Terms of Service
Last updated: 6 October 2026
These Terms govern your use of PixelMagic (the “Service”), an AI image studio operated by PixelMagic, S2 Jayavijaya Apts, 5th St, Chennai, Tamil Nadu, 600100, India (“we”, “us”). By creating an account or using the Service you agree to these Terms, our Privacy Policy and our Refund & Cancellation Policy.
1. The Service
PixelMagic lets you create and edit images and videos with artificial intelligence, including:
- generating images from text prompts;
- editing images with instructions, masks and reference images, and removing objects;
- removing and replacing backgrounds;
- upscaling and enhancing images;
- translating text inside images and splitting images into layers;
- turning photos into videos with music, effects and text;
- karaoke lyric videos: separating a song’s vocals from its music and transcribing its lyrics;
- bulk processing and automated workflows.
The Service is available through our website, a public API, an MCP server for AI agents, a command-line tool and plugins for third-party applications. Some features are marked Beta and may change or be removed.
2. AI models and third-party providers
The Service uses a mix of models we host ourselves (for example the open-source Lucida model for background removal) and models provided by third parties, which we access through third-party AI infrastructure providers. Which model handles a request depends on the tool and the model you choose. Third-party models may have their own usage restrictions, which apply to your use of them through PixelMagic. We may add, change or remove models at any time.
3. Accounts
You must provide a valid email address and are responsible for all activity under your account, including by team members you invite, and for keeping your password and API keys secret. You must be at least 18 years old, or have consent from a parent or guardian. You may not share an account to avoid plan limits.
4. Credits, plans and billing
- The Service runs on credits shared across all tools. Each tool and model shows its credit cost before you run it; for example, one standard background removal uses one credit, while premium models, high resolutions, reference images and videos can use more.
- Credits come from the free plan, a monthly plan (or 1-month pass) or one-off credit packs. They are non-transferable, have no cash value, and expire 12 months after they are granted unless stated otherwise at purchase.
- The free plan includes a limited monthly allowance; we may change free allowances at any time.
- Plans bought in Indian rupees (INR, via Razorpay) are one-time 1-month passes: you pay once, the plan lasts one month, and it does not renew automatically — we email you a few days before it ends and you can buy another month from the Pricing page.
- Plans bought in other currencies (via Paddle) are monthly subscriptions that renew automatically each month until cancelled. You can cancel at any time from Dashboard → Billing; you keep the plan until the end of the period you have paid for. Prices may change; a new price applies from your next billing period after we give notice.
- Payments in India are processed by Razorpay. Payments from other countries are processed by Paddle.com, which acts as the merchant of record and reseller of the Service, collects applicable sales tax or VAT, and whose buyer terms also apply to that purchase. Applicable taxes (including GST) may be added.
- If a job fails because of an error on our side, its credits are returned to your balance automatically.
5. Your content and outputs
“Inputs” are the images, videos, audio, prompts and other material you provide. “Outputs” are what the Service generates for you.
- You keep all rights you have in your Inputs. You confirm you have the rights and any necessary consents (including from people shown in them) to use them with the Service.
- As between you and us, you own the Outputs, and you may use them for personal and commercial purposes, subject to these Terms and any restrictions of the model that produced them.
- AI outputs may not be unique: other users may receive similar outputs. We do not guarantee that Outputs are protected by copyright or that they do not resemble existing works, and you are responsible for checking them before you publish or rely on them.
- You grant us a limited licence to store, process and transmit your Inputs and Outputs, including sending them to the model providers that process your request, solely to provide and secure the Service.
- We do not use your Inputs or Outputs to train AI models, and we do not sell them.
5A. Songs, music and karaoke
Some tools (for example Karaoke Lyrics) let you upload a song or music video, separate the singer’s voice from the music, transcribe the lyrics and show them in your video.
- Upload only songs, recordings and videos you own or have permission to use. Most commercial songs are protected by copyright in both the recording and the composition and lyrics.
- Removing the vocals, making a music-only track, transcribing or displaying the lyrics does not give you any rights in the song. These features are a technical service; they are not a licence from us or from the rights holders.
- If you publish, share publicly, perform or monetise a video made from a song you do not own (for example on YouTube, Instagram or in a paid event), you are responsible for getting the necessary permissions — typically from the music publisher (composition and lyrics) and the record label (recording), or through a licensing body. Without them, keep such outputs to private, personal use.
- Vocal separation and lyric transcription are produced by AI and can be imperfect: traces of the voice may remain, and words or timings may be wrong. Check the result before you rely on it.
- We may remove content, disable sharing links or suspend accounts in response to a valid copyright complaint. Rights holders can contact us at the address below.
6. Acceptable use
You may not use the Service to create, upload or share content that:
- is unlawful, or infringes anyone’s copyright, trademark, privacy or publicity rights;
- depicts or sexualises minors in any way;
- is sexual or intimate imagery of a real person without their consent, including “deepfakes”;
- impersonates a real person or organisation, or is designed to deceive (for example fake documents, IDs, receipts or misleading news);
- harasses, threatens, or promotes violence, hate or self-harm;
- removes watermarks or other rights information from images you do not have the right to edit;
- distributes vocal or instrumental tracks separated from recordings you do not have the rights to, or publishes another person’s lyrics or music without permission.
You also may not reverse-engineer the Service, bypass rate limits, credit metering or safety filters, use it to build a competing model, or resell it as a substantially identical product without our written permission. We may refuse prompts, block outputs, and suspend accounts that break these rules, and we will report child sexual abuse material to the authorities.
7. API, MCP server and integrations
API keys are personal to your account. API and MCP usage uses the same credits as the website and is subject to per-minute rate limits for your plan. We may change or retire endpoints with reasonable notice. Third-party platforms you connect (such as Shopify, Zapier, Make, Figma, Google Drive, Dropbox or your own storage) are governed by their own terms.
8. Music, templates and stock assets
Built-in music tracks, video templates, effects and other assets we provide are licensed to you only for use within Outputs you create with the Service, not for separate redistribution.
9. Learn hub and third-party links
Guides, model comparisons, model listings and news on our Knowledge pages are for general information only. Some news summaries are prepared with AI assistance from public sources, which we link to. Model names belong to their owners, and we are not responsible for third-party websites.
10. Availability and changes
We aim for high availability but do not guarantee uninterrupted service. We may modify, suspend or discontinue features with reasonable notice where practical. We may update these Terms; we will post the new version here with a new date and email you about material changes. Continuing to use the Service after changes take effect means you accept them.
11. Intellectual property
The Service, its software, design, branding and content (other than your Inputs and Outputs) are owned by us or our licensors. The Lucida background-removal model is used under its open-source MIT licence.
Speech recognition (video captions and transcripts): English speech is transcribed with Phonon-2 by Fermion Research, a model derived from NVIDIA’s parakeet-tdt-0.6b-v3 and used under the Creative Commons Attribution 4.0 (CC BY 4.0) licence; we use the model unmodified. Other languages are transcribed with OpenAI Whisper, used under its MIT licence.
AI voice (video narration): Powered by Stellon Labs. Voices are generated with KittenTTS 2 by KittenML / Stellon Labs, used unmodified under the Stellon Labs Community License (This Stellon Labs Model is licensed under the Stellon Labs Community License. Copyright © Stellon Labs Inc. All Rights Reserved.), with the speech decoder from ResembleAI’s Chatterbox (MIT). AI voice outputs may not be used to train or improve other AI models. Indian-language voices are generated with Indic Parler-TTS by AI4Bharat, IIT Madras, used unmodified under the Apache License 2.0. Native Chinese, Japanese, Korean, French, German and Spanish voices and voice cloning use Ming 3.0 TTS by inclusionAI (Ant Group), used unmodified under the MIT licence.
Voice cloning: you may only clone your own voice. You record yourself reading a consent statement; that recording is stored privately in your account, used only to generate narration for you, and deleted when you delete the voice or your account. Cloning or impersonating anyone else's voice is not allowed and may lead to account suspension.
Karaoke Lyrics uses these open-source components, each under its own licence:
- Vocal separation: the Mel-Band RoFormer vocals model by Kimberley Jensen (MIT), run with model code from Music-Source-Separation-Training by Roman Solovyev (ZFTurbo) (MIT); Meta’s Demucs (MIT) as a fallback.
- English lyrics: Phonon-2 by Fermion Research (CC BY 4.0, as above).
- Hindi lyrics: Qwen3-ASR-1.7B by the Qwen team, Alibaba Cloud (Apache License 2.0).
- Other Indian languages (Tamil, Telugu, Malayalam, Kannada, Bengali, Marathi, Gujarati, Punjabi, Urdu and others): IndicConformer by AI4Bharat, IIT Madras (MIT).
- Other languages, word timings and pasted-lyrics alignment: OpenAI Whisper with faster-whisper and stable-ts (MIT).
These components are provided by their authors “as is”, without warranty, as their licences state. Their names and trademarks belong to their owners and do not imply endorsement of PixelMagic.
12. Disclaimers and liability
To the maximum extent permitted by law, the Service and all Outputs are provided “as is” and “as available”, without warranties of any kind. Our aggregate liability for any claim is limited to the amount you paid us in the three months before the claim arose. We are not liable for indirect, incidental or consequential losses, or for loss of data, profits or goodwill. Nothing in these Terms limits liability that cannot be limited by law.
13. Indemnity
You agree to cover our reasonable losses and costs arising from your Inputs, your use of Outputs, or your breach of these Terms.
14. Termination
You may stop using the Service and delete your account at any time from your Account settings. We may suspend or terminate accounts that breach these Terms. Unused credits are handled under our Refund & Cancellation Policy.
15. Governing law
These Terms are governed by the laws of India, and the courts of Chennai, Tamil Nadu have exclusive jurisdiction, subject to any mandatory consumer protections in your country of residence.
16. Contact
Questions about these Terms: contact us or email support@pixelmagic.studio.