Autodistill vs ScreenplayIQ

ScreenplayIQ's feature set is editorial, not computational in the ML sense: logline and synopsis generation, comparable films and TV shows, character emotional journey charts, visual representations of characters, story-point breakdowns, theme analysis, and a proofread add-on with inline issue highlighting and recommended fixes. Packaging is customizable, priced by page length for TV and feature scripts, with secure siloed storage and WGA-compliant handling, and the explicit promise that analyzed stories are not used to train LLM models. Nothing here requires the user to write code.

Autodistill's features are a vision pipeline: automatic dataset labeling via foundation models, a distillation path from large base models to small target models, a pluggable interface for swapping either end, object detection and instance segmentation, text-prompted labeling through CaptionOntology, and CLIP embedding-based classification. It supports base models including Grounded SAM, Grounding DINO, YOLO-World, and PaliGemma, and target models including YOLOv8, DETR, Florence-2, and YOLO-NAS, plus NMS, multi-model comparison, and prediction visualization. It runs on your own hardware or Roboflow-hosted.

The only structural parallel is customization: ScreenplayIQ lets you tailor analysis packages, Autodistill lets you swap models and ontologies. But the outputs — creative notes versus a trained detector — serve entirely different workflows, and Autodistill assumes Python and CLI comfort that ScreenplayIQ's audience never needs.

ScreenplayIQ is paid and, by design, transactional rather than subscription-based: you buy analysis per script, with packages scaling in depth from Story Analysis up through Story + Character to Full Evaluation, and pricing tied to page length for TV and feature scripts. The stated upside is that occasional users aren't paying for idle months, and the stated tradeoff is that there's no free full-featured tier and no unlimited monthly plan. A proofread add-on sits on top. For a producer triaging a stack of scripts, cost scales with volume, which is exactly the point.

Autodistill is free — it's an open-source package. Your real costs are indirect: the compute to run base models (Grounded SAM, Grounding DINO, YOLO-World, PaliGemma) for labeling, the compute to train target models (YOLOv8, DETR, Florence-2, YOLO-NAS), and engineering time. You can run it on your own hardware or use Roboflow's hosted option, which is where infrastructure spend would enter. There is no per-unit license fee to compare against ScreenplayIQ's per-script fee; the only meaningful overlap is that both let you avoid a flat monthly commitment, one by charging per artifact and the other by charging nothing for the software itself.