WHAT IT DOES
Features
in detail.
- Creating images from text descriptions
- Existing and added compatible LoRAs
- Hires Fix for upscaling during generation
- Separate upscaling of a finished result
Here, the task starts with text
This YesAI mode supports text-based generation; input references and inpainting are unsupported. To change an existing photograph, choose an editor that supports source images.
For a new image, define the subject, setting and visual technique. Then add lighting and composition. 'An illustration of a room' leaves the space to the model, while describing a window, desk and evening light establishes a clearer intention. Do not pack the prompt with random adjectives: consistency among key features matters more.
Why choosing a LoRA deliberately matters
A LoRA helps steer the model toward particular visual features. YesAI describes using existing add-ons and installing compatible new ones. Before experimenting, check the selected LoRA's base model: it must match the Stable Diffusion branch selected in the service.
Compare results by changing one variable at a time. Keep the prompt's foundation, then change a style or add-on. This makes the useful influence easier to identify. Changing subject, lighting and LoRA together makes a successful result harder to repeat. Keep short notes about effective descriptions and settings.
Upscaling and further processing
The service page lists Hires Fix and separate image upscaling for this model. These address size and detail but do not replace selecting a suitable frame. First check the pose, object count and overall composition. Only then decide whether a larger image is needed. Otherwise, enlargement also makes its errors more noticeable.
For further edits, open the finished result in another supported YesAI tool. The prompt below does not require a source photograph: describe the desired image in words. In the service, check the active mode and add-on compatibility, then compare the result with your original intent.
Suitable tasks
- Illustrations for a personal project
- Background and setting variations
- Exploring an artistic style
- Working with a LoRA from the correct branch
Things to consider
- This model currently does not accept input references in YesAI.
- Area editing through inpainting is not supported in the current mode.
- Choose a LoRA for the specific Stable Diffusion branch.
FROM AN EXAMPLE TO YOUR OWN IDEA
Where to start a prompt
Choose an example and insert your own details.
Cozy interior
A small bookbinder's workshop: wooden table, orderly tools, a stack of books and a tall window on the left. Soft morning light, a calm painterly illustration, natural materials and a clear composition. No people or lettering.
Environment concept
A rocky island with a small research pavilion by the water. An overcast day, a low sky and a cool gray-and-blue palette. A wide frame, with the building occupying little space. Concept art with clear, large forms.
Style check
Create an illustration of a teapot and two cups on a simple table. Focus on silhouettes and areas of color. Diffused daylight, a white background, no patterns or text. This subject will be used to compare selected styles.
Workflow
- 01
Prepare a text description of the scene.
- 02
Choose a compatible model and a LoRA in YesAI if needed.
- 03
Evaluate the image first, then decide about upscaling.
A FEW MORE ANSWERS
Questions about Stable Diffusion
Can I edit a photograph in this mode?
Input references and inpainting are unsupported in this YesAI mode. For photographs, choose an editor such as Qwen Images or Flux.
How does this page differ from SDXL?
In YesAI, these are different family branches with separate settings and compatible LoRAs. Add-ons cannot automatically be transferred between them.
When should I upscale?
After checking content and composition. Fix object, face and pose errors before Hires Fix or upscaling so you do not enlarge a failed image.
