[AI Image Generation] Cool Illustrations Now Available: For Those Whose Illustrations Keep Falling Apart 🔰 Recommended Correction Order: PixAI🐳
Thank you for finding this article🥰
I'm Natsume, walking alongside AI while using it as it evolves day by day😊
The rainy season has started, but it's not raining and it's scorching hot💦
So, this time I'll be sending you some illustrations to help you "cool down."
Well then, let's get started🥰
[Woman Walking on Water]

[Fireworks and Memories ~Early Version~]

[Maiden of Foam]

[Bear, bear!]

[Waiting for you]

[Bon appétit?]

When you get tired, when it gets hot, come back here again.
I have released a few cool-looking works.
Yes, Natsume doesn't end things just like this🐳
When I first used PixAI, the collapse of the illustrations was really severe.
Why is that?
I didn't understand it, so I stepped away for a while. A little slump, you could say💦
And then, through another turn of fate, I returned to PixAI.
Thinking about it now, I thought that maybe it was still 'too early' for the previous me.
The 'spirit' and 'way of thinking' that come from facing AI, Natsume includes various things in
How to approach it
I use that term.
Perhaps there are people who are struggling like Natsume, or I decided to leave this for those who might come along in the future.
Here, regarding the image generation AI 'PixAI',
When generating images, not only choosing a 'model' but also
LoRA

There are also elements that affect the "depiction" of the "illustration" itself, which we call it.
At first, you might think, "I want to add all sorts of things, let's just add them," and generate it that way.
Please take a look at this🥰
[First illustration]

This is the result of adding everything I like at full capacity.
If you ask me if this is also art, I'd say "( ^ω^ )..."💦
I don't dislike it, but it's not what I'm looking for.
So, I'm putting this in the "illustration is falling apart" category.
Next, here is the same model and prompt as the first illustration,same model, promptbut withonly the LoRA changed.
[Second illustration]

Can you tell the difference??
The colors are clearer, and you can see what kind of prompt I was using to convey it.
This is just a result of changing a single LoRA.
It is still in development, but it changes like this and is more settled compared to the first one.
It is still in development, but it changes like this and is more settled compared to the first one.
💡Reasons for falling apart (First illustration)
・LoRAs are incompatible with each other
・The prompt is not good
・The model doesn't match
Now, here is the key point: at this stage, the creator has options.
① Try changing the prompt (to match the LoRA)
② Try changing the LoRA (to match the prompt)
③ Try changing the model
Here, we have "options".
Once you get used to image generation, you end up doing everything you "can do" at "once" without intending to.
In this case, what do you think it is?
It is doing ① and ②, and hopefully all of ③ as well.
Then, you end up back at "square one" after all.
I recommend that you always investigate the cause of "what" went wrong with the "result" one step at a time.
Getting stuck in a swamp.
This term carries both the meaning of having fun and the meaning of "Why? Why can't I do it?"🥰
I'm mentioning this so you don't end up in the latter kind of "swamp"🥰
First of all, as you use a "model" repeatedly, you should come to understand "this is what I like," so I recommend not changing the "model," which is your main axis, even if the illustration falls apart👆✨
※If you want to change your mood or go on an adventure, that's a different story🥰 Please feel free to try changing various things~🥰
[Illustration 3]

This is also from the first illustration.
I only changed one LoRA.
That's why LoRA is so important🥰
💡Priority for changes
・LoRA > Prompt > Model
And,
・Change them one by one
※This is just my personal opinion as a beginner, Natsume. Please understand that it varies depending on the creator🥰
Everyone has a different way of approaching image generation🥰
🐳Existing AI might change too.
🐳New AI keeps coming out. I want to try them.
In response to these changing things, prompts and operation methods are also constantly changing.
What doesn't change is,
the way we approach trying to know and understand AI, isn't it?
That is what I, Natsume, think🥰
I hope this can be of some help to everyone🥰
That's all for this time~🥰
Thank you always for your likes, comments, and adding to your magazines🥰
It is very encouraging🥰✨
Well then, have a great AI image generation life~💐

🔽PixAI Related Articles
いいなと思ったら応援しよう!
応援をどうぞよろしくお願い致します🥰頂きましたチップは創作活動に使用させて頂きます🥰✨