{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreiherbdbpde777y6vd3yflhmcudqfewiqvvm4ack2ywpqvf3hmokyy",
    "uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3mgwmllcre6a2"
  },
  "path": "/t/possible-regression-in-reference-image-based-image-edit-regenerate-behavior-since-around-mar-12-4-00-pm-jst/1376557#post_3",
  "publishedAt": "2026-03-13T07:02:33.000Z",
  "site": "https://community.openai.com",
  "textContent": "Additional observation:\n\nI also observed cases where I cropped and enlarged only the area that needed correction, then attached that partial image and requested a localized fix.\n\nEven in those cases, the output did not preserve the input image ratio. A 2:3 cropped input could come back as a 1:1 square output instead.\n\nThis suggests the issue is not limited to full-frame reference edits. It may also affect partial or localized edit flows, where non-target areas are effectively discarded and the result is re-emitted on a fixed square canvas.\n\nAt least from what I can observe externally, the current behavior does not appear to preserve source dimensions reliably. So far, I have only observed outputs collapsing into fixed formats such as 1024×1536, 1536×1024, or 1024×1024, rather than preserving the original input frame.\n\nI am not claiming this as a confirmed internal mechanism. This is simply the repeated pattern I have observed from the outside.",
  "title": "Possible regression in reference-image-based image edit/regenerate behavior since around Mar 12, 4:00 PM JST"
}