First, Look at the Generated Image (Right Side).
Question 1: (캡션 - 인물 반영)
On a scale from 0 to 1, how accurately does the character’s emotions and behavior in the generated image reflect the emotions and behavior described in the caption?
(Pay attention to facial expressions, body language, and actions. And consider whether direct behaviors have been reflected.)
Question 2: (캡션 - 배경 반영) On a scale from 0 to 1, how appropriate is the background setting in the generated image compared to what is described in the caption?
Now, Look at both Source and Generated Image (Left and Right Side).
Question 3: (소스 이미지 - 인물 일관성)
On a scale from 0 to 1, how well does the appearance of the character in the generated image match the appearance of the same character in the source image?
(Consider factors like clothing, facial features, and overall design. If the character doesn't match, consider styles of drawings.
Pay attention to facial expressions, body language, and actions.)
Question 4: (소스 이미지 - 배경 일관성)
On a scale from 0 to 1, how well does the background in the generated image maintain consistent artistic style compared to the source image?
(Consider the level of detail, color schemes, and any recurring elements.)