The current Leonardo image workflow

Leonardo’s current image tools separate fast ideation from more controlled creation. The home prompt can choose a model automatically, while Image Creation exposes model, dimensions, image count and other model-dependent settings. Flow State is useful for rapidly exploring directions, and newer Image Guidance workflows can combine text with multiple image references. Settings and supported controls vary by model, so the creative brief should remain more permanent than any one interface.

1. Start with the visual rules

If those rules are still vague, return to the Mood Board Guide. Generating more images will not solve an undefined direction.

2. Use Flow State for exploration, not final authority

Leonardo currently positions Flow State as a rapid ideation stream. It can help reveal compositions, lighting ideas and stylistic directions quickly. Use it to compare possibilities, then move the strongest direction into a controlled generation workflow rather than treating every attractive result as a new brand direction.

3. Choose the model by the job

Do not preserve old universal recommendations such as one “best” model or one fixed quality setting. Leonardo’s available models and controls evolve. Decide whether the job requires general image generation, stronger reference understanding, a specific aesthetic treatment or editing, then use the model picker and current capability labels to make the technical choice.

4. Use references to preserve decisions

Newer Leonardo workflows support contextual image references alongside text instructions, while some older guidance modes remain model-dependent. A reference should answer a specific question: which subject identity, composition, palette, environment or visual relationship must survive?

5. Generate a small comparison set

1. Write one clear production prompt from the approved brief.
2. Generate enough variation to compare the important decision, not enough to restart the project.
3. Choose a winner against explicit criteria such as subject consistency, composition, lighting and brand fit.
4. Save the winning version before experimenting further.
5. Use editing or reference-based refinement to correct the specific failure rather than rerolling blindly.

6. Edit the winning direction

Leonardo’s current viewer includes editing workflows that can accept direct instructions and additional references on supported models. Use editing when the overall image is right but a defined element needs correction. If the composition itself is wrong, regenerate from the brief instead of repeatedly patching a weak foundation.

7. Organize for reuse

Keep the relationship between brief, prompt, references and final asset visible. Leonardo’s Library and Collections can help organize generations, but your project should also preserve meaningful filenames, approved exports and the source/reference record needed for later brand work.

Common failures

• Starting in Leonardo before defining what the project should look like.
• Changing model, palette, references and composition simultaneously, then not knowing what improved the result.
• Using another creator’s work as a reference without considering rights, provenance or how closely the final output reproduces protected material.
• Building a permanent training rule around a temporary interface setting.
• Treating motion as an afterthought instead of planning the sequence in the storyboard first.

Official current references

Leonardo’s official help center currently documents Image Creation, Flow State, Image Guidance and its newer contextual editing/reference workflows. Use those current docs when a model-specific control matters; use this page for the durable creator workflow that connects the tool to your brand.