Text-to-Image
This pipeline generates an image from a text description.
Use command /image in a format /image prompt_text --argument 1. Parameters are always added to the end of a prompt. You can add multiple parameters to each prompt.
Arguments
Keyword Strength
Adjust keyword strength is to use () and []. (keyword) increases the strength of the keyword by a factor of 1.1 and is the same as (keyword:1.1). [keyword] decrease the strength by a factor of 0.9 and is the same as (keyword:0.9).
You can use multiple of them, just like in algebra. The effect is multiplicative.
(keyword) is equivalent to (keyword: 1.1)
((keyword)) is equivalent to (keyword: 1.21)
(((keyword))) is equivalent to (keyword: 1.33)
Similarly, the effects of using multiple [] are:
[keyword] is equivalent to (keyword: 0.9)
[[keyword]] is equivalent to (keyword: 0.81)
[[[keyword]]] is equivalent to (keyword: 0.73)
Example: /image (((Phil))) dressed as a tiny vampire, standing under a full moon, surrounded by bats and glowing pumpkins, in a foggy graveyard ((distorted perspective, unique perspective, side profile, viewed from above, (ultra-wide angle lens), vanishing point))
Aspect Ratio
-ar and --ar change the aspect ratio of an image. -ar 1:1 will produce a square image, -ar 16:9 will generate a standard wide image. Values from -ar 4:1 to -ar 1:4 are supported.
Batch Generation
We currently offer Batch Generation to Elite tier users.
-n and --n change the number of images to generate per command. For example, /image anime cat -n 2 will generate two images.
Best Practices
Describe, don't direct
Describe the image you want to see rather than instructing it directly to draw or illustrate. Just provide vivid details of the scene, and the AI will interpret them into a visual output
❌ Don't: /image draw a cat sitting by the window
✅ Do: /image a fluffy cat sitting by a sunlit window, gazing outside with a calm expression
❌ Don't: /image draw a picture of brett, add a mountain in the background
✅ Do: /image Brett stands triumphantly on the peak of a rugged mountain. His broad shoulders and powerful arms are bare, except for tattered blue jeans that cling to his frame, torn from his intense journey to the top. Snow and wind swirl around him, yet he remains unfazed, his gaze fixed on the vast, breathtaking horizon.
.
Stick to visual details
Focus only on visual elements rather than conceptual or background information about the characters or scene. The AI doesn’t interpret relationships, histories, or abstract concepts—it generates based solely on what you describe visually.
❌ Don't: /image brett is the father of vitalik buterin
✅ Do: /image brett standing next to vitalik buterin
Be detailed
The more details in the prompt, the better.
✅ Do: /image (((Phil))) dressed as a tiny vampire, standing under a full moon, surrounded by bats and glowing pumpkins, in a foggy graveyard ((distorted perspective, unique perspective, side profile, viewed from above, (ultra-wide angle lens), vanishing point))
Quotes only for visible text
Don’t use quotation marks around the entire prompt. Quotation marks can confuse the AI, making it think you want the text itself to appear in the image. Only use quotes for specific text you want shown within the image (e.g., a sign or ticker).
❌ Don't: /image "Phil dressed as a werewolf, howling playfully at the full moon, standing on a cliff with misty mountains and eerie forests in the background"
✅ Do: /image Phil dressed as a werewolf, howling playfully at the full moon, standing on a cliff with misty mountains and eerie forests in the background
We are working on an automatic prompt enhancer, coming soon.
Examples
Last updated
