new function! Validate the image prompts provided in the store! Logo design edition
Store function released!
This is Ushiyama from OpenFashion.
This time, I would like to take a look at what you can do using the "store function" added to MaisonAI on February 14th.
There are now 20 prompts in three categories: general tasks, fashion-specific tasks, and image generation.
Despite the name "Store", all prompts inside are provided for free. 
First, let's take a look at what "Image Generate" was before the store was created.
If you write your own English words in a blank space like this, you won't know what to write.
Unlike AI agents and templates, there is the hurdle of having to write in English, so it was difficult to generate images.
Try the Store Image Prompt: Logo Design
Now let's try out some of the image prompts in the store!
This is a prompt that will help you create a logo.
Instructions for placing the logo in the center of the screen and the atmosphere of the logo are specified in advance.
Try changing the motif
Now, let's swap the words.
The eagle came out properly! The quality is such that it can be used as an icon image. There's only one change to make, so it's very easy to make.
Next, I chose "Torso wearing a jacket". It has a stylish look.
I chose "a man wearing brown glasses." The "brown" color is canceled out in the color scheme instructions in the second half, but it looks like it will be used as a profile picture.
Can you enter text?
Let's change the word to "Pineapple".
Since it's a logo, can you actually draw a design with text?
The letters have appeared...! However, unfortunately there is a spelling mistake. Most of the logos are misspelled.
Our MaisonAI uses Stable DiffusionXL. I've heard that the accuracy of character input has improved, but it seems like there are still some questionable aspects.
By the way, it was impossible to make Japanese characters appear.
It seems like it can be used as a pattern with text to generate ideas, but it seems difficult to use it as is.
However, it is an "icon generator" that can be used immediately by replacing the words and color scheme instructions, so please take advantage of it!
summary
I think it's great that you can now quickly try generating images without any prompts.
However, there still seems to be some room for improvement.
-
Challenge: Users don't know what to change
- If you can read English, you will be able to pinpoint what needs to be changed, but for those who can't read it, there may still be parts that are difficult to use.
- In the future, I hope to improve the prompt syntax so that you can intuitively understand what needs to be changed, so please look forward to improvements!
- Issue: Cannot support Japanese input
- Also, writing in English itself can be a hurdle for users. You will end up typing in even the slightest difficult words using a translation tool in one hand.
- We will continue to develop this part so that it can be used in Japanese in the future, so please look forward to it!
How to use guide
We have prepared a guidebook here. We will explain what functions it has and how to use it, so please use it as a reference!
