About ZenCtrl
ZenCtrl is an open source model for consistent AI image generation from a single input image, published by Fotographer.ai. It regenerates a subject, such as a product, person or food item, from other angles and in different scenes without any additional training, and it can swap backgrounds or clothing.
The vendor positions it as an alternative to LoRA training and ControlNet, requiring one image where LoRA needs dozens and avoiding a training step. It is described as working with control methods such as Canny edge, and Canny and multi view generation are available in the Fotographer.ai web application. Output types include product shots, CG renders, fashion images and illustrations from a single model. The model is distributed through HuggingFace and GitHub, hosted on Baseten, and also reachable through a web app and an API, with a Discord community. The site notes that some control and editing options are available only through the API and the web platform.
It is aimed at professionals in marketing, advertising and creative work who need consistent subject imagery.
The model weights are published openly. Hosted API plans are listed monthly: Basic is $40 for 1,000 credits, Plus is $200 for 5,000 credits and Pro is $400 for 10,000 credits. Enterprise starts at $1,000 per month, sized around expected request volume. For reference, the vendor states its ZenInpaint tool uses 0.04 to 0.05 credits per call.