Black Forest Labs has released FLUX.2, their most advanced image generation model yet. This release brings significant improvements in image quality, editing capabilities, and enterprise-grade efficiency. FLUX.2 is available to run on Replicate today.
FLUX.2 comes in three variants:
FLUX.2 [pro] generates images in 6 seconds, or 9 seconds with an input image. It accepts up to 8 input images, and costs $0.015 + $0.015 per input and output megapixel.
FLUX.2 [flex] generates higher quality images, especially typography and fine-grained details. It generates images in 22 seconds, or 40 seconds with an input image. FLUX.2 [flex] accepts up to 10 input images, and costs $0.06 per input and output megapixel.
FLUX.2 [dev] is an open-source, distilled version of FLUX.2 [pro] that we have worked with Pruna.ai to optimize. It generates images in 2.5 seconds, or 5.5 seconds with an input image, and costs $0.012 per input and output megapixels.
FLUX.2 produces realistic image details up to 4MP resolution. The model excels at rendering accurate hands, faces, fabrics, logos, and small objects that other models often miss. This level of fidelity makes it ideal for creative photography, e-commerce product shots, marketing materials, interior design, and image restyling.
FLUX.2 delivers state-of-the-art visual intelligence at low latency and predictable cost. Whether you’re generating new images or editing existing ones, the model handles complex visual problems efficiently, making professional-quality AI accessible even at large image volumes.
One of FLUX.2’s standout features is its ability to reference up to 8 images simultaneously while maintaining consistent identities, products, and styles. This means you can create 50 ad variants using the same face, generate product mockups in any context, or produce dynamic fashion editorials where the model stays consistent across all shots.
FLUX.2 offers direct pose control for exact positioning, hex-code color accuracy for brand guidelines, and the ability to combine visual cues with input images for localized editing. The model handles clean background replacement and object insertion that match perspective, with localized edits that leave the rest of the image intact. All of this is available at any aspect ratio.
The model includes structured prompting for programmatic workflows, 32K prompt tokens for detailed creative direction, and generative expand/crop capabilities for post-generation edits. Speed and quality tiers adapt to your specific workflow needs.
FLUX.2 understands context, tone, and trends well enough to create engaging posts, memes, and editorial visuals from short prompts in a single generation.
FLUX.2 brings several key improvements over its predecessor:
Multi-reference support: Reference up to 8 images simultaneously with improved character, product, and style consistency.
Image detail and photorealism: Higher fidelity on skin, hair, fabric, materials, and hands, with fewer artifacts and more stable lighting. The gap between generated images and real photography has never been smaller.
Text rendering: More reliable rendering of complex typography, infographics, and UI mockups. The model generates clean, legible text for memes and designs from pure text prompts or input images.
Enhanced prompt following: Significantly better accuracy and responsiveness to complex instructions.
World knowledge: The model is more grounded in real-world knowledge, lighting, and spatial logic, resulting in more coherent scenes.
Higher and flexible output resolutions: FLUX.2 can edit images up to 4MP and can be used to resize or extend images.
FLUX.2 is designed for anyone generating professional visual content at scale: marketing teams, product designers, content creators, social media managers, e-commerce brands, agencies, developers building visual applications, publishers, and educators.
Here’s how to run FLUX.2 using JavaScript and the Replicate API:
Or with Python:
FLUX.2 represents a significant leap forward in AI image generation and editing. With its combination of photorealistic detail, multi-reference consistency, and enterprise-grade efficiency, it’s ready to handle professional visual content creation at any scale.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。