text_to_image
Generate images from a text description. Customize resolution, sampling steps, and more to get the exact image you need.
Instructions
Generate images from a text description.
All parameters except prompt are optional with sensible defaults. Change any parameter to customize the generation.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| cfg | No | Classifier-free guidance scale. Default: 8.0. Higher = closer to prompt. Typical: 1.0-20.0. | |
| seed | No | Random seed for reproducibility. Default: -1 (random). | |
| steps | No | Sampling steps. Default: 20. Higher = better quality, slower. Typical: 15-50. | |
| width | No | Image width in pixels. Default: 512. Must be multiple of 8. | |
| height | No | Image height in pixels. Default: 512. Must be multiple of 8. | |
| prompt | Yes | Text description of the desired image. | |
| sampler | No | Sampler algorithm. Default: "euler". Valid: "euler", "dpmpp_2m", "dpmpp_sde", "ddim", "uni_pc", etc. Use list_samplers_and_schedulers() to see all. | euler |
| lora_name | No | LoRA model to apply. Default: "" (none). Use list_models(folder="loras") to see options. | |
| scheduler | No | Noise scheduler. Default: "normal". Valid: "normal", "karras", "sgm_uniform", "exponential", etc. Use list_samplers_and_schedulers() to see all. | normal |
| batch_size | No | Number of images to generate. Default: 1. | |
| checkpoint | No | Checkpoint model name. Default: "" (auto-selects first available). Use list_models(folder="checkpoints") to see options. | |
| lora_strength | No | LoRA influence. Default: 1.0. Range: 0.0-2.0. | |
| negative_prompt | No | Things to avoid. Default: "" (none). |
Output Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||