Pruna AI family
# P Image Edit High-speed multi-image editing for professional creative workflows. ## Overview P Image Edit is a high-performance, production-grade AI model designed to handle complex multi-image editing tasks with sub-second latency. Unlike standard generative models that struggle with consistency, P Image Edit leverages reference-based architecture to perform precise relighting, style transfer, and subject consistency tasks. It is specifically engineered for marketing teams and ad agencies that require rapid iteration without sacrificing visual fidelity. By utilizing multiple reference images, the model maintains structural integrity and brand identity across various creative assets. Whether you are adapting a product shot for different seasonal campaigns or ensuring a character remains consistent across multiple social media frames, P Image Edit provides the speed and reliability necessary for high-volume production environments. Its optimized inference path makes it one of the most cost-effective solutions for scaling creative operations. ## Use Cases * **E-commerce Relighting:** Adjust product lighting to match seasonal campaign aesthetics without reshooting. * **Brand Consistency:** Maintain subject identity across diverse ad formats, from Instagram Stories to wide-screen banners. * **Style Transfer for Campaigns:** Apply specific brand-aligned artistic styles to raw photography for cohesive visual storytelling. * **Rapid Asset Iteration:** Generate multiple variations of a scene for A/B testing ad creatives in real-time. * **Texture Extraction & Application:** Seamlessly map textures onto 3D models or product surfaces for high-fidelity mockups. * **Scene Adaptation:** Transform white-background product shots into lifestyle scenes or stylized environments instantly. ## Parameters This model offers a robust set of controls to fine-tune your output. Please refer to the Input Parameters section in the Pixloop interface to configure your seed, enable turbo mode for faster inference, manage aspect ratios, and select specific task weights such as 'relight' or 'subject_consistency'. ## Tips for Best Results * **Reference Order:** Always place your primary subject or base image as the first item in the 'images' input array to ensure the model anchors the edit correctly. * **Prompt Specificity:** When describing your edit, explicitly reference the input images (e.g., 'Apply the lighting from image 2 to the subject in image 1') to guide the model's attention. * **Turbo Mode Strategy:** Use 'Turbo' mode for rapid prototyping and initial drafts; disable it for complex, high-detail edits where maximum precision is required. * **Aspect Ratio Matching:** Use the 'match_input_image' setting to preserve the composition of your original assets, which is critical for maintaining professional layout standards. * **Task Selection:** Choose the specific 'replicate_weights' option that matches your goal—such as 'subject_consistency' or 'relight'—rather than relying on the default setting for specialized tasks. ## About P Image Edit is developed by Pruna AI and optimized for high-speed inference on the Replicate platform. Built specifically for production pipelines, it represents a significant advancement in multi-image conditioning, allowing creative teams to bridge the gap between raw photography and polished, AI-enhanced marketing assets. For those interested in pure generation tasks, the provider also maintains the P Image model for standalone image creation.
Model input reference derived from preset schema.
| Name | Type | Required | Default | Description |
|---|---|---|---|---|
| seed | integer | No | null | Random seed. Set for reproducible generation. |
| turbo | boolean | No | false | If turned on, the model will run faster with additional optimizations. For complicated tasks, it is recommended to turn this off. |
| images | image | Yes | null | Images to use as a reference. For editing task, provide the main image as the first image. |
| prompt | string | Yes | null | Text prompt for image generation. Make sure to describe your edit task clearly. You can refer to the images as 'image 1' and 'image 2' and so on. |
| aspect_ratio | select | No | "match_input_image" | Aspect ratio for the generated image. `match_input_image` will match the aspect ratio of the first image. |
| replicate_weights | select | No | "default" | Task to perform with P-Edit. |
| disable_safety_checker | boolean | No | true | Disable safety checker for generated images. |