AI Transparency Statement

Last Updated: August 5, 2026

1. How png3D Uses AI

png3D uses advanced AI models, including depth estimation networks and novel view synthesis, to perform 2D to 3D image conversion — a technology that estimates depth information from a single 2D image and generates a 3D representation of the scene.

Our AI models are designed and trained to:

• Estimate depth maps from single 2D images
• Generate 3D representations from 2D inputs (novel view synthesis)
• Produce 3D parallax views for display on compatible devices
• Process images at high resolutions for quality output

png3D's AI uses depth estimation and view synthesis technology: given a user-provided 2D image, the AI estimates scene depth and generates a 3D representation. This is a form of generative AI that is constrained by the original image content — the AI derives 3D structure from the user's own image and does not create unrelated content.

2. What png3D Does NOT Do

It is important to clarify the boundaries of our AI capabilities:

• No text-to-image generation: png3D does not generate images from text descriptions or prompts
• No text-to-video generation: png3D does not generate video content from text
• No synthetic image creation: png3D does not create fictional people, scenes, or objects from scratch
• No face swapping: png3D does not provide face swapping or facial reenactment technology
• No deepfake generation: png3D does not create or facilitate deepfake content
• No AI avatar generation: png3D does not generate AI avatars from user photos
• No voice cloning: png3D does not process audio or voice content
• No user-uploaded models: Users cannot upload custom AI models or training datasets
• No user fine-tuning: Users cannot fine-tune our AI models with their own data

3. AI Safety and Responsible Use

png3D implements several measures to ensure our AI technology is used responsibly:

3.1 Purpose Limitation

Our AI models are designed and trained exclusively for 2D to 3D image conversion. The models' generative capability is limited to deriving 3D structure from the user's own 2D image. They have no capability to generate entirely new content from scratch, produce synthetic media, or perform tasks outside their intended purpose.

3.2 Content Policy Enforcement

We maintain and enforce a comprehensive Content Policy that prohibits the use of our service for illegal, harmful, or abusive purposes. We implement automated content moderation measures to detect and prevent prohibited content. Accounts found violating these policies are subject to suspension or termination.

3.3 User Controls

Users have full control over what areas of their images are processed. Images and processed results are deleted immediately after processing. No user content is retained or stored beyond the processing session.

4. Transparency and Disclosure

png3D is transparent about its use of AI technology:

• Our website and marketing materials clearly describe the AI-powered nature of our service
• We do not misrepresent AI-processed images as authentic unmodified photographs
• Our Terms of Service and Content Policy are publicly available
• This AI Transparency Statement is publicly accessible

5. Compliance with EU AI Act

png3D acknowledges the evolving regulatory landscape for artificial intelligence, including the EU AI Act. Based on our current understanding of the regulatory framework:

5.1 Risk Classification

2D to 3D image conversion technology, as provided by png3D, does not fall within the high-risk or prohibited AI categories contemplated by the EU AI Act. Our service performs a targeted image conversion function (deriving 3D structure from user-provided 2D images) rather than engaging in high-risk AI use cases such as biometric categorization, social scoring, or unrestricted content generation.

5.2 Transparency Obligations

While png3D's AI generates new pixel content within user-designated removal areas (a form of constrained generation), this content is derived from and bounded by the surrounding image context. The service operates as an image editing tool, not as an unrestricted content generation platform.

We voluntarily commit to transparent disclosure of our AI capabilities and limitations, as outlined in this statement.

5.3 Ongoing Monitoring

We continuously monitor regulatory developments in AI governance and will update our practices as legal requirements evolve.

6. Content Authenticity

png3D does not apply watermarks to processed images. The platform processes images as uploaded and preserves existing metadata where technically feasible.

Users are responsible for ensuring they have the legal right to process images they upload. png3D does not encourage or facilitate the removal of copyright protection measures from copyrighted content without authorization.

7. Model Training and Data

png3D's AI models are trained using lawfully obtained datasets. We do not use user-uploaded images to train or improve our AI models unless explicit, informed consent is obtained from users. User privacy and data rights are respected in accordance with our Privacy Policy.

8. Continuous Improvement

AI technology and its regulatory landscape are rapidly evolving. png3D is committed to:

• Regularly reviewing our AI practices against emerging best practices
• Updating this transparency statement as our services and applicable regulations evolve
• Engaging responsibly with the broader discourse on AI governance
• Maintaining open channels for user feedback on AI-related matters

9. Contact

© 2026 png3D. All rights reserved.