AI Drawing modifications 人工智能電繪生成及修改圖

Patreon優先文章/更多圖片及高解像圖只在Patreon： https://www.patreon.com/posts/67685219

As AI modelling/generation of AI artwork, painting skill etc becomes more sophisticated, many artists are talking about it and some fear it would put many artists or illustrators out of business. I checked them out of course… yes they are amazing. In some areas they look like a CG masterpiece. But it’s even colder than many CG drawings — you can feel the lack of human touch. Here I use “Disoc Diffusion” to create some pieces and then edit the AI rendering paintings to see how they look and feel.

隨著人工智能（AI）生成藝術/繪圖等技術變得越來越成熟，許多畫師都在談論它，有些人擔心這會使許多藝術家或插畫家減少工作機會。我當然看看究竟是什麼回事了 — 無可否認，現在的AI真的很厲害，那些生成作品看起來就真的像是某位畫師的電腦繪圖作品一樣。不過我仍覺得生成出來的作品很冷，你能感覺到它缺乏「人」的情感。 這裡我嘗試使用“Disoc Diffusion”創建一些圖像，然後自己再用電繪後期加工，看看感覺如何。

I would say yes they could really replace a lot of artists, but artists can also use it to help finish their CG painting faster, or to get the fundamental of the piece for the larger work frame. The world is changing fast, we need to adapt and instead of being left out of the trend, try to think positively — maybe we can learn new tools and adopt them with our current skills.

Another thing is that they can’t be made accurate enough to show emotion on the pieces (at least in my opinion) or match a client’s specifications compared to what an artist has in mind. They might get very close to what we have in mind, but at the end of the day, they still can’t directly read your mind, and even if the generation goes in a slightly different direction, the hours of modelling can be wasted and you have to do another rendering. So there is still some room for improvement — I know the technology is advancing fast and may soon be a lot closer to what we envision, but again it also depends on the artist/illustrator whose skills do the retouching of the image would affect piece.

What I think about this whole AI painting is — don’t give up the traditional art! What you can do with the brain and hand will help your CG painting. What you can do in CG painting will help your enhance and edit your CG rendering paintings. The human brain is still developing.

我認為，AI的確可以取代很多繪師，然而繪師也可以用它來幫助更快地完成他們的電繪作品，或者為更大的工作框架獲得作品的基礎。世界瞬息萬變，我們需要適應，而不是被排除在潮流之外，所以我試著積極思考：也許我們可以學習新工具，並以我們現有的技術讓它們變成我們的其中一個技能。

另一件事是，與繪師的想法相比，AI始終無法表達出作品的情感（至少在我看來），又或與客戶的規格相匹配。AI可能會非常接近我們的想法，但歸根結底Ai仍然無法直接讀懂你的想法，如果Ai朝著跟我們事前想像的繪圖稍微不同的方向發展，連建模的時間也會被浪費掉，你必須做另一個生成，所以仍有一些改進的空間。

當然我知道科技一日千里，AI繪圖可能很快就會更接近我們的設想，但這也取決於藝術家/插畫師，他們的圖像修飾技更能影響作品的最終效果。我對這AI生成繪畫開始普及的看法是：不要放棄傳統藝術！CG繪圖本來也基於畫師本身的技術，那你的基本功也對於編輯那些AI生成圖有關鍵影響。人腦仍會不停進化呢。

Below are my modifications compare with the rendering pieces: 以下是我的修改與電腦生成的比較：

AI can “deftly” create background objects, but everything related to humans and animals is still a bit out of shape, thus artists should focus on adjusting the creature part and you’ll have an amazing result. AI可以巧妙地創造出背景物體，但對一切有關人和動物的生成仍然有點走樣，因此繪師應該專注於調整作品中「生物」的部分，結果就會很不錯。

Below：The AI can almost read your mind just by typing in keywords, but you still have to move things around. 下圖：人工智能幾乎可以通過輸入關鍵字來讀懂你的想法，但你仍然需要動手改動一下。

Below: Based on the AI generated image, I adjusted the hue to my liking and added a woman’s face. This part requires illustrator skills. 下圖：AI 生成的圖像，我根據自己的喜好調整了色調，並加了一張人臉。 這部分需要插畫師的技能。

But I would say it does something amazing. Sometimes like below I can only edit a small part. But did it really produced what I have in mind? I would say not 100% — and during the modification I feel like the AI already generated something, and therefore I had to follow it’s art direction to finish or edit the image. Hence in a way, the AI is guiding/leading me instead of just helping me. You have to remember what you have in mind before the generation or it will control or limit your creativity. Is that good or bad? See how you use it… Very interesting.

但我仍會說AI真的很厲害。 像下面這張，我只需要編輯一小部分而已。然而這生成的結果真的是我輸入關鍵字時的想法嗎？ 我會說不是百份百 — 在修改過程中，我覺得只是因為AI 已經生成了一些東西，因此我必須按照它的art direction來編輯完成圖像。因此在某種程度上，人工智能在指導/控制我，而不僅僅是幫助我。

你必須記住在圖像生成前你自己其實想要什麼樣的作品，否則它將控制或限制你的創造力。是好還是壞？ 看看你如何使用AI吧……非常有趣。

I might occasionally use AI from for some of my artwork, but I won’t rely on it too much. After all, art (if it’s fine art) is supposed to express the artist’s emotions and creativity, when it was the computer getting half or even more than half of my work done, it doesn’t feel quite right. However like I said, it helps when I need to quickly do a few things for a promotional/commercial item.

我可能偶爾會在一些作品中使用 AI，但我不會過分依賴它。畢竟，美術應該表達藝術家的情感和創造力，當我一半甚至超過一半的工作是由AI完成時，感覺總不太對勁。然而像我說的那樣，當我需要快速做一些事情如商業項目時，它可能會有所幫助。

＝＝＝

以下分享其他我沒有編輯過的生成作品： Here are other generated works that I have not edited:

======

Those two were the very first test. I was scared… WTF are they

Diamonds and flowers etc. They may look good, but they are very far from what I have in my mind.

Below: The sort of human looking creatures are fooking scary.

Below: Face off… I used “Victoria Harbour” in one of the keywords.

Below: TOO BRIGHT

Below: All those very dark ones.

Below: Ghibli Studio?

Below: They are all very nice, but I can’t use them.

Below: I used “Madonna” and “Michael Jackson” and AI gave me this:

Below: Again very nice background, I could use them for something else.

Below: They look like real watercolour paintings, but they are nowhere near what I imagine

＝＝＝＝

Below are made by a mobile app! Dream by wombo

What’s even more scary is this app is very quick and you don’t need to know any coding (For disco diffusion, you need some basic knowledge )

Just out of curious, what if I type in my artist name and my Ladykylie alias?

Now something completely random.

=====

【如果您喜歡我的文章/繪圖，請支持我的Patreon，可以看到昔日及不公開的其他Patrons Only文章。謝謝！https://www.patreon.com/Ladykylie 】

＝＝＝＝