Last updated 08-20-2026
Category:
Reviews:
Join thousands of AI enthusiasts in the World of AI!
InstructPix2Pix
InstructPix2Pix edits a photo from a plain-language instruction such as "make the sky sunset orange" or "turn this into a watercolor." You provide an input image and a short text command, and the model returns an edited version in seconds without training on that specific photo first. It was published as a CVPR 2023 Highlight by researchers at UC Berkeley.
Mask-based editors make you paint regions before changing anything. InstructPix2Pix runs a single forward pass from text alone, skipping per-image fine-tuning or inversion steps that slow other diffusion edit workflows. The trade-off shows up in the paper's failure cases: viewpoint changes, object swaps, and tight local edits remain weak spots.
The project ships open source code on GitHub and a Hugging Face demo space for trying edits in a browser. Training combined GPT-3 and Stable Diffusion to synthesize instruction pairs, and the released model generalizes to real photos up to about 768 pixels wide despite 256x256 training resolution.
Edit an uploaded image from a written instruction without manual masking or per-image fine-tuning
Runs edits in a forward pass and returns results in seconds rather than minutes-long inversion workflows
Generalizes to real photos up to about 768 pixels wide despite 256x256 training resolution
Open source code on github.com/timothybrooks/instruct-pix2pix with a Hugging Face demo space
Allows chained instructions where multiple sequential edits compound on the same photo
Edits photos from text instructions without drawing masks or fine-tuning on each image.
Open source release with a free Hugging Face demo lowers the barrier to testing edits.
Forward-pass inference completes edits in seconds on consumer hardware setups.
Cannot reliably change camera viewpoint or swap objects within a scene.
Sometimes applies broader changes than the instruction intended.
Project site is a research page, not a hosted editor with ongoing product support.
What is InstructPix2Pix?
InstructPix2Pix is a conditional diffusion model that edits images from human-written instructions. InstructPix2Pix takes an input photo plus a text command and returns an edited image in seconds.
Is InstructPix2Pix free to use?
Yes. InstructPix2Pix is a free research release with open source code on GitHub and a public Hugging Face demo. InstructPix2Pix has no paid subscription or commercial pricing page on its project site.
How does InstructPix2Pix work?
InstructPix2Pix was trained on image editing pairs generated by combining GPT-3 and Stable Diffusion. At inference time, InstructPix2Pix follows your written instruction in one forward pass without per-example fine-tuning.
Where can I try InstructPix2Pix?
You can run InstructPix2Pix in the Hugging Face Space at timbrooks/instruct-pix2pix or download the code from GitHub. InstructPix2Pix links both from its project homepage.
What are InstructPix2Pix limitations?
InstructPix2Pix struggles with viewpoint changes, excessive unintended edits, isolating specific objects, and swapping objects within a scene. InstructPix2Pix documents these failure modes on its results page.
Who created InstructPix2Pix?
InstructPix2Pix was created by Tim Brooks, Aleksander Holynski, and Alexei A. Efros at UC Berkeley. InstructPix2Pix appeared as a CVPR 2023 Highlight paper with code and demo links on the project site.