What I would say are real limits to creativity are the arbitrary and absurd evaluation metrics we use. Publish or perish paradigm encourages quick work rather than good work, which correlates with a decrease of novelty and substance. Makes the process more incremental. It also encourages railroading as exploration requires more time.
The review process also exacerbates these problems, as the system is not actually aligned with the goals. They are arbitrary noisy filters rather than quality filters, and we have substantial evidence that, putting it nicely, reviewers are good at identifying bad works but not good at identifying good works (i.e. default to reject). This causes one to continually rework a project rather than moving on or even making it better, as the feedback generally isn't about how to make the work better and each set of reviewers wants to add more and more.
The pursuit of novelty is also not only fundamentally anti-scientific but the basis of an existential crisis. To get funding or to pass review, we must be creative in convincing reviewers that we are clever while ideas generally far more obvious after they are revealed to you (regardless of lack of prior works). The need to innovate also prevents replication, which is the basis of science! It is far more difficult, often nearly impossible, to get funding or even approval to replicate works. Science is the process of ruling things out, not confirming things. Each replication has its own set of noise and biases, and this stochastic nature helps reveal hidden confounders. It is also quite essential to the learning process and failing to perform this task results in information being passed like a game of telephone.
But by nature, science is anti-authority and highly creative. You _must_ challenge pre-existing notions (trust but verify) and you must think outside the box. Though I do not blame anyone for missing this given the structures and incentives (shackles) the bureaucratic system has imposed upon us.