Re: [請益] 只會寫程式的話該怎麼解決素材問題?

看板GameDesign (遊戲設計)作者ZMTL (Zaious.)時間1年前 (2023/01/06 11:03)推噓11(11推 0噓 11→)

留言22則, 13人參與討論串5/6 (看更多)

※ 引述《ddavid (謊言接線生)》之銘言： : ※ 引述《googled (15357)》之銘言： : : 想請問個很魯的問題 : : 小弟本身只會寫程式，一直有個遊戲設計的夢想給他圓起來 : : 從以前玩RPG Maker到Unity，一直到最近的Godot : : 在摸索時總會想到素材的問題不知該怎麼解決 : : 本身有找過網路上付費或免費的素材 : : 但都會離自己的需求差那麼一些 : : 可能A作者的素材包裡面有30%符合自己需求 : : B作者的素材也有40%符合自己的要的內容 : : 可是A跟B的素材風格卻不一樣 : : 兩款素材放在同一款遊戲感覺很怪 : : 同樣的音樂跟音效的部分也有同樣的問題 : : 不知道各位大大都是怎麼解決這類問題的? : : 是到最後乾脆自學繪畫，還是會直接請一個御用畫師量身訂造? : : 或者是不管風格問題，網路上有什麼素材符合就買什麼資源? : : 謝謝 : 關於素材運用，同人或獨立遊戲中有大量運用案例。不過近年我看過最傑出的例 : 2. 其實生在目前這個年代非常幸運，AI 繪圖已經有遊戲拿來實用化了，花點時間 : 學習詠唱一下，就可以產生一定水準的圖片。當然如果要畫同一個人物的多張圖可能 : 就比較不容易，若有此類要求，就要更有耐心地大量產生 AI 圖後仔細篩選更相近的 : 人物圖（提升詠唱技術也會有點幫助但還是要靠運氣）。 (略) 聊聊這部份，雖然我沒正經的遊戲製作經驗，但也參與過一些課堂專案，後來還幫某個朋友當貓手用Unity趕過遊戲畢制，同時還幫他整理素材，那時候他的選擇是用Unity Store用一部份免費的、去Booth買一些素材、立繪自己畫。但如果是現在，真的可以大部分靠AI，我自己現在是AI繪圖板板主，剛剛偶然點進這個板看到這個議題，剛好可以拿一些現有的例子來講。先隨手找些例子給大家，我直接貼未放大版的四格圖，都可以再出大圖。以下都來自AI繪圖服務Midjourney的官方畫廊（關鍵字：UI,icon,video,game,rpg...），部份是這幾天的熱門部份是我的書籤： ICON: 「Ui icons, buttons and assets plus a screen concept for a game in the style of legends of runeterra on a white background, 4K」 https://cdn.midjourney.com/9adbba8b-540c-4b98-bd87-69d4ce9329d5/grid_0.png

「ui elements, frames, buttons, rpg game」 https://cdn.midjourney.com/fd9732cd-d472-48c5-b426-e754d0afe9c1/grid_0.png

「pickaxe, rocks rare gems icon for a rpg game - dark fantasy, rpg, atmospheric, detailed, intricate, brown and gray, delicate light, circle shape with rounded corners, pieces of cloth, stone background」 https://cdn.midjourney.com/d13d5f4d-32b5-45fb-b881-9c9aa27fb461/grid_0.png

「Ui icons, buttons and assets plus a screen concept for a game in the style of legends of runeterra on a white background, 4K」 https://cdn.midjourney.com/9adbba8b-540c-4b98-bd87-69d4ce9329d5/grid_0.png

UI：「pokemon themed UI UX mobile interface statistics charts infographics sharp lines vector no text」 https://cdn.midjourney.com/d13ab871-d818-4ef6-b59d-401975377e88/grid_0.png

「Beautiful hq dashboard, design, valleys / mountains, face of norse god / crows, graphs images and stats, ux/ui」 https://cdn.midjourney.com/f7eec853-c6c6-4032-93ce-2ee0a09f70f5/grid_0.png

「UI for an app that tracks your daily tasks like a fantasy video game skills」 https://cdn.midjourney.com/077b756f-557e-47d1-a469-ab06c91ea868/grid_0.png

「role-playing character sheet, ui/ux」 https://cdn.midjourney.com/1aaba7fb-1d1d-4e08-a2fe-f06314e0dd82/grid_0.png

「interface for a role-playing game set in a fantasy world. The interface should include elements such as a character stats display, a inventory menu, and a map of the game world. The style should be inspired by medieval fantasy art, with a color palette of rich blues, purples, and golds」 https://cdn.midjourney.com/024e5e72-cd4d-4bd4-a6f4-9fd4b3a581ea/grid_0.png

道具素材：「Sprite sheet with game visuals, game assets, 3D, cinematic lighting, designed, pc game, steampunk, gray background, intricate detail, 8k,」 https://cdn.midjourney.com/aea2642c-857f-4b48-bdcf-1aecd7f35471/grid_0.png

「Sprite sheet with game visuals, game assets, 3D, cinematic lighting, designed, pc game, steampunk floor tiles, gray background, intricate detail, 8k,」 https://cdn.midjourney.com/d315fa5e-b7aa-4bdc-bc36-3328b5314a74/grid_0.png

角色：「cute elf warrior, full body shot, bow and arrows, forest, fantasy, cinematic, tarot card, anime, kawaii, HD, highly detailed, detailed hands」 https://cdn.midjourney.com/4520cb44-1e51-4c8a-8355-423a60bc038a/grid_0.png

「collection of islamic avatars for a video game,full body ,Disney style」 https://cdn.midjourney.com/ab70a57c-fce6-41d1-9150-8c4df2146899/grid_0.png

「In the Dungeons & Dragons fantasy role-playing game, a female gnome artificer is a character class that specializes in creating magic items and other magical devices. Artificers are highly skilled craftsmen and engineers who use their knowledge of magic and technology to create powerful and often complex magical items. They are able to imbue their creations with magical energy, and may also have the ability to repair and modify existing magic items. Artificers are often depicted as being highly intelligent and resourceful, with a deep understanding of the principles of magic and technology. They may also have abilities that allow them to imbue their own bodies with magical energy, or to enhance the abilities of their allies, cinematic effects, octane render」 https://cdn.midjourney.com/072c8936-17a6-464b-89b8-5f69796e0ca2/grid_0.png

「cute child boy cyborg meka mecha with wings, winged avatar, feet to head picture whole body, futuristic character using modern clothes, dynamic standing pose, using semi half mecha eyewear, modern fashion, semi half mecha accessories, the character has a semi mecha gun, wepaon, knives, a circle shape in the background, sword or katana in his backside, main focal point is the character, colorful, high contrast, good lighting, good shadow, thick bold textures, thick bold outlines, eyes looking at the front, full body, detailed, by valorant game, cartoonic style」 https://cdn.midjourney.com/d18d4e6b-92dd-4d29-a875-53133b5b9f34/grid_0.png

「Deadpool as a long white hair vampiric chibi female, concept design sheet, white background, warhammer 40k watercolor」 https://cdn.midjourney.com/8165d433-0d81-495c-a2dd-acfa78766b1c/grid_0.png

角色插畫：「evil queen sitting on her thrown in her ancient palace, looking down at viewer, stone gargoyles, tapestries, braziers, elaborate background, magical particles of light, smoke wisps, pointing at viewer, microdetails, vibrant colors, bold linework, digital illustration, concept art, fantasy theme, pop art influence, eye-catching composition, strong contrasting colors, dynamic poses, dynamic perspective, playful, energetic」 https://cdn.midjourney.com/669822b7-13f2-4849-8a29-c02584aadef2/grid_0.png

「Female Dark Magician, Shadow Magic, Dramatic, Dark Magic Fire, concept art, Fantasy, Frank Frazetta, Aaron Griffin, Lucas Graciano, Magali Villeneuve, Pino Deani」 https://cdn.midjourney.com/9a190ff5-44da-4229-a938-8f475e24eb5e/grid_0.png

「full body portrait of Hecate Greek Goddess of Witchcraft, Magic, Crossroads, beautiful greek woman with half black, half white hair, wearing all black, two black dog, holding torch, detailed, dramatic lighting, by J.C. Leyendecker 」 https://cdn.midjourney.com/99f7e657-df5c-47ff-9692-b30db096a5e2/grid_0.png

場景插畫: 「a futuristic and devastated street with destroyed skyscrapers and destroyed cars by night, no light, cyberpunk style, very realistic, very detailed」 https://cdn.midjourney.com/924febb1-7ac1-4e0a-9dc4-7e054338e835/grid_0.png

「Japanese Garden, pond, water lillies, orchids, lotus flowers, sand, rocks, boulder, lush shrubs, hyperdetailed, Erin Hanson/Donato Giancola/Nicolas de Stael, cinematic lighting, long shadows, saturated contrast」 https://cdn.midjourney.com/8992bd6c-5080-46a6-8893-dd0b550a6453/grid_0.png

「a shot of the Secret Sacred Tree, hidden within a cavern, guarded by powerful Arcana wielders, by Andrew Kurniawan, pexels contest winner, fantasy theme, high fantasy time period, intricate details, glowing aura, dark and mysterious atmosphere, ancient and sacred, hidden and protected, powerful magic emanating from the tree, ethereal light」 https://cdn.midjourney.com/4701dc06-df9d-4d94-a93c-e3a7c32cdaf7/grid_0.png

場景: 「8 bit office interior」 https://cdn.midjourney.com/4f56508c-c836-4dd0-9018-9ad56039b97e/grid_0.png

「An isometric map of the Golden Oasis, a popular tavern & inn located in the heart of Al-Rashid, is depicted in this digital art piece. The detailed map showcases the various rooms and features of the establishment, including the common room, bar, stage for entertainment, and guest rooms. The Golden Oasis is depicted in a fantasy setting, likely inspired by the tabletop role-playing game Dungeons & Dragons, adding to its allure and intrigue. The intricacies of the map make it a visually interesting image, inviting the viewer to explore and discover all that the Golden Oasis has to offer.」 https://cdn.midjourney.com/4a37c8cd-125c-4c6c-97c3-8791630c5ae4/grid_0.png

「abandoned room with secrets, flat style, illustration, limited colour palette, overhead view as in role-playing games.」 https://cdn.midjourney.com/6c2fe5ec-8386-4cad-8fcd-4bdeb3882327/grid_0.png

其他素材：「isometric beehive village, game design asset, environment, hand drawn, don't starve together style, on white background」 https://cdn.midjourney.com/e34ffd96-6e35-4215-88c9-f56a3a72358c/grid_0.png

地圖應該也可以，但當前找不到範例，前幾天我做的「map design, fantasy world map of video games, D&D Style, Pangaea, Historical fantasy, parchment scroll, 8k」 https://images.plurk.com/6ZrX8BizwoFrlYo45ODv7v.png

之前存的角色設計稿（我忘記咒語了） https://cdn.midjourney.com/36ca1d7b-7f0a-499d-a856-693ef7dadfa3/grid_0.png

請養成每天逛畫廊隨手按書籤的好習慣。＼至於想要知道怎麼樣風格統一，可以參考人家賣的穩定召喚咒語：（我自己是買了幾組後就大概抓到訣竅可以自己施法了） Fantasy Landscapes https://promptbase.com/prompt/fantasy-landscapes-2 Fantasy Gem Game Assets https://promptbase.com/prompt/fantasy-gem-game-assets Isometric Floating Cities https://promptbase.com/prompt/isometric-floating-cities ＼之前參加AIGC年會，跟本板這個主題比較相關的部份： https://www.businessyee.com/article/933-AI-Generated-Content-Event 學習模式與工作流程因 AI 產生變革在政大教授數位內容圖表的李怡志分享自己如何幫助學生利用 AI 學習設計，雖然學生學習圖表就像學習書法一樣要經過「臨摹」的鍛鍊（或說 Copy），與其在 Behance、 Pinterest 上瀏覽案例，不如利用 AI 臨摹來學習效果更好。例如在 Midjourney 輸入 inforgraphic 加上圖表主題關鍵字，「一小時就能生成相當一本書厚度的範例。經由認識本體，了解結構，最後組合出自己的作品。AI 可以幫助學生掌握結構，更快達成理想的成果。」雷亞遊戲執行長游名揚則從遊戲產業的角度討論，AI 會如何改變遊戲開發的流程。例如遊戲產品的構成包括人物、道具、icon，以及遊戲廣告、登入頁面的製作等，透過與 AI 協作，將使開發速度提升百倍以上。例如可以讓 ChatGPT 生成角色的招式（例如巫女的招數），或是提供角色的特徵，讓它生成可用 AI 繪圖的 prompt／咒語，人只要發想以及熟練如何跟 AI 溝通即可，而新的工作如「AI 溝通師」因此而生。游名揚指出，有 AI 的協助，工作流程將產生變革，以往製作遊戲角色通常是決定人設，繪製草稿，最後完稿。現在可以決定人設後直接使用 AI 完稿，再來調整人物設定。「重新洗牌的時間即將到來，沒有跟上 AI 導入的公司將被淘汰。」補充雷亞的文字應用部分: 「這個角色的背景是OOO 職業XXX 請根據此幫角色設計10個招式名稱並翻譯成日韓英文」＼也歡迎來AI_Art板我們聊聊你想要什麼、我們可以怎麼做：） -- AI_Art AI藝術與創作板歡迎各方前來討論AIGC創作與相關議題！位置：戰略高手 > DigitalWare > AI_Art -- ※ 發信站: 批踢踢實業坊(ptt.cc), 來自: 123.194.160.178 (臺灣) ※ 文章網址: https://www.ptt.cc/bbs/GameDesign/M.1672974235.A.D77.html

推

coolrobin

01/06 13:01, 1年前 , 1^F

01/06 13:01, 1^F

→

ZMTL

01/06 18:20, 1年前 , 2^F

01/06 18:20, 2^F

推

aegis123321

01/06 19:56, 1年前 , 3^F

01/06 19:56, 3^F

推

iLeyaSin365

01/06 20:27, 1年前 , 4^F

01/06 20:27, 4^F

→

ZMTL

01/06 21:17, 1年前 , 5^F

01/06 21:17, 5^F

→

ZMTL

01/06 21:19, 1年前 , 6^F

01/06 21:19, 6^F

→

ZMTL

01/06 21:19, 1年前 , 7^F

01/06 21:19, 7^F

→

ZMTL

01/07 11:21, 1年前 , 8^F

01/07 11:21, 8^F

→

ZMTL

01/07 11:21, 1年前 , 9^F

01/07 11:21, 9^F

推

milk830122

01/07 13:18, 1年前 , 10^F

01/07 13:18, 10^F

→

testPtt

01/07 13:47, 1年前 , 11^F

01/07 13:47, 11^F

→

ZMTL

01/07 14:38, 1年前 , 12^F

01/07 14:38, 12^F

→

ZMTL

01/07 14:38, 1年前 , 13^F

01/07 14:38, 13^F

AI繪圖用於遊戲素材製作，剛剛看到更接近成品的例子了: Midjourney x Stable Diffusion 協作繪製紅色警戒風格的遊戲素材文章連結: 讓遊戲公司震驚！可微調效果，AI畫出高品質遊戲美術素材了 http://www.gamelook.com.cn/2022/12/504273 先讓MJ跑底圖（MJ在概念創作上比SD強很多） https://images.plurk.com/6sOzd6C7452ottNzBKOFkJ.jpg

整理風格相近的，餵給SD訓練模型（SD在統一風格上比MJ強很多）生一個掩體： https://images.plurk.com/GQWghSPj5A6WGiEINeFv3.jpg

生一個碉堡： https://images.plurk.com/27RysOOwMKJm7symR9DGK1.jpg

生一個核電廠、雷達站、煉油廠： https://images.plurk.com/2aFR9HDoLZM85lqlfj8coT.jpg

範例的創作人是Scenario聯合創始人Emmanuel de Maistre https://twitter.com/emmanuel_2m Scenario是一個事前登錄中的AIGC工具，將專門用來生產遊戲美術素材（我猜就是這個案例的模組化之類？） https://www.scenario.gg/ ※ 編輯: ZMTL (123.193.180.122 臺灣), 01/08/2023 15:25:54

推

wangm4a1

01/08 23:56, 1年前 , 14^F

01/08 23:56, 14^F

推

googled

01/09 10:49, 1年前 , 15^F

01/09 10:49, 15^F

推

nicetw20xx

01/10 00:07, 1年前 , 16^F

01/10 00:07, 16^F

→

nicetw20xx

01/10 00:09, 1年前 , 17^F

01/10 00:09, 17^F

→

nicetw20xx

01/10 00:10, 1年前 , 18^F

01/10 00:10, 18^F

Midjourney是30$美元/月訂閱制，你想有你生成的圖片的所有權，可用於商業。但你運算的過程跟產物有機會被官方公開並被他人看到，並公開咒語，默認任何人都可以二次利用。 60$美元/月的方案加上成果保密(Private)模式，另外營利超過多少的大公司要另外付費。 Stable Diffution生成的圖片也可以用於商業，但默認是CC0授權，也就是其他人可以直接拿你的圖片去用。目前商用許可其實都沒問題（Adboe Stock也在販售AI生成圖片），但要保護產出的圖片被他人使用時主張自己的權利目前缺乏法律實務判例。

推

Lhmstu

01/10 02:05, 1年前 , 19^F

01/10 02:05, 19^F

推

ctrlbreak

01/10 15:57, 1年前 , 20^F

01/10 15:57, 20^F

MJ官方有說他們的下一步是3D建模然後3D模型的材質運算現在有這個，是用SD做到的Blender外掛 https://www.youtube.com/watch?v=05YMQs9Drb4