What is ControlNet Pose?
ControlNet Pose is an open-source model for generating images using human poses from a reference image and a text prompt. ControlNet Pose adapts Stable Diffusion with ControlNet, adding a human pose map as an input condition. The model is intended for image generation involving humans, where the pose in an existing image provides guidance beyond the written prompt.
What ControlNet Pose can do
ControlNet Pose combines image-based pose information with text-based generation instructions. The source image supplies the human pose, while the prompt describes the image to generate in the same manner as a Stable Diffusion prompt. Openpose handles pose detection automatically, so the documented workflow begins with an image rather than requiring a separately prepared pose map.
- ControlNet Pose detects human poses in an input image using Openpose.
- ControlNet Pose uses a human pose map to condition Stable Diffusion.
- ControlNet Pose accepts a text prompt alongside the input image.
- ControlNet Pose generates an output image from the combined inputs.
How ControlNet Pose works
ControlNet Pose follows the documented workflow of supplying an image and prompting the model as for Stable Diffusion. Openpose detects the human pose, and ControlNet adds that pose information to the generation process. The Replicate page provides Playground, API and examples, while a GitHub link points to the original model and code. Local execution is also available through Docker.
Who ControlNet Pose is for
ControlNet Pose is relevant to people generating images of humans who want an input image to guide the pose. The model addresses a specific form of conditional image generation: combining a detected human pose with a written description. The source presents other ControlNet models for drawings, edges, depth and other conditions separately, rather than describing those methods as capabilities of ControlNet Pose.
- ControlNet Pose supports generating human images with pose guidance from a reference image.
- ControlNet Pose supports exploring text prompts alongside an image-derived human pose.
- ControlNet Pose supports local experimentation with the open-source model through Docker.
ControlNet Pose pricing and plans
ControlNet Pose uses per-run pricing on Replicate rather than a subscription plan described on the page. The stated approximate cost is $0.15 per run, or 6 runs per $1, with costs varying according to the inputs. The source also identifies the model as open source and says users can run the model on their own computer with Docker.
ControlNet Pose strengths and limitations
ControlNet Pose provides a concrete additional control for Stable Diffusion: a human pose detected from an image. Automatic pose detection and the choice between hosted execution and local Docker execution are practical strengths. On Replicate, the model runs on Nvidia A100 (80GB) GPU hardware, and predictions typically complete within 108 seconds. The page cautions that prediction time varies significantly depending on the inputs.
- ControlNet Pose hosted runs incur a cost that varies with the inputs.
- ControlNet Pose prediction times are not fixed.
- ControlNet Pose requires Docker for the stated local execution option.
ControlNet Pose FAQ
What is ControlNet Pose?
ControlNet Pose is an open-source image generation model that adapts Stable Diffusion to use a human pose map from an input image alongside a text prompt. Openpose detects the pose automatically.
Is ControlNet Pose free?
ControlNet Pose costs approximately $0.15 per run on Replicate, with the actual cost depending on the inputs. ControlNet Pose is also open source and can run on a user's own computer with Docker.
How do you use ControlNet Pose?
ControlNet Pose takes an input image and a text prompt written as for Stable Diffusion. Openpose detects the human pose, and the model uses the resulting pose map together with the prompt to generate an output image.
Can ControlNet Pose run locally?
ControlNet Pose can run on a user's own computer with Docker, according to the model page. The page also links to the original model and code on GitHub and offers hosted execution through Replicate.
How long does ControlNet Pose take to generate an image?
ControlNet Pose predictions typically complete within 108 seconds on Replicate, where the model runs on Nvidia A100 (80GB) GPU hardware. The page states that prediction time varies significantly based on the inputs.
Does ControlNet Pose have an API?
ControlNet Pose has an API option on its Replicate model page, alongside Playground and examples. The documented model workflow uses an input image and a text prompt, with Openpose detecting the human pose automatically.
From the community
No reviews yet
Be the first to review this tool. What you write helps the next person choose.




