Phase B shipped — chaining shots for model/outfit consistencyFollowing up on Phase A (jewelry design locking).Problem: Even with jewelry d

Problem: Even with jewelry design locked, each shot in a showcase (closeup, mid-length, extreme closeup) was still generated independently — so the model's face, outfit, and scene would sometimes shift between shots within the same showcase.
Fix: The mid-length generation now references the closeup's output alongside the original photo, instead of just the original alone. This should carry the same model, outfit, and scene across shots, while jewelry design stays anchored to Phase A's locked description.
Status: Just shipped, running test showcases now to confirm face/outfit consistency actually holds before calling this validated.
jewelviz.com