I'd recommend an ordinal scale.
A nominal scale just labels classes without a < or > relation. Using this you would be limited to "immersive" and "not immersive" type of answers. This runs into the danger that experimental subjects might classify all your conditions as immersive or non-immersive respectively. That is, you might get little information.
An ordinal scale is a set of ordered levels with a < or > relation. A 5-point Likert scale like "not immersive at all / a little immersive / moderately immersive / highly immersive/complete immersion" would be an example. In case immersiveness is binary, it would show as all answers falling into only two levels of your scale. That is, you do not lose anything. As an experimental subject, I'd understand such instructions and a rating scale. This is what I'd take.
An interval scale requires a quantitative comparison. For example, the difference between 350ms and 370ms reaction time is as large as between 200ms and 220ms. For immersiveness I do not know of such a scale. Therefore it does not seem an option for you.
There is a further rational scale. But it is even more demanding.
A nice explanation of scales is on https://www.mymarketresearchmethods.com/types-of-data-nominal-ordinal-interval-ratio/