P H3 Image Edit LoRA API Documentation

P H3 Image Edit LoRA API Documentation

Playground

P-H3-Image by Pruna AI, built on MiniMax H3 open weights, edits images from natural-language instructions while helping preserve faces, products, and other important visual details. It is well suited for creative edits, product imagery, branded content, and consistency-focused image workflows. Ready-to-use REST inference API.

Features

P-H3-Image Edit LoRA edits an existing image using text instructions and custom LoRA weights. The first input image is treated as the image to edit, while up to 8 additional images can be used as references. Load up to 3 LoRA weights per request and generate edited images from 1k up to 4k.

Why Choose This?

• LoRA support — Load up to 3 LoRA weights in one request, each with its own scale.
• Instruction-based editing — Describe what should change while using the first image as the main editing target.
• Multi-image references — Use up to 9 images, addressed as <Picture 1>, <Picture 2>, and so on.
• Reference consistency — Use additional images to guide people, products, clothing, hairstyles, and other visual details.
• Up to 4K output — Choose 1k, 1.5k, 2k, 3k, or 4k.
• Flexible framing — Keep the first image’s aspect ratio or select another supported ratio.
• Multiple output formats — Export as JPEG, PNG, or WebP.

Parameters

How to Use

1. Upload images — Place the image to edit first, followed by any reference images.
2. Write an edit instruction — Describe what should change and what should remain consistent.
3. Reference images clearly — Use <Picture 1>, <Picture 2>, and so on when explaining the role of each image.
4. Add LoRAs optional — Load up to 3 LoRA weights and configure their scales.
5. Choose resolution — Select 1k through 4k depending on the required detail.
6. Choose framing optional — Keep the first image’s aspect ratio or select another supported ratio.
7. Submit — Generate and retrieve the edited image.

Pricing

Pricing is based on the selected resolution, with a fixed $0.015 LoRA surcharge per generation.

The first input image is included in the base price. Each additional input image costs $0.005.

The LoRA surcharge is fixed per generation and does not increase with the number of LoRAs used within the supported limit.

Additional Images

Aspect ratio, output format, and seed do not add separate charges.

Best Use Cases

• LoRA-guided editing — Apply custom styles, characters, products, or visual identities during editing.
• Portrait editing — Change clothing, environment, lighting, or styling while keeping the original subject.
• Product editing — Use references and LoRAs to guide product appearance and presentation.
• Multi-reference composition — Combine visual elements from several reference images.
• Advertising creatives — Adapt existing assets for campaigns, promotions, and social content.
• High-resolution editing — Generate detailed edited images up to 4k.

Pro Tips

• Clearly describe what should change and what should remain unchanged.
• Refer to each important image explicitly, such as the jacket from <Picture 2>.
• Adjust each LoRA scale based on how strongly you want it to influence the result.
• When combining multiple LoRAs, keep the prompt clear about the role of each concept.
• Use fewer references when several images provide conflicting visual guidance.
• Use 3k or 4k when finer visual detail is important.
• Use a fixed seed when comparing prompt, reference, or LoRA changes.

Notes

• prompt and images are required.
• images supports between 1 and 9 inputs.
• The first image is the primary image being edited.
• Additional images are treated as references.
• Up to 3 LoRA weights can be used in one request.
• Each LoRA uses a {path, scale} configuration.
• aspect_ratio defaults to the first image’s aspect ratio.
• resolution defaults to 1k.
• output_format defaults to jpeg.
• Use seed=-1 for random seed selection.
• LoRA support adds a fixed $0.015 per generation.
• Each additional input image after the first adds $0.005.

Related Models

• P-H3-Image Text-to-Image — Generate images directly from text prompts.
• P-H3-Image Edit — Edit existing images without LoRA guidance.
• P-H3-Image Text-to-Image LoRA — Generate images with custom LoRA guidance.
• MiniMax H3 Text-to-Video — Generate videos with the MiniMax H3 model family.

Reference

• How to train your LoRA?
• How to use your LoRA?

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"
  ],
  "aspect_ratio": "1:1",
  "resolution": "1k",
  "output_format": "jpeg"
}
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/wavespeed-ai/p-h3-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