0126테스트

1. Insert the entire head from Image B (face + hair) into the white masked area in Image A. 2. Mask-only interaction: The composited head must exist only within the white masked area. • No interaction outside the mask: no overlap, no blending, no shadow, no color spill, no edge extension beyond the white boundary. • Anything outside the white masked area must remain exactly as Image A. 3. Strict fit: the inserted head must stay fully inside the white area. No overflow beyond the white boundary. 4. Slightly smaller than the mask: Scale the inserted head to be slightly smaller than the white mask area, leaving a small inward margin (about 2–5%) from the mask edges. The head should not touch the white boundary. 5. Seam & background integration (boundary-only): Apply only minimal boundary blending to remove harsh edges. Also fix any unnatural artifacts around the head/face edges (halo, jagged cutout, color mismatch) by doing local, background-aware blending limited to the edge transition region so it matches Image A’s lighting/tone/DOF/noise. • Do not alter the face identity or facial features—only the edge transition region. 6. Fill remaining white area naturally: After inserting the head, any remaining visible white mask background should be blended/reconstructed to match the surrounding background from Image A, removing any awkward seams so it looks natural. • This background fill must stay consistent with Image A (same texture/lighting/noise) and must not modify non-mask areas. 7. Keep the rest of Image A unchanged (background/body/clothes/pose) outside the mask region. 8. Pose & expression: The inserted head should be oriented slightly toward 7 o’clock (down-left). Set the expression to a subtle surprised look: slightly widened eyes, gently raised brows, and a small “oh” mouth—keep it natural and not exaggerated, and do not change identity.

1. Insert the entire head from Image B (face + hair) into the white masked area in Image A. 2. Mask-only interaction: The composited head must exist only within the white masked area. • No interaction outside the mask: no overlap, no blending, no shadow, no color spill, no edge extension beyond the white boundary. • Anything outside the white masked area must remain exactly as Image A. 3. Strict fit: the inserted head must stay fully inside the white area. No overflow beyond the white boundary. 4. Slightly smaller than the mask: Scale the inserted head to be slightly smaller than the white mask area, leaving a small inward margin (about 2–5%) from the mask edges. The head should not touch the white boundary. 5. Seam & background integration (boundary-only): Apply only minimal boundary blending to remove harsh edges. Also fix any unnatural artifacts around the head/face edges (halo, jagged cutout, color mismatch) by doing local, background-aware blending limited to the edge transition region so it matches Image A’s lighting/tone/DOF/noise. • Do not alter the face identity or facial features—only the edge transition region. 6. Fill remaining white area naturally: After inserting the head, any remaining visible white mask background should be blended/reconstructed to match the surrounding background from Image A, removing any awkward seams so it looks natural. • This background fill must stay consistent with Image A (same texture/lighting/noise) and must not modify non-mask areas. 7. Keep the rest of Image A unchanged (background/body/clothes/pose) outside the mask region. 8. Pose & expression: The inserted head should be oriented slightly toward 9 o’clock (left). Set the expression to a subtle happy smile: gently upturned lips, relaxed brows, and natural eyes—keep it soft and not exaggerated, and do not change identity.

1. Insert the entire head from Image B (face + hair) into the white masked area in Image A. 2. Mask-only interaction: The composited head must exist only within the white masked area. • No interaction outside the mask: no overlap, no blending, no shadow, no color spill, no edge extension beyond the white boundary. • Anything outside the white masked area must remain exactly as Image A. 3. Strict fit: the inserted head must stay fully inside the white area. No overflow beyond the white boundary. 4. Slightly smaller than the mask: Scale the inserted head to be slightly smaller than the white mask area, leaving a small inward margin (about 2–5%) from the mask edges. The head should not touch the white boundary. 5. Seam & background integration (boundary-only): Apply only minimal boundary blending to remove harsh edges. Also fix any unnatural artifacts around the head/face edges (halo, jagged cutout, color mismatch) by doing local, background-aware blending limited to the edge transition region so it matches Image A’s lighting/tone/DOF/noise. • Do not alter the face identity or facial features—only the edge transition region. 6. Fill remaining white area naturally: After inserting the head, any remaining visible white mask background should be blended/reconstructed to match the surrounding background from Image A, removing any awkward seams so it looks natural. • This background fill must stay consistent with Image A (same texture/lighting/noise) and must not modify non-mask areas. 7. Keep the rest of Image A unchanged (background/body/clothes/pose) outside the mask region. 8. Pose & expression: The inserted head should be oriented slightly toward 11 o’clock (left). Set the expression to a subtle happy smile: gently upturned lips, relaxed brows, and natural eyes—keep it soft and not exaggerated, and do not change identity.

1. Insert the entire head from Image B (face + hair) into the white masked area in Image A. 2. Mask-only interaction: The composited head must exist only within the white masked area. • No interaction outside the mask: no overlap, no blending, no shadow, no color spill, no edge extension beyond the white boundary. • Anything outside the white masked area must remain exactly as Image A. 3. Strict fit: the inserted head must stay fully inside the white area. No overflow beyond the white boundary. 4. Slightly smaller than the mask: Scale the inserted head to be slightly smaller than the white mask area, leaving a small inward margin (about 2–5%) from the mask edges. The head should not touch the white boundary. 5. Seam & background integration (boundary-only): Apply only minimal boundary blending to remove harsh edges. Also fix any unnatural artifacts around the head/face edges (halo, jagged cutout, color mismatch) by doing local, background-aware blending limited to the edge transition region so it matches Image A’s lighting/tone/DOF/noise. • Do not alter the face identity or facial features—only the edge transition region. 6. Fill remaining white area naturally: After inserting the head, any remaining visible white mask background should be blended/reconstructed to match the surrounding background from Image A, removing any awkward seams so it looks natural. • This background fill must stay consistent with Image A (same texture/lighting/noise) and must not modify non-mask areas. 7. Keep the rest of Image A unchanged (background/body/clothes/pose) outside the mask region. 8. Pose & expression: The inserted head should be oriented slightly toward 4 o’clock (down-right). Set the expression to a subtle tired/strained look: slightly lowered eyelids, gently furrowed brows, and a small pressed-lip or faint frown—keep it natural and not exaggerated, and do not change identity.

1. Insert the entire head from Image B (face + hair) into the white masked area in Image A. 2. Mask-only interaction: The composited head must exist only within the white masked area. • No interaction outside the mask: no overlap, no blending, no shadow, no color spill, no edge extension beyond the white boundary. • Anything outside the white masked area must remain exactly as Image A. 3. Strict fit: the inserted head must stay fully inside the white area. No overflow beyond the white boundary. 4. Slightly smaller than the mask: Scale the inserted head to be slightly smaller than the white mask area, leaving a small inward margin (about 2–5%) from the mask edges. The head should not touch the white boundary. 5. Seam & background integration (boundary-only): Apply only minimal boundary blending to remove harsh edges. Also fix any unnatural artifacts around the head/face edges (halo, jagged cutout, color mismatch) by doing local, background-aware blending limited to the edge transition region so it matches Image A’s lighting/tone/DOF/noise. • Do not alter the face identity or facial features—only the edge transition region. 6. Fill remaining white area naturally: After inserting the head, any remaining visible white mask background should be blended/reconstructed to match the surrounding background from Image A, removing any awkward seams so it looks natural. • This background fill must stay consistent with Image A (same texture/lighting/noise) and must not modify non-mask areas. 7. Keep the rest of Image A unchanged (background/body/clothes/pose) outside the mask region. 8. Pose & expression: The inserted head should be oriented slightly toward 1 o’clock (up-right). Set the expression to a subtle happy smile: gently upturned lips, relaxed brows, and natural eyes—keep it soft and not exaggerated, and do not change identity.

1. Insert the entire head from Image B (face + hair) into the white masked area in Image A. 2. Mask-only interaction: The composited head must exist only within the white masked area. • No interaction outside the mask: no overlap, no blending, no shadow, no color spill, no edge extension beyond the white boundary. • Anything outside the white masked area must remain exactly as Image A. 3. Strict fit: the inserted head must stay fully inside the white area. No overflow beyond the white boundary. 4. Slightly smaller than the mask: Scale the inserted head to be slightly smaller than the white mask area, leaving a small inward margin (about 2–5%) from the mask edges. The head should not touch the white boundary. 5. Seam & background integration (boundary-only): Apply only minimal boundary blending to remove harsh edges. Also fix any unnatural artifacts around the head/face edges (halo, jagged cutout, color mismatch) by doing local, background-aware blending limited to the edge transition region so it matches Image A’s lighting/tone/DOF/noise. • Do not alter the face identity or facial features—only the edge transition region. 6. Fill remaining white area naturally: After inserting the head, any remaining visible white mask background should be blended/reconstructed to match the surrounding background from Image A, removing any awkward seams so it looks natural. • This background fill must stay consistent with Image A (same texture/lighting/noise) and must not modify non-mask areas. 7. Keep the rest of Image A unchanged (background/body/clothes/pose) outside the mask region. 8. Pose & expression: The inserted head should be oriented slightly toward 7 o’clock (down-left). Set the expression to a blank / neutral stare: relaxed brows, neutral mouth (no smile), and unfocused-looking eyes—keep it natural and not exaggerated, and do not change identity.

Prompt placeholder

Prompt placeholder

Prompt placeholder

Prompt placeholder

Prompt placeholder