• Login
    • Search
    • Categories
    • Recent
    • Tags
    • Users
    • Groups
    • Rules
    • Help

    Do more on the web, with a fast and secure browser!

    Download Opera browser with:

    • built-in ad blocker
    • battery saver
    • free VPN
    Download Opera

    Extra Unidentifiable Foot in Generated Silhouette

    Bugs
    image pr analysis
    1
    1
    1719
    Loading More Posts
    • Oldest to Newest
    • Newest to Oldest
    • Most Votes
    Reply
    • Reply as topic
    Log in to reply
    This topic has been deleted. Only users with topic management privileges can see it.
    • xionix4
      xionix4 last edited by leocg

      Hi everyone,

      I’m reporting a bug I encountered while using the AI image generation feature in Opera. I’m using Windows 11 and Opera One(version: 114.0.5282.235).

      Issue:

      When I used the following prompt: "create a new image of a silhouetted man laying on his back on a giant blue sphere," the generated silhouette had an extra foot that wasn’t present in the prompt. This extra foot seems to be invisible to the system both when confronted after generation as well as after uploading the image to refresh context. Aria argued that it didn't exist and that I was confused because of the lighting. I continued to push back until we agreed to generate a bug report. Part of this text was written by Aria and later edited by me to fill in recommended information.

      Expected Behavior:

      The system should accurately identify and depict all elements of the silhouette in the generated or uploaded image, including the man’s [extra] foot.
      The silhouette should only reflect the elements present in the original prompt, without adding any extraneous features [e.g. head-foot].

      Actual Behavior:

      The system correctly identifies one foot in the silhouette, but it erroneously misses an additional foot that appears to be coming out of the man's head and concludes the user is incorrect.
      This additional foot is not present in the original prompt, creating a surreal and confusing representation.
      The additional foot seems to be invisible to the system when the image is analyzed, indicating a potential flaw in the detection algorithm.

      Summary
      This situation highlights a potential issue with how the system interprets and generates silhouettes [images in general?], especially regarding distinguishing between actual elements and those that should not be included.

      Screenshots and relevant image(s):

      25d5f22e-6f3e-488c-be8c-90637a6a305b-image.png

      P.S. Aria has also since informed me that it should have been spelled "lying" rather than "laying." I included the original wording for potential reproducibility despite the blow to my pride. 😛

      I hope this helps!

      Reply Quote 1
        1 Reply Last reply
      • First post
        Last post

      Computer browsers

      • Opera for Windows
      • Opera for Mac
      • Opera for Linux
      • Opera beta version
      • Opera USB

      Mobile browsers

      • Opera for Android
      • Opera Mini
      • Opera Touch
      • Opera for basic phones

      • Add-ons
      • Opera account
      • Wallpapers
      • Opera Ads

      • Help & support
      • Opera blogs
      • Opera forums
      • Dev.Opera

      • Security
      • Privacy
      • Cookies Policy
      • EULA
      • Terms of Service

      • About Opera
      • Press info
      • Jobs
      • Investors
      • Become a partner
      • Contact us

      Follow Opera

      • Opera - Facebook
      • Opera - Twitter
      • Opera - YouTube
      • Opera - LinkedIn
      • Opera - Instagram

      © Opera Software 1995-