Pretty good.
It got all of the details of the outfit correct, including necklace and hairfeathers, even though the base image was kinda low res.
The artistic quality of the output image is also quite nice.
The hands are garbage and the legs are a total loss, but that's a limitation of most models even today. a few re-runs should yield a workable result. (I should also note that the hands and legs are not visible in the input image - so that might be the cause)

Imagine- Now anyone can create an entire world worth of ai images from only a single starting image. no loras. no training. no "hq character collection download lora training [virus free].exe". o n l y a s i n g l e i m a g e