The concept is valid, which is to seek a result based on a reference. But of course it needs to be broken down into measurable points.
For example: the acceptance test will be like:
**Achieve the same exact effect as in screen x in app y in terms of:**
**1\. Dimming effect hex value is #9999**
**2\. Area affected by dimmed light is 70% of the screen**
**3\. Colorful items are in black and white...etc**
The main goal of any acceptance test is to help the team to mark the task or feature done. So it has to be measurable not subjected to personal opinions.