日本に特化した精度 (Unable to generate Japanese specific imagery)

[NorySkywalker]

お世話になっております。

ベータ版から Adobe Firefly を使っていますが、日本に特化したコンテンツをつくる際、なかなかうまく行かないことが多いです。

特に、『スイカ割り』『盆踊り』等、その国の文化、民族、伝統に関することですが、これは、AIが対応できないのか、スイカ割りや盆踊り等に関するデータが少ないために、AIが表現できないということでしょうか？

AIの精度を上げていくためには、やはりその国の文化や民族、伝統等に関する画像生成を行う必要があるのでしょうか？

ただ単に、私のプロンプトに問題があるだけなのかもしれません。

どなたかお分かりの方はいらっしゃいますか？ 

[Preran · Accepted solution]

Thank you for your feedback. 

Copying from this article: https://www.adobe.com/sensei/generative-ai/firefly.html#:~:text=Where%20does%20Firefly%20get%20its,c....

The current Firefly generative AI model is trained on a dataset of Adobe Stock, along with openly licensed work and public domain content where copyright has expired.  

As Firefly evolves, Adobe is exploring ways for creators to be able to train the machine learning model with their own assets so they can generate content that matches their unique style, branding, and design language without the influence of other creators’ content. Adobe will continue to listen to and work with the creative community to address future developments to the Firefly training models.

It is possible that Adobe Firefly doesn't have sufficient data to handle requests like the ones that you refer to in your post. However, I would suggest trying keywords that are less-specific for now. For example, you can substitute the names of flowers with something that is more generally available across the world.

Let me know if this helps.

[NorySkywalker]

Understood.
thank you.