Bria Increase Resolution API

Name: Bria Increase Resolution
Brand: Bria
Price: 0.04 USD
Availability: InStock

Image·bria·by Bria

Upscale an input image with BRIA Increase Resolution, preserving the original content while improving image sharpness and size.

Try it now →

API reference

Runtime (p50): 20s
Estimated price: $0.04

Call the API

prediction.sh

curl -X POST \
  -H "Authorization: Bearer $EACHLABS_API_KEY" \
  -H "Content-Type: application/json" \
  --data '{
    "model": "bria-increase-resolution",
    "version": "0.0.1",
    "input": {
        "image": "https://cdn-us.eachlabs.ai/defaults/c7e7203a371345f69ab2545e7ab2f752.png",
        "desired_increase": 2,
        "visual_input_content_moderation": true,
        "visual_output_content_moderation": true
    },
    "webhook_url": ""
}' \
  https://api.eachlabs.ai/v1/prediction/

Documentation8 sections

Overview
Bria Increase Resolution Overview

Bria Increase Resolution is an image upscaling model from Bria that enlarges an input image while preserving the original content, structure, and visual intent. It is designed for users who need sharper, higher-resolution outputs without introducing a full regeneration step, which makes it well suited for product imagery, marketing assets, and production workflows on each::labs.

Its primary differentiator is that it uses a dedicated upscaling approach rather than recreating the image from scratch, so the output is focused on retaining the source image while improving clarity and detail. Bria’s documentation describes the model as part of its AI image editing and super-resolution pipeline, with TensorRT engine builds available for 2x and 4x upscaling.
Capabilities
Capabilities
- Upscales an input image while preserving the original content.
- Provides dedicated 2x and 4x super-resolution options.
- Fits production pipelines that require image enlargement without scene regeneration.
- Supports deployment in a GPU-accelerated TensorRT environment.
- Works as part of Bria’s broader AI image editing and super-resolution tooling.
- Helps improve apparent sharpness, texture clarity, and overall image size for downstream use.
- Is suitable for preserving composition in commercial or catalog-style images.
Use cases
Use Cases for Bria Increase Resolution

Creators: A photographer can enlarge a finished shot for print while keeping the original composition stable. Example prompt: “Upscale this portrait to a higher resolution while preserving skin texture and framing.” This uses the model’s content-preserving upscaling behavior.

Marketers: A brand team can prepare a campaign image for larger placements without re-shooting or re-editing. Example prompt: “Increase resolution of this product banner for a billboard layout, keeping the product shape and colors unchanged.” That relies on the model’s dedicated enlargement workflow.

Designers: A designer can refine web or mockup assets before export to print-ready sizes. Example prompt: “Upscale this UI illustration for a brochure cover while preserving edges and typography structure.” The model is useful when fidelity matters more than generation.

Developers: A platform team can integrate the Bria Increase Resolution API into a media pipeline for automated enhancement of user-uploaded images. This is useful when the workflow needs consistent super-resolution on a GPU-backed deployment.
Tips & tricks
Tips and Tricks

Use Bria Increase Resolution when the source image is already compositionally correct and only needs quality improvement. Because the model is meant to preserve the original image, the best results usually come from clean inputs with minimal compression artifacts and as much source detail as possible.

For prompt-aware workflows in the broader Bria image-to-image family, keep instructions specific and minimal so the model does not drift from the source. If your pipeline exposes size or output-resolution controls, choose the smallest upscale that meets your target, then compare against a higher tier only if the extra detail is worth the added cost and processing time.

Example prompts: “Upscale this product photo while preserving the original label and lighting.” “Increase resolution of this portrait without changing facial features.” “Enhance this storefront image for large-format print while keeping the scene intact.”
Technical spec
Technical Specifications
- Model type: Image super-resolution / image-to-image upscaling.
- Upscale modes: 2x and 4x TensorRT engine builds are available for the Increase Resolution pipeline.
- Input: A source image for enhancement and enlargement.
- Output: An upscaled image that preserves the original content rather than generating a new scene.
- Architecture/runtime: TensorRT engines tied to specific GPU, CUDA, and TensorRT runtime combinations in BYOC deployment.
- Processing expectations: Exact timing is not publicly specified in the available documentation; runtime will depend on deployment environment and GPU support.
- Format support: The public sources reviewed do not list file-format constraints for this model specifically.
Things to be aware of
Things to Be Aware Of

Bria Increase Resolution is not designed to invent new scene content, so inputs with severe blur, heavy noise, or missing detail may not recover cleanly. Results are also dependent on the deployment environment because the released engines are tied to specific GPU, CUDA, and TensorRT combinations.

Users sometimes expect an upscaler to fix every flaw in the source image, but this model is best treated as a resolution and clarity enhancer. For the best outcome, start with the cleanest possible image and avoid using it as a replacement for major retouching or generative editing.
Key considerations
Key Considerations

Bria Increase Resolution is best when you need to improve image sharpness, enlarge artwork, or prepare assets for higher-resolution delivery without changing the scene. It is especially useful when content fidelity matters more than creative variation, because the model is positioned as a dedicated upscaler rather than a generative editor.

For deployment, the biggest practical consideration is runtime compatibility. The available Hugging Face assets are TensorRT engines tied to a specific GPU and CUDA/TensorRT stack, so integration on each::labs should account for environment matching. If your workflow needs flexible scene editing or new visual content, a broader image generation or editing model may be a better fit than Bria Increase Resolution.
Limitations
Limitations

Available public documentation confirms 2x and 4x TensorRT engine variants, but it does not provide a full public parameter sheet for every integration setting. The model is also constrained by its role as a super-resolution tool: it preserves the original image rather than transforming it into a new composition.

Because the engines are environment-specific, portability is limited compared with fully abstracted hosted APIs. It is therefore strongest for controlled production use, not for workflows that require broad runtime flexibility or creative image generation.