The tiering is stated plainly and it decides the cost: resizing, cropping, watermarking and format conversion run on your machine for nothing, background removal and upscaling get three free runs a day, and only the generation work — product shots, avatars, prompt edits, video — draws credits, at rates the tool descriptions state up front. In practice that means an agent can do the entire dull half of image work before an account exists.
Image tooling in three tiers: local edits that run offline with no account, AI tools with a small free daily allowance, and paid generation — product photos, try-on, scenes, avatars, video — that needs a PixelPanda token and spends credits.
- Local edits with no account and no network: resize, crop, rotate, flip, grayscale, invert, rounded corners, borders, compression, and conversion between PNG, JPEG, WebP, BMP and TIFF
- Several images merged horizontally, vertically or as a grid, with a gap between them
- Opacity, Gaussian blur, pixelation, text watermarks placed centred, cornered or tiled, and brightness, contrast, sharpness and saturation adjusted
- A profile picture cut as a circle, square or rounded square at the size you name
- An image's dimensions, format, file size, colour mode and DPI
- AI background removal to a transparent PNG, 2x or 4x upscaling with Real-ESRGAN, text and watermark removal by OCR and inpainting, and image analysis for objects, colours, composition, quality and mood — free for 3 uses a day without an account
- Product photography: a product uploaded, then product photos with an avatar, try-on shots, or lifestyle, studio and detail scenes with no avatar at all
- Reusable AI avatars generated, either guided or matched to the look of a reference photo
- An image reverse-engineered into four ready-to-use prompt variants — general, Flux, Midjourney v6 and Stable Diffusion
- Text-prompt image editing through FLUX.1 Kontext Pro, keeping the subject while colours, scenes or elements change
- A UGC-style video animated from a still, with native audio and lip-sync generated from a script
- Six enhancement variants from one image — AI Restore, Vivid Restore, Conservative, Face Restore, Face Enhance and GFPGAN Restore
- Job, avatar and video generations polled to completion, recent jobs listed, and the credit balance checked
Uvx pixelpanda-mcp, or pip install pixelpanda-mcp. The token is optional: every local tool and the free AI tier work without one, and PIXELPANDA_API_TOKEN — a pk_live_ key — is needed only for the paid tools. The free AI tools are capped at 3 uses a day without an account. Credits are what the paid tools spend: 1 per generated product photo, try-on shot, scene, avatar or image-to-prompt run, 2 for a prompt-driven edit, and 50 for a video. PIXELPANDA_API_URL overrides the API base, which defaults to https://pixelpanda.ai.
One command — uvx pixelpanda-mcp