Originally posted to Huggingface.
Full Flux Tools Guide, including example workflows, live now!
If you're looking to get started using Flux, check out our Guide here.
FLUX.1 Canny [dev] is 12 billion parameter rectified flow transformer capable of generating an image based on a text description while following the structure of a given input image. For more information, please read our blog post.
Key Features
Cutting-edge output quality.
It blends impressive prompt adherence with maintaining the structure of source images based on canny edges.
Trained using guidance distillation, making
FLUX.1 Canny [dev]more efficient.Open weights to drive new scientific research, and empower artists to develop innovative workflows.
Generated outputs can be used for personal, scientific, and commercial purposes as described in the
FLUX.1 [dev]Non-Commercial License.
Usage
We provide a reference implementation of FLUX.1 Canny [dev], as well as sampling code, in a dedicated github repository. Developers and creatives looking to build on top of FLUX.1 Canny [dev] are encouraged to use this as a starting point.
Description
FAQ
Comments (5)
Is this a controlnet? If yes, why is it as big as a full-fledged non-quantized Flux model? Other Flux controlnets are no bigger than a few gigabytes. The Shakker Union Pro controlnet is ~6GB and it includes 4-5 different types of controlnets.
No, this is a kind of Flux Dev model that has been trained to work with masks created with preprocessors such as canny or depth without using any additional ControlNet models.
@13_february Ah, I see. Too bad. This would mean that I can't use it with any model that I like, correct? The base Flux Dev model is good, but lacks the detail and diversity of some of the newer fine tuned models that I like to use.
@mmdd2543 Yes, that's right. However, if we are talking about details, then when using FLux canny, all the details are worked out at the preprocessor stage, based on your sketch or reference image. So, all the details present in the reference will take place in the final image.
@13_february Oh, I didn't know that. Maybe I should give it a shot then! Thanks! 🙂