使用midjourney作图的短板The shortcomings of using 'midjourney' for graphics...

  Рет қаралды 26

了不起的胡子爸爸

了不起的胡子爸爸

Күн бұрын

我的网站:www.hu-zi.com/
因为我自己写了个小说,使用AI制作的很多图画还是挺符合我的小说内容的,概念意境上能够满足我的要求,甚至超过,省了我很多时间,短时间做出来很多成品图,当看到自己的脑海里的东西跃然屏上的时候,还是挺开心的。
不过慢慢地就发现AI,在具体的细节就不太聪明了。
比如,我说“三只阿拉斯加犬一只萨摩耶一只黑色中华田园犬的合影”
AI就又进入糊涂模式了,要不就是数字就是搞不对,要不就是对应的犬种搞不明白,简直白痴一样,
我又说“一只兔子对着大猫喷火!”然后AI就给了我一堆图,要不就是谁对谁喷火搞不清楚,要不就是兔子和猫谁大谁小搞不清楚,
我寻思它数字可能是弱点,于是就搞个简单的,要“一只阿拉斯加犬头顶扣一个铁锅”,它给了我如下这些图,我简直笑喷了,无论我怎么讲那锅倒扣在狗脑袋上,它都听不明白!反倒给我添加了一个汤勺!
没办法,我让我媳妇帮我画个概念图,她几分钟就画好了,
可这也不是长久之计啊,
不过跟AI一起工作还是蛮有趣的,好像跟一个智力超群的孩子工作,他能给你超过你要求的东西,也能面对一些基本的信息手足无措,
如果你不要求细节,让它自由发挥,那它就来劲了,放飞自我,它一次出四张图,各种炫酷,
可能AI目前比较适合概念性的内容发挥,一旦你需要一些细节的东西,AI就变成小白痴了,完全是个傻白甜的形象!
Because I wrote a novel myself, many illustrations created by AI are quite in line with the content of my novel. Conceptually, they meet my requirements and even exceed them, saving me a lot of time. I can quickly produce many finished illustrations. When I see things from my imagination vividly displayed on the screen, it makes me quite happy.
However, gradually I discovered that AI is not very smart when it comes to specific details. For example, when I say "a group photo of three Alaskan Malamutes, one Samoyed, and one black Chinese rural dog," AI enters a confused mode. Either the numbers are incorrect, or it can't distinguish the corresponding dog breeds. It's like a complete idiot.
I also tried saying, "a rabbit breathing fire in front of a big cat," and AI gave me a bunch of images, either unclear about who is breathing fire at whom or unable to determine the size of the rabbit and cat. I thought maybe the issue was with numbers, so I tried something simple, like "an Alaskan Malamute with a frying pan on its head." It gave me the following images, and I couldn't help but burst into laughter. No matter how I explained it, with the pot inverted on the dog's head, it couldn't understand. Instead, it added a ladle!
There's no way around it. I had my wife help me draw a conceptual sketch, and she finished it in a few minutes. However, this is not a long-term solution.
Nevertheless, working with AI is quite interesting. It's like working with an exceptionally intelligent child. It can provide you with things that exceed your expectations, but it can also be completely clueless about some basic information. If you don't demand details and let it freely express itself, it gets enthusiastic and creative. It can generate four cool images at once.
Perhaps AI is currently more suitable for conceptual content. Once you need some detailed elements, AI becomes a little clueless, resembling a cute but foolish character!

Пікірлер
《扎妃传》第二集  十字桃花
9:35
了不起的胡子爸爸
Рет қаралды 32
Новый уровень твоей сосиски
00:33
Кушать Хочу
Рет қаралды 4,3 МЛН
Alpha Waves Heal Damage In The Body, Brain Massage While You Sleep, Improve Your Memory
1:22:11
How to Use Midjourney V6 for Ultra Realistic Product Photography
17:49
Vidu Ai | Create Pro Animations For Free | Vidu Ai Tutorial
14:01
Atomic Gains
Рет қаралды 61 М.
猫と「秋麗」
2:56
中村ケンタ
Рет қаралды 9