Loccra is designed to make local image generation feel like a Mac product rather than a collection of AI components. You choose what you want to make, provide a useful amount of visual direction, and let the installed model work on your device.

Complete the one-time local setup

Loccra needs its local model before Create and Edit can run. During first setup, keep the Mac connected to power and online while the model downloads and verifies. The download is large, so make sure meaningful disk space is available before starting.

When setup is complete, the status changes to Ready. From that point, ordinary generation and editing can work offline. You do not need Xcode, Homebrew, a separate Python installation, or an account for the creative workflow.

If a model download is interrupted, reopen Loccra and continue. The installer is designed to reuse valid files and resumable download data rather than marking an incomplete model as ready.

Choose the destination before the prompt

Format decisions are easier when you know where the image will live. Choose Square for a flexible social or concept image, Portrait when height matters, and Landscape for scenes, headers, and wider compositions.

Start at 640 px while finding the idea. It is a useful balance for iteration. Use 480 px when speed and memory pressure matter most, then move to 1080 px only after the visual direction works. Larger output is not a substitute for a clearer prompt.

Write a prompt the way a photographer would brief a scene

You do not need a long list of model keywords. Give Loccra five useful kinds of information: the subject, the setting, the composition, the light, and the visual character.

A small orange robot sitting beside a quiet Scandinavian lake at sunrise, eye-level composition, natural photography, soft atmospheric light, subtle mist

The first noun phrase establishes the subject. The setting creates context. Composition tells the model where the imagined camera belongs. Lighting creates time and mood. The final visual direction keeps the result coherent.

Use Avoid for visible problems

The Avoid field is useful for unwanted content such as text, watermarks, or common distortion. Keep it short and concrete. It should remove distractions, not become a second contradictory prompt.

Generate locally

  1. Select the aspect ratio, visual style, output size, and one image.
  2. Enter the prompt and optional Avoid text.
  3. Choose Loccra Standard in Create.
  4. Press Generate and follow the existing progress states.
  5. Use Cancel if you want to stop the current job and return to the workspace.

The first generation after opening Loccra can take longer because the local model must load into unified memory. A warm second generation can be faster. Exact time depends on the Apple Silicon chip, available memory, output size, and thermal conditions.

Review the image before adding more words

When the result appears, separate the idea from the execution. First ask whether the subject, framing, and atmosphere are going in the right direction. Then inspect smaller details.

  • If the overall idea is wrong, rewrite the subject or composition.
  • If the mood is wrong, change light and atmosphere.
  • If the image is too busy, remove adjectives instead of adding more.
  • If the result is close, change one phrase and generate again.

Multiple images are useful after the prompt is stable. Beginning with one image makes each creative decision easier to understand and uses less unified memory.

Every successful Create result is saved into Loccra’s local project store and appears in Gallery. Open Gallery to search, filter, and select individual creations. The prompt, style, format, seed, model identifier, and local file reference remain associated with the project.

Quit and reopen Loccra to confirm the result persists. That small check is also a useful way to understand the product: your gallery is local history, not a remote account feed.