AD SPACE AVAILABLE · BRAND PARTNERS WELCOME

How to Control Craiyon Image Output Using Style Presets and Negative Prompts

365Finds Editorial
Indexed 11,430 Wiki articles

Put 365 Wiki to work with your AI

Copy the prompt below, paste it into the AI you already use, then ask your question.

Download full Skill

Try it with the prompt first. If you use 365 Wiki often, install the full 365Wiki Skill so you do not have to paste it again. No regular AI? Gemini or Microsoft Copilot on the web will work.

365Wiki Skill

Craiyon (originally developed as DALL-E mini) offers one of the lowest-barrier entry points to generative AI art: anyone can generate a 3x3 grid of nine images directly in a browser without creating an account or paying a subscription. However, unguided prompts frequently produce frustrating AI artifacts—muddled textures, distorted facial anatomy, extra fingers, or unintentional cartoon styling.

To overcome these distortions and achieve predictable results, visual creators and hobbyists can leverage two built-in features: Style Presets and Negative Prompts (labeled in the interface as Negative words or Exclude). By combining explicit visual styles with targeted exclusion rules, you can guide the diffusion model to produce cleaner compositions, preserve artistic consistency, and minimize post-generation clean-up.


1. Choosing the Right Style Preset

Rather than crowding your primary prompt with contradictory descriptive terms (such as "hyperrealistic oil painting sketch"), Craiyon provides preset style buttons that bias the underlying diffusion model toward a specific visual medium:

Creator Tip: Select your style preset first, then write your prompt focusing primarily on the subject, setting, action, and lighting. Avoid repeating the preset type inside the text box (e.g., avoid writing "a photo of a..." when the Photo preset is already toggled), which saves token weight for descriptive nouns and lighting conditions.


2. Refining Output with Negative Prompts

Negative prompts allow you to define what the AI must actively omit during the image generation process. In Craiyon, the dedicated Negative words field functions as an inverse filter. Instead of hoping the model naturally avoids common pitfalls, you explicitly steer the sampling path away from them.

For practical workflows, negative words are most effective when grouped into four targeted categories:

A. Anatomy and Character Glitches

Generating humans, animals, or expressive characters is notoriously prone to structural deformation. When working on character art or portraits, include targeted anatomical exclusions:

bad anatomy, extra fingers, missing fingers, mutated hands, fused fingers, distorted face, unnatural eyes, malformed limbs

B. Image Quality and Artifacts

Craiyon's free tier prioritizes generation accessibility, which can occasionally yield low-contrast or pixelated thumbnails. To raise baseline rendering clarity:

blurry, low quality, pixelated, grainy, jpeg artifacts, low resolution, out of focus

C. Watermarks, Signatures, and Text

Diffusion models trained on broad web imagery often reproduce simulated typography or painter watermarks. Filter these out with:

watermark, signature, text, username, logo, border, frame, cropped, out of frame

D. Style Cross-Contamination

When you want a distinct medium, use the negative prompt to suppress competing styles. For example:


3. The "Less is More" Negative Prompt Rule

A frequent mistake among hobbyists transitioning from advanced tools like Stable Diffusion or Midjourney is pasting massive, 50-word negative prompt blocks into Craiyon. Overloading Craiyon's negative input field can backfire:

  1. It dilutes the model's focus on your positive prompt.
  2. It can flatten dynamic color contrast, resulting in dull, washed-out grey tones.
  3. It can cause unexpected compositional shifts.

Best Practice: Start with a focused baseline of 3 to 5 keywords (such as blurry, bad anatomy, text, extra limbs). Only add specific negative terms if an unwanted artifact persistently appears across your nine-image generation grid.


4. Step-by-Step Creator Workflow

Here is an efficient, five-step loop to produce refined graphics in Craiyon:

  1. Formulate the Positive Core: Write a clear, concise text description focusing on subject, setting, and lighting.
    Example: A robotic cybernetic fox perched on a mossy neon log in a misty pine forest, cinematic backlight, volumetric fog
  2. Set the Preset: Choose Photo for realism or Art for conceptual sci-fi illustration.
  3. Configure Negative Words: Insert high-priority exclusions:
    blurry, low quality, cartoon, watermark, distorted, extra limbs
  4. Generate and Evaluate the 3x3 Grid: Click Draw. Craiyon will deliver nine variations. Scan the grid to identify which composition, posture, and lighting best match your creative intent.
  5. Upscale and Export:
    • Click the preferred thumbnail to open the detail view.
    • Click Upscale to sharpen the image resolution and refine micro-details.
    • Use the Screenshot or download function to save the asset as a clean PNG.
    • If a result is close but needs fine-tuning, check Craiyon's suggested alternative prompts below the grid or adjust your negative words to prune lingering defects.

5. Creator Verdict: When to Use Craiyon

Craiyon is not designed to replace high-end commercial rendering suites like FLUX, Midjourney v6, or DALL-E 3 for billboard-ready commercial photography. Its real competitive advantage lies in zero-friction ideation:

By combining the precision of Style Presets with the defensive filtering of targeted Negative Words, visual creators can tame Craiyon's unpredictability and turn a quick brainstorming tool into a dependable asset in their creative toolkit.

References