Horror movie alternate endings follow a pattern that the genre’s own production record makes visible, and Obsession is the clearest diagnostic case available right now.
Director Curry Barker filmed two significantly different versions of his ending before arriving at the theatrical cut through elimination. Both discarded versions represent recognisable horror archetypes. Both were rejected.
The Ending That Wasn’t, Obsession’s Post-Production Reinvention
Barker originally shot what he called a “Romeo and Juliet” ending: protagonist Nikki also dies by suicide, mirroring her obsessive subject’s fate. Tragic symmetry, clean closure, thematic logic intact.
It was abandoned after collaborators, including his own father, gave him consistent feedback. According to Dread Central:
People kept telling him that it was ‘way more disturbing if she just survives this thing,’ advice he ultimately agreed with.
The second discarded version took a different route entirely. As Dread Central reported, “There was once an earlier version of Obsession where, like many A24 movies that Barker grew up watching, the writer-director kept us guessing whether everything was in Bear’s head or not.” Interpretive openness: the audience gets to decide; the horror becomes optional. Also abandoned.
Beyond the endings, additional photography, editorial adjustments, rewrites focused on character clarity around Cooper Tomlinson’s role, work balancing horror and comedy elements, and a reported uncredited editorial pass all shaped the final film. Obsession did not arrive at its ending through a single authorial decision, but through elimination.
Not Sure What to Watch?
Answer 4 questions, get curated horror movies to watch.
Horror’s Third Act Problem
This is where the two camps of existing coverage stop talking to each other, and where the actual argument lives.
Den of Geek has catalogued 32 films whose endings changed via test-screening feedback, including horror entries like The Descent and Misery. Comic Basics documents reshoot-driven changes across genres, such as Get Out, World War Z, as a useful inventory.
Fiction Horizon tracks horror films that traded darker endings for softer ones, establishing a directional pattern. Meanwhile, Counter Arts and Cultured Vultures have both observed independently that horror third acts tend to collapse structurally, writing from the audience side, with no production evidence underneath the observation.
The directional pattern in the reshoot record, horror endings consistently moving away from symmetrical closure toward something more destabilising, maps directly onto the structural observation that horror’s power depends on what remains unresolved. Closure, whether tragic or ambiguous, can defuse exactly what a horror film has spent ninety minutes building. Test screenings catch this because audiences feel the deflation in real time; the writers’ room caught it in theory first and got it wrong anyway.
Prestige and elevated horror are absent from the reshoot listicles, the A24 wave, Hereditary, Midsommar, despite being the tier where post-production ending revision is most discussed. Barker cited A24 films as the template for his discarded ambiguity ending, which means the influence ran in both directions: prestige horror taught filmmakers to reach for ambiguity, and test-screening feedback pushed at least one of them back toward something more concrete and more disturbing. That’s a significant blind spot in the existing catalogue work.
For comparison: the Michael biopic required a full third-act reshoot due to legal constraints, not genre logic. That distinction matters. When horror films find their endings in post-production, the driver is not external pressure. It is something the genre itself seems to demand.
What Obsession’s Final Ending Tells Us About Horror’s Real Grammar
Nikki lives. That is the ending. And according to everyone who weighed in during post-production, it is the most disturbing outcome available, more disturbing than her death, more disturbing than ambiguity about whether any of it was real.
The reason is structural. Death provides resolution, however tragic. A reality-ambiguity ending provides interpretive distance, and the viewer can decide events were not real and exit the horror cleanly. Neither is equivalent to a survivor whose survival confirms something irreversible and wrong. The ending that works is not the most logically complete or thematically symmetrical; it is the one that keeps the horror metabolically active after the credits. Scripted resolution, however skillfully constructed, tends toward the wrong kind of closure for the genre.
Obsession discovered its theatrical ending through a process of elimination that the reshoot record suggests is endemic to prestige horror specifically. As elevated horror continues to dominate critical conversation, the process by which these films find their endings deserves more sustained attention than listicles and production footnotes have provided. The alternate endings that don’t survive are not failures. They are the genre’s diagnostic record.