Skip to main content
Participant
May 8, 2024
Open for Voting

Custom AI for Product Insertion in Photoshop

  • May 8, 2024
  • 4 replies
  • 1259 views

Hi Everyone,

 

I'm reaching out to the community with an idea for a new AI feature in Photoshop.

 

Problem: Currently, integrating custom product images seamlessly into client photographs can be a time-consuming process.

 

Solution: I propose the development of a custom AI tool specifically trained on user data. This tool would allow users to:

  • Upload high-resolution images of their own products.
  • Train the AI on these images.
  • Utilize the trained AI to seamlessly insert the product images into client photographs, ensuring accurate details and material textures.

Benefits:

  • Improved efficiency and workflow for product-based businesses.
  • Enhanced realism and accuracy during product integration.
  • Potential for a wider range of creative applications.

I believe this feature could be incredibly valuable for product photographers, graphic designers, and businesses like mine (World of Stones). I'd love to hear your thoughts on this idea and gauge potential interest from the community.

 

Would a custom AI tool for product insertion be helpful for your workflow? Share your comments and feedback below!

 

Thanks,

Gourav Vishwakarma

World of Stones

4 replies

D Fosse
Community Expert
Community Expert
May 9, 2024

No, you're absolutely right, it will never meet those expectations.

 

There's a perfect metaphor for AI from Greek mythology: Pandora's box.

 

"It contained all the evils of the world, which were released when Pandora opened the box. However, it also contained hope, which remained inside the box. Symbolically, the box represents the curiosity and desire for knowledge that can lead to both negative consequences and positive outcomes. The evils inside the box can be seen as the challenges and difficulties of life, while the hope represents the optimism and resilience to overcome those challenges."

c.pfaffenbichler
Community Expert
Community Expert
May 9, 2024

Alas, can a »"fix everything" button« ever meet the expectations of the people asking for it? 

 

It seems character consistency is an issue being worked on with current generative AI imagery software (and I guess this may have implications for product images as well) but when the output is an image instead of a fullblown 3D-model then any changes of larger or smaller details risk also causing changes of things one did not want to change. (edited)

D Fosse
Community Expert
Community Expert
May 9, 2024

In the old days (ten years ago) we used to joke about this, the magic "fix everything" button where all the user had to do was push a button and sit back. And we all laughed.

 

And now it's here <shakes head sadly>

 

The only thing  I want to ask is, please stop using the word "creative" in this context. That's what gets me every time.

 

😕😕

c.pfaffenbichler
Community Expert
Community Expert
May 9, 2024

This seems potentially unreasonable to me at current. 

But maybe I misunderstand the process you propose … what would be the ultimate input you want to use to create a new image? One photohraph for the background and one for the object? 

 

Edit: With product imagery details can matter a great deal. I think leaving that up to AI instead of creating actual photographs (of products or prototypes or models) or up-to-date 3D models could lead to output quality issues.