Pruna AI P Image Edit LoRA API Documentation

Pruna AI P Image Edit LoRA API Documentation

Playground

Pruna AI P-Image Edit LORA is a fast AI image editing model that edits and transforms images with LORA-based customization. Ready-to-use REST inference API for text-guided image editing, style changes, character consistency, product image updates, marketing assets, and custom AI editing workflows with simple integration, no coldstarts, and affordable pricing.

Features

Pruna AI P-Image Edit LoRA edits one or more input images using a natural-language instruction, with optional LoRA guidance for stronger style or edit control. It is designed for workflows where you want prompt-based image editing together with a LoRA trained specifically for the Pruna p-image-edit-lora pipeline.

Why Choose This?

• LoRA-guided image editing Edit images with natural-language instructions while steering the result with a compatible LoRA.

• Multi-image reference support Use one to five input images to guide appearance, structure, composition, or scene transformation.

• Edit-specific LoRA control Apply lora_weights and tune lora_scale for stronger stylistic or transformation control.

• Flexible aspect ratio handling Use match_input_image to follow the first input image by default, or select a preset aspect ratio when needed.

• Private LoRA support Use hf_api_token when accessing a private or gated Hugging Face LoRA repository.

• Simple fixed pricing Each run uses a flat per-image price.

Parameters

How to Use

1. Upload your reference images — provide one to five images you want to use for the edit.
2. Write your prompt — describe what should change and what should remain unchanged.
3. Add a LoRA (optional) — provide lora_weights if you want LoRA-guided editing.
4. Adjust LoRA strength (optional) — set lora_scale to control how strongly the LoRA affects the result.
5. Choose aspect ratio — keep match_input_image to follow the first input image, or select a preset ratio if needed.
6. Add a Hugging Face token (optional) — use hf_api_token if your LoRA is private or gated.
7. Choose output format — select png, jpeg, or webp.
8. Set a seed (optional) — use -1 for random output, or a fixed value for more reproducible edits.
9. Submit — run the model and download the edited image.

Example Prompt

Make this scene look like the next scene style.

Pricing

Just $0.01 per generated image.

Best Use Cases

• Scene-to-scene style transfer — Rework one scene to match the visual style of another.
• Character-consistent editing — Preserve identity while changing mood, styling, or composition.
• Reference-guided transformations — Use multiple images and a LoRA for tighter edit control.
• Creative adaptation — Turn existing images into new variants for storytelling, design, or campaign work.
• Edit-specific LoRA workflows — Apply LoRAs trained for editing rather than text-to-image generation.

Pro Tips

• Use a LoRA trained specifically for p-image-edit-lora for best compatibility.
• Text-to-image LoRAs should be used with p-image-lora, not this model.
• When using multiple images, explain each image’s role clearly in the prompt.
• Use match_input_image when you want to preserve the framing of the first input image.
• Adjust lora_scale gradually to balance prompt influence and LoRA influence.
• If your LoRA is private or gated on Hugging Face, provide hf_api_token.
• Reuse the same seed when you want more consistent edit iterations.

Notes

• Both prompt and images are required.
• images supports one to five input images.
• lora_weights is optional.
• LoRAs for this model should be trained for p-image-edit-lora.
• Text-to-image LoRAs should be used with p-image-lora instead.
• aspect_ratio defaults to match_input_image, which follows the first input image.
• seed uses -1 for random generation.
• Turbo mode is off and the safety checker is disabled by default; these are not user-facing controls.
• Pricing is fixed at $0.01 per generated image.

Related Models

• Pruna AI P-Image Text-to-Image — Generate images directly from prompts.
• Pruna AI P-Image Text-to-Image LoRA — Generate images from prompts with LoRA style control.
• Pruna AI P-Image Edit — Edit images with natural-language instructions without a LoRA.
• Pruna AI P-Image Upscale — Enhance and enlarge existing images.

Authentication

For authentication details, please refer to the Authentication Guide.

API Endpoints

Submit Task & Query Result

set -euo pipefail

export WAVESPEED_API_KEY="your-api-key"

REQUEST_BODY=$(cat <<'JSON'
{
  "prompt": "A cinematic ocean wave at sunrise, highly detailed",
  "images": [
    "https://interactive-examples.mdn.mozilla.net/media/cc0-images/painted-hand-298-332.jpg"
  ],
  "lora_scale": 1,
  "aspect_ratio": "match_input_image",
  "output_format": "png"
}
JSON
)

# 1. Submit the prediction once. Never retry this request automatically:
# a retry after an unclear failure can create a second, billed task.
CHECK='The task may already exist. Check History or the predictions list before submitting again.'
if ! SUBMIT_RESPONSE=$(curl --silent --show-error --fail-with-body \
  --connect-timeout 10 --max-time 60 \
  -X POST "https://api.wavespeed.ai/api/v3/pruna-ai/p-image/edit-lora" \
  -H "Authorization: Bearer ${WAVESPEED_API_KEY}" \
  -H "Content-Type: application/json" \
  -d "${REQUEST_BODY}"); then
  printf 'Submit failed. %s\n' "${CHECK}" >&2
  exit 1
fi

TASK=$(printf '%s' "${SUBMIT_RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
PREDICTION_ID=$(printf '%s' "${TASK}" | jq -r '.id // empty')
if [ -z "${PREDICTION_ID}" ]; then
  printf 'Submission response did not contain a prediction id. %s\n' "${CHECK}" >&2
  exit 1
fi
RESULT_URL="https://api.wavespeed.ai/api/v3/predictions/${PREDICTION_ID}/result"

# 2. Poll until the prediction reaches a final status.
BODY_FILE=$(mktemp)
trap 'rm -f "${BODY_FILE}"' EXIT
while true; do
  # Transient errors (429, 5xx) are retried, honoring Retry-After.
  HTTP_CODE=$(curl --silent --show-error --fail-with-body \
    --connect-timeout 10 --max-time 60 --retry 5 --retry-delay 2 \
    --output "${BODY_FILE}" --write-out '%{http_code}' \
    "${RESULT_URL}" \
    -H "Authorization: Bearer ${WAVESPEED_API_KEY}" || true)
  case "${HTTP_CODE}" in
    2??) ;;
    000|429|5??)  # dropped connection or temporary error: keep polling
      printf 'Task %s: HTTP %s, polling again in 10s\n' "${PREDICTION_ID}" "${HTTP_CODE}" >&2
      sleep 10; continue ;;
    *) printf 'Task %s: HTTP %s\n' "${PREDICTION_ID}" "${HTTP_CODE}" >&2
      cat "${BODY_FILE}" >&2; exit 1 ;;
  esac
  RESULT=$(jq 'if type == "object" and has("data") then .data else . end' "${BODY_FILE}")
  STATUS=$(printf '%s' "${RESULT}" | jq -r '.status // empty')

  case "${STATUS}" in
    completed) printf '%s\n' "${RESULT}" | jq '.outputs'; break ;;
    failed|cancelled|timeout|deleted)
      printf 'Task %s ended with status %s\n' "${PREDICTION_ID}" "${STATUS}" >&2
      printf '%s\n' "${RESULT}" | jq . >&2; exit 1 ;;
    *) sleep 2 ;;
  esac
done

Parameters

Task Submission Parameters

Request Parameters

Every model also accepts the global request options enable_sync_mode and enable_base64_output; see their pages for details.

Response Parameters

Result Request Parameters

Result Response Parameters