Look at the image and answer ONLY with a JSON object, no other text, using exactly these keys:
{"animal": "<species in one lowercase English word>", "count": <number of animals visible, integer>, "eye_color": "<one lowercase word>", "coat_pattern": "<one lowercase word>", "posture": "<sitting|standing|lying>", "background_sky": <true|false>}