DeepInfra raises $107M Series B to scale the inference cloud — read the announcement

Bria/

fibo-1.5

Partner

$0.04

/ image

FIBO 1.5 is Bria's JSON-native text-to-image model, distilled to a few sampling steps and post-trained for sharper realism and texture, with structured-prompt control over lighting, composition and camera.

Public

Paper

api

Input

Prompt

Text-based instruction. Use it alone to generate from text, or together with `image` or `structured_prompt` as a refinement command.. (Default: empty)

ImageUpload Image File

Please upload an image file

You need to log in to use this model

Settings

Structured Prompt

{"short_description": "A sleek, futuristic bullet train, pristine and gleaming, speeds through a vibrant, swirling cosmic wormhole. The wormhole is a tunnel of brilliant blues, purples, and greens, filled with distant stars and nebulae. The train is perfectly intact, leaving a trail of light rather than debris, symbolizing a journey through space and time with advanced technology.", "objects": [{"description": "A futuristic bullet train, with a streamlined, aerodynamic design. Its surface is polished and reflective, suggesting advanced materials. It has no visible damage, unlike the reference image.", "location": "center-right, moving towards the center", "relationship": "The primary subject, actively moving through the cosmic wormhole.", "relative_size": "large within frame", "shape_and_color": "Elongated, metallic silver with subtle blue accents", "texture": "smooth, highly reflective", "appearance_details": "Sleek, modern design with glowing blue lights along its sides and front. The windows are dark and tinted, giving it a mysterious yet advanced look.", "orientation": "facing forward and slightly to the left, indicating motion"}, {"description": "The cosmic wormhole, a tunnel of swirling energy and light, composed of blues, purples, and greens. It appears stable and inviting, not destructive.", "location": "center to left, extending into the background", "relationship": "The environment through which the train is traveling, providing a path.", "relative_size": "very large, encompassing most of the background", "shape_and_color": "Tubular, swirling blues, purples, and greens", "texture": "ethereal, gaseous, luminous", "appearance_details": "Filled with sparkling stars and faint nebulae, creating a sense of deep space. The edges are soft and glowing.", "orientation": "extending diagonally from the foreground to the deep background"}, {"description": "A trail of shimmering light and energy emanating from the rear of the bullet train, indicating its rapid movement.", "location": "behind the train, extending towards the right background", "relationship": "Generated by the train's movement, contrasting with the debris in the reference image.", "relative_size": "medium", "shape_and_color": "Stream of white and light blue energy", "texture": "luminous, flowing", "appearance_details": "Suggests speed and advanced propulsion, without any destructive elements.", "orientation": "trailing horizontally behind the train"}], "background_setting": "The vastness of space, visible through the translucent walls of the cosmic wormhole. Distant galaxies, stars, and nebulae are faintly visible, adding to the sense of an interstellar journey. The wormhole itself forms the primary background structure, a tunnel of vibrant, swirling cosmic energy.", "lighting": {"conditions": "luminous, internal glow from the wormhole", "direction": "omni-directional, emanating from the wormhole's energy", "shadows": "soft, minimal shadows on the train, primarily self-cast, due to the ambient cosmic light"}, "aesthetics": {"composition": "dynamic diagonal composition, with the train and wormhole creating leading lines towards the deep background", "color_scheme": "cool complementary colors, dominated by blues, purples, and greens with metallic silver accents", "mood_atmosphere": "futuristic, awe-inspiring, adventurous, and serene", "preference_score": "very high", "aesthetic_score": "very high"}, "photographic_characteristics": {"depth_of_field": "deep, with both the train and the wormhole in focus", "focus": "sharp focus on the bullet train and the immediate wormhole, with a slight blur on the distant cosmic elements to enhance depth", "camera_angle": "eye-level, slightly below the train, emphasizing its grandeur and speed", "lens_focal_length": "standard lens (e.g., 35mm-50mm)"}, "style_medium": "3D render", "context": "This is a concept art piece for a science fiction movie poster or a video game loading screen, depicting advanced space travel.", "artistic_style": "futuristic, vibrant, dynamic"}

A string containing the structured prompt in JSON format. Use one from a previous generation to recreate or refine that image exactly.. (Default: empty)

Resolution

Output image resolution. '4MP' improves detail, especially for photorealism, but adds about 30 seconds of latency.

Aspect Ratio

The aspect ratio of the generated image.

Seed

Seed for deterministic generation. A random seed is used if omitted. (Default: empty, 0 ≤ seed ≤ 2147483647)

Output Type

The desired output image format.

Ip Signal

If true, returns a warning for potential IP content in the prompt.

Prompt Content Moderation

If true, returns 422 on prompt moderation failure. Default: true

Visual Input Content Moderation

If true, returns 422 on visual input moderation failure. Default: true

Visual Output Content Moderation

If true, returns 422 on visual output moderation failure. Default: true

Output

Model Information

FIBO 1.5 is the latest release in Bria's FIBO family, an open-source, JSON-native text-to-image model trained on long structured descriptions rather than short captions. It builds on FIBO with two post-training stages: distillation into a few-step, CFG-free student, and a refinement pass that sharpens realism and texture. Like every Bria model it is trained exclusively on licensed data, so the output is commercially safe to deploy.

Structured prompts

Send a plain prompt and the model writes the structured JSON description for you, returning it as structured_prompt alongside the image. Feed that JSON back in, edited and with the same seed, to change one attribute at a time — camera angle, lighting, a single object — without disturbing the rest of the scene.