Midjourney V5 發(fā)布!AI繪畫圈再次炸鍋 - 深度體驗(yàn)V5

繼chatGPT V4之后,Midjourney也在昨日發(fā)布了V5 測(cè)試版,雖然還不是完整的版本,但是已經(jīng)可以在畫質(zhì)上碾壓之前所有的版本了!
廢話不多說(shuō)~ 直接開(kāi)始體驗(yàn)哈o(* ̄︶ ̄*)o
同一段Prompt:
【official game art, landscape, highly detailed, 3D game render, Age: 8 years old, Hair: Messy brown, Eyes: Big and bright green, conveying curiosity and wonder, Clothing: Practical and well-worn shirt and pants, along with a backpack or satchel to hold essential items, Accessories: Slingshot or wooden sword strapped to his back, indicating readiness for a fight or challenge; a loyal dog or mischievous monkey as a pet or companion, Environment: A lush and vibrant forest with towering trees, babbling brooks, and hidden paths, Pose: The boy should be shown on a journey, walking along a winding trail or standing at the edge of a rushing river, with a look of determination and excitement on his face, trending on Artstation, --ar 2:1】
V4 版本出的圖:

V5 版本 - style med [這里需要在/Settings 里面選擇]

V5 版本 - style high

V5 版本 -style very high

可以看到,從畫質(zhì)的提升,對(duì)prompt提詞的讀取,還有細(xì)節(jié)處理上都有質(zhì)的飛躍!
測(cè)試了多種風(fēng)格以后,發(fā)現(xiàn)V5對(duì)于偏寫實(shí),設(shè)定方向的圖效果要好很多,但是二次元的圖,依然不是很理想,以下是我的一些測(cè)試圖,附上相應(yīng)的prompt提詞,感興趣的小伙伴們可以自行嘗試下o(* ̄︶ ̄*)o





以上的圖都是基于V5 -style med,感興趣的小伙伴們也可以嘗試-style high 和 -style very high,效果應(yīng)該會(huì)更驚艷!
也歡迎大家在評(píng)論區(qū)多多討論更多的prompt提詞,繪更多更棒的圖~