Re: [閒聊] 有人嘗試用ai繪畫還原小說著名場景嗎

※ 引述《meblessme (人身攻擊是爛人)》之銘言：
: 剛剛用Bing image creator 嘗試還原過去科幻小說的著名場景，似乎還原的還可以，有人
: 嘗試過嗎
: 下面是某個小男孩在廢棄星球上修著機械零件
: https://reurl.cc/gZpeyR
: https://www.bing.com/images/create/vast-nebula-with-colorful-suspended-gas-forma
: tions/6435753b626d42d89fea22c9a14f3dac?id=FXJH0unhf4CAWTORhXuQHw%3d%3d&view=deta
: ilv2&idpp=genimg&FORM=GCRIDP&mode=overlay
: https://i.imgur.com/3TrrdsN.jpg

剛用我自己的SD模型跑了一下

用的模型有放在CIVITAI

https://civitai.com/models/1540?modelVersionId=40626

有稍微加工了一下prompts

實話說,目前最容易出好結果的還是midjourney.但是要錢

自由度最大可以自己弄模型裝本機端的的是Stable diffusion,不要錢但是
對顯卡有需求
產圖低標1660S
訓練低標2060 12G

有興趣可以到AI繪圖板看看

prompts:masterpiece, best quality,
vast nebula with colorful suspended gas formations in the distant background,futuristic interstellar spaceport top-left, young boy with mechanical arm
wearing futuristic yet slightly grease-stained clothing, sitting on the
ground, organizing scattered mechanical parts, ,
best quality,masterpiece,highly detailed,ultra-detailed
, unreal engin 5,octane render,(2D:1.2),
science fiction,

by __artist-scifi__ __artist-scifi__ __artist-scifi__

--