How do Key Prompts work?
A Key Prompt keeps all of your work to generate a specific unit of media together. The text, yes, but also the model it was written for, the parameters it generates with, the reference images it points at, and when part of a Brief, the context in which it is contained.
Generating with Key Prompts
Section titled “Generating with Key Prompts”You rarely write one from scratch. A shot on your project’s shot list is parsed into a Key Prompt by shot composition, already knowing the world from your brief and the craft from its Skill. You then dial it in rather than compose it.
Key Prompts can contain image, video and audio prompts, which helps when you’re looking to create an image to be used in a video along with audio for that clip. While it’s common to use a prompt for one specific media, just know that you can group these media together for one shot if necessary.
The Prompt Agent does the heavy lifting, in the Studio using natural language conversation. It helps you describe what you’re after to the model selected, sets the parameters, reads the references you have given it, and draws on our catalog of styles, cameras, lenses and looks. You can overrule any of it. The Prompt Agent can also ensure references are worked into your prompt correctly for any given model.
Each Key Prompt has its own Library, with the compiled prompt, a cover image, the grid of reference images, and the image, video and audio library of everything generated from it. Status and gallery visibility live there too, so a prompt can be finished, working, or kept out of your gallery.
Copy for platform provides you a means to use Lorvara to generate prompts meant for outside generative tools. Within the editor, you are given a menu to select the model for which the prompt is destined to be used. Note that while we can pass media to Midjourney (copy & paste workflow), that same copy/paste workflow does not support adding references to most other models. Image upload must become manual at that time.
There is no cap. Unlimited Key Prompts on every plan. Credits meter the model work (editing a prompt with the Prompt Agent, and generating media from our internal models). Creating a prompt yourself costs nothing; creating one with the Prompt Agent costs credits, and the app shows the cost. The prompts can be edited manually, and the images generated into the Library are only limited by the amount of media your tier offers.
For finer control
Section titled “For finer control”Skills can be used for specific prompts, in the Studio or in the Prompt Agent dialog, you can select to use a skill. A Skill is passed to the Prompt Agent to dial in a prompt using a set of instructions. Lorvara has existing skills available, or you can import or create your own with the Skill Agent.
Mode is reactive, not cosmetic. Switching between image and video re-drives the compiled prompt and the parameters card. A Key Prompt written for an image model is not the same string when aimed at a video model, and the page does not pretend otherwise.
References appear in the Library. The reference grid reports which assets a Key Prompt uses in the final generation; it is a statement of what is attached, not a place to attach them. That’s done in either the Brief or the Studio.
Assign a Cover Image in the Library. The Library has some unique functions including setting a cover image. This image will appear as thumbnails in Dashboard and Project tables. The cover will also appear in the Gallery should you set a prompt to be visible there. You can also favorite images in the library and set the status of any prompt (Working | Completed) to help manage your workflow.
What to read next
Section titled “What to read next”- How a prompt is assembled from a shot — what the composed text is made of.
- How do Skills work? — what shapes the way a Key Prompt is written.
- How do Briefs work? — the world every Key Prompt already knows.
- The shot list — where Key Prompts come from.