Black Forest Labor (BfL), the startup, which was founded by the creator of the favored stable diffusion model, began a brand new image generation model called Flow.1 context. This model not only generates and processes photos, but in addition enables users to alter them with each text and other images.
The company also announced its latest BFL playground, where people can try the BfL models before they could be removed to corporate applications.
BfL published two versions of the model: Flux.1 Context (Pro) and Flux.1 Context (max). A 3rd version, Flux.1 Context (DEV), will probably be available on private beta. Both the professional and the Max versions are actually on platforms akin to B. available CheatPresent FreepikPresent Leaf tricksPresent Open type And Leonardoai. With these models, Enterprise creative teams and other developers can edit images with precision and faster pace.
Flux.1 Context can perform an in-context generation. This implies that the model could be generated from a reference or situation. It doesn’t generate latest from scratch.
The company said in a single post X That 4 things make context “special”:
- Character consistency and preserving elements across scenes
- Local processing that goals at certain parts without influencing the remaining “
- Style reference that produces scenes in existing styles, and
- Minimal latency
Developers can test applications and play with the models on the BFL playground before accessing the total BFL -API.
The pro and Max models
Companies can use the Pro version for fast and iterative processing. Users can enter each text and reference images and make local changes. The company said context (pro) “as much as a size, faster than previous models at the newest state-of-the-art” and is considered one of the primary models with which the processing of several curves could be processed.
On the opposite hand, Flux.1 context (max) is the faster version with maximum performance. The company said that it could be more of the request, makes typography readable and is consistent in processing without compromises.
Of course, many other image generation models also can generate photos from uploaded files. Midjourney's AI image editor can use a reference picture after which edit certain regions of it. Also Adobe's Firefly, to which many individuals use the Adobe popular image and video platforms.
Flux.1 Context (Dev), the third version of the models context family, is an open model with 12 billion parameters.
Generative flow
BfL said river.1 Context is a river model that offers it more flexibility to perform the tasks mentioned above.
Flow models learn from a continuous data flow and define a path between loud data and useful information. This differs from diffusion, the model architecture, which supports many image and video models Stability AiMidjourney and even Openais sorawhat data “describes”.
BfL said in a blog post that the context models are progress for flow models.
“Flux.1 Context models transcend text-to-image,” said the corporate. “In contrast to previous flow models, which only enable a pure text-based generation, understand and also can understand context models of existing images.
In the text-to-image-benchmark test, BFL claimed the river.1 Context models can compete against other models by way of aesthetics, realism and typography.
Generate interest
BfL published the text-to-image modEl Flux 1.1 Pro in October last yr. It also contained an API for third -party developers to integrate them into their apps.
Thanks to the BFL playground, some users have already began to mess around with the context models and report that they’re impressed.
Of course, it still has to compete with other available image models, especially with those that have been around for several years and have further improved.
