Copy link to clipboard
Copied
Hello all
I`m using latest phothsop beta v. 26.5 on Windows - selecting the people options it`s seems like a game changer - great work for your team to develop this.
However I`ve noticed a kind of a problem here is a description how to reproduce :
1. Choose Object selection tool
2. Select people - choose a person and the garment e.g Upper Clothes
3. Click generative fill
4. Upload a picture of upper clothes
5. Click generate
Accual result - not the same picture is used which was attached - picture is similar but not the same.
Expected result - Uploaded garment eg Upper Clothes should be used - not the similar one
Is there possibility to force to use the attached picture ?
Thanks
Copy link to clipboard
Copied
Thank you @Web_Style3248
Testing here and it seems to be working as expected. One thing to make sure to do is select Cloud to get better results. Give this a try and let us know if you also see better results.
Thanks,
Cory - Photoshop Product Manager
Copy link to clipboard
Copied
Hello @CShubert
Thank you for your reply - I did a try with detailed setting but that doesn`t work for me. I recorded loom video with the details pls have a look:
https://www.loom.com/share/6cde48a53fb54ca7a40a686903808e04
Best regards
Copy link to clipboard
Copied
Hi @Web_Style3248
Currently our Reference Image (beta) for Generative Fill does not make an exact match.
The problem is not with the Object Selection tool.
I got similar results as you did (the best being selecting the model's shirt and midriff via the Object Selection tool).
Is this a common scenario for you?
I'd love to see more examples of how you use Reference Image.
If you're interested in testing out the latest changes, I could invite you to a private test group. I'll send you a direct message. 🙂
Thank you,
Meredith
Photoshop Product Manager for Reference Image
Copy link to clipboard
Copied
@MeredithStotzner the functionality is great but the result I`m getting are still not to be used in a live cycle.
Please invite my to the group and I will share few movies with examples to support this functionality