"

"abonded vw bus"

how these prompts work?

More comparison.

Cubism style, lowpoly style

More comparison.

MJ V4 generated portraits with better aesthetics, more consistent results and accurate facial anatomy, except not look like Gal Gadot at all.

@_akhaliq @LambdaAPI a black cat wearing suit, typing on a laptop

花5分钟画了一张巨角鹿头顶落日的草图， 作为 inti image 在 DFserver 里跑了个 low poly 风格。结果还挺挺不错的 （成品图用PS简单修补了一下）。 |

DFserver 现已支持以图生图。对结果有更高可控要求的玩家，用数字绘画工具随便画个草稿丢进去跑，最好玩不过了。Github 链接 1/2

#Stabledifffusion

Spent 5 minutes on a quick sketch of a giant antlered deer with a sunset over its head as an init image, then used Img2img in DFserver to generate a low-poly style outcome. the Final result was simply fixed in PS.
Quite satisfied with the results.
#DFserver #Stablediffusion

还是回来玩我的 stable diffusion， generated in DFserver