Ultra-Resolution Text-to-Image Specifications
Last verified: August 2, 2026
Flux Pro V1.1 Ultra Text To Image converts a written scene specification into one image under a fixed ultra-quality profile. Black Forest Labs documents the underlying Ultra model as capable of outputs up to 4MP while retaining prompt fidelity, making it relevant when textures, edges, and small compositional details matter.
Architecturally, the safest public context comes from BFL's open FLUX.1 dev sibling, which the lab describes as a 12-billion-parameter rectified-flow transformer. The proprietary Ultra documentation does not expose a matching parameter count, so that family-level architecture is useful context rather than an exact specification for the hosted Ultra endpoint.
The creator-level advantage is the ability to place more production direction into the initial brief instead of relying on generic quality keywords. BFL introduced Ultra at four times the standard FLUX1.1 pro resolution without sacrificing prompt adherence, giving users more room to inspect fine features in the generated image.
Explore More Text To Image
Verified Run Specifications
One text prompt produces one fixed-profile image with documented time and credit expectations.
Required Input
Text prompt
Output Per Run
1 image
Quality Profile
Ultra quality, fixed
Expected Time
20 seconds
Seed Control
Supported
Constrain the Run to One Required Input
Lock an Iteration With Seed Control
Pre-Budget the Complete Generation Run
Run Flux Pro V1.1 Ultra Text To Image in Four Steps
Four actions take you from a written brief to a downloaded image.
Step 1: Write the complete scene prompt
Describe the main subject first, then add its action, environment, spatial layout, lighting, visual style, materials, and finishing treatment.
Step 2: Set a seed when repeatability matters
Optionally choose and record a seed if you expect to rerun the prompt or compare controlled wording changes.
Step 3: Click Generate
Submit the prompt with the fixed ultra-quality profile. The run is expected to use 36 credits, take about 20 seconds, and produce one image.
Step 4: Review and download the image
Inspect subject accuracy, composition, surfaces, and unwanted artifacts, then download the final output when it meets the brief.
Production Fit Matrix for Single-Image Runs
Use these factors to decide whether a deliberate ultra-quality generation is the right workflow.
| Criterion | Our Tool | Alternatives | Best For |
|---|---|---|---|
| Starting material | Starts from one required written prompt without relying on source media. | Use an image-guided or editing workflow when an existing composition, product, or identity must be preserved. | Creating an original scene from a complete art-direction brief. |
| Quality strategy | Uses a fixed ultra-quality profile, removing the need to select a quality tier. | Choose an adjustable-quality workflow when you need explicit draft-versus-final tradeoffs. | Users who want a final-oriented render rather than a low-quality preview. |
| Iteration economics | A run is expected to require 36 credits and approximately 20 seconds. | A lower-cost or faster model is more suitable for broad exploratory prompt sweeps. | Deliberate prompts with defined subjects, layouts, and visual direction. |
| Variation count | Returns one image per generation for focused review. | Use a batch or multi-output workflow when several candidates are required simultaneously. | Reviewing one detailed candidate before making the next controlled change. |
Choose this workflow when one carefully directed, ultra-quality image is more valuable than rapid batches of exploratory variations.
Flux Ultra Preflight Checks for Prompt Accuracy
Before submitting, verify these four prompt and repeatability conditions.
1. Before you generate, verify the subject hierarchy
Cause: The main subject can lose priority when style, environment, and props appear first or receive equal emphasis.
Fix: Open with the subject, action, and defining attributes; place environmental and finishing details afterward.
Retry: Retry after shortening or reordering the brief rather than resubmitting unchanged text.
2. Before you generate, verify the spatial map
Cause: Conflicting positions, unclear object counts, or multiple camera directions can produce an unstable composition.
Fix: Define foreground, middle ground, and background separately, then state where each important subject belongs.
Retry: Retry after removing contradictions and reducing the scene to one clear viewpoint.
3. Before you generate, verify the detail instructions
Cause: Generic phrases such as high quality do not define the material, edge, texture, or lighting information the image needs.
Fix: Name specific surfaces, visible micro-details, light direction, shadow character, focus behavior, and desired retouching level.
Retry: Retry when the revised prompt contains concrete visual evidence for the detail you expect.
4. Before you generate, verify the seed plan
Cause: Omitting or changing the seed makes controlled comparison between prompt revisions less repeatable.
Fix: Record one seed and retain it while adjusting a single prompt component, such as lighting or camera position.
Retry: Retry with the same seed after one controlled change; switch seeds only when you want a new interpretation.
Frequently Asked Questions
What should I include in a Flux Pro V1.1 Ultra Text To Image prompt?
Use an ordered brief: main subject, action or pose, environment, object relationships, viewpoint, lighting, visual style, materials, color treatment, and final polish. Because the quality profile is fixed, prompt structure is the primary way to direct the image.
Does this workflow accept an image reference?
The documented input for this page is one required text prompt. It does not list image, video, or audio inputs for this workflow, so use an image-guided tool when preserving an existing visual source is essential.
Can I change the output quality setting?
No editable quality choice is documented. The page uses a fixed ultra-quality profile, so you can concentrate on the prompt and seed rather than choosing between multiple render tiers.
Is this model suitable for rapid idea batching?
It is better aligned with deliberate single-image generation. One run returns one image and is expected to require 36 credits and about 20 seconds, so a cheaper or multi-output workflow may fit broad early-stage exploration better.
What does the seed control do?
The platform supports seed entry, and BFL documents the model seed as an integer used for reproducible generation. Reuse a recorded seed while changing one prompt detail at a time to make comparisons more controlled.
Will every generated image be exactly 4MP?
No exact pixel dimensions are stated in the supplied tool configuration. BFL documents the underlying Ultra model as supporting images up to 4MP, while this page specifically guarantees a fixed ultra-quality profile rather than a stated resolution.
Can I use a negative prompt or automatic prompt enhancement?
Neither capability is part of the documented workflow for this page. Put the required subject, composition, style, and cleanliness instructions directly into the main prompt using positive, concrete language.
Can I use the generated image commercially?
The supplied tool configuration does not state commercial-use or ownership terms, so review the terms attached to your account and intended use before publishing. In the United States, commercial permission and copyrightability are separate issues: Copyright Office guidance says human-authored elements can qualify for protection, while purely AI-generated material or prompt-only control may not.