Variopinta

Variopinta is a CPU image-augmentation compiler: define a pipeline in Python, then compile it for optimized native execution in Rust, with reproducible randomness and an inspectable execution plan.

Variopinta is pre-alpha. The public API may change between 0.y.0 releases; patch releases preserve documented signatures and data contracts unless a correctness or security fix requires otherwise.

Why Variopinta?

  • One pipeline, from input to output. Decode, augment, and deliver arrays, tensors, or encoded images in one native call. Compilation plans buffer reuse, copy avoidance, and output layout; each target is transformed once for all its outputs. See pipelines and outputs.
  • Reproduce results regardless of worker order. Use a pipeline seed and a per-call key to replay an augmentation independently of call order or worker assignment, within the same release and execution environment. See control randomness.
  • See what the compiler actually does. explain() exposes the execution plan, including pixel passes, buffers, copies, and layout changes, so you can inspect which optimizations your pipeline uses. See execution-plan inspection.

Install

Variopinta supports CPython 3.10–3.13 on 64-bit x86 Linux with glibc 2.34 or newer and on macOS 11 or newer running natively on Apple Silicon.

python -m pip install variopinta

PyTorch is optional and needed only for ReturnTensor outputs. Platform details and source-build instructions are in Getting started.

Quick start

import numpy as np
import variopinta as vp

pipeline = vp.Pipeline(
    [
        vp.RandomCrop(256, 256),
        vp.Resize(224, 224),
        vp.HorizontalFlip(p=0.5),
        vp.Normalize(),
    ],
    seed=42,
).compile()

image = np.random.default_rng(0).integers(0, 256, (320, 320, 3), dtype=np.uint8)
output = pipeline(image, key=0)
replayed = pipeline(image, key=0)

assert np.array_equal(output, replayed)
print(output.shape, output.dtype)  # (224, 224, 3) float32
print(pipeline.explain())

Pipeline is the semantic reference executor. .compile() selects the optimized execution plan while keeping the same call signature and keyed result.

Keep the pipeline, input, seed, and key fixed to replay a result. Exact replay is not guaranteed across releases, builds, or platforms.

Documentation

Current scope includes images and semantic masks with NumPy, encoded-buffer, and local-path inputs. It does not include boxes, keypoints, GPU execution, native batches, or Python callbacks inside a pipeline.

Project information

Performance depends on the pipeline, image sizes, output formats, and hardware. The benchmark harness and recorded evidence support comparisons for specific tested configurations.

Variopinta is licensed under the Apache License 2.0. Native-wheel attributions are in THIRD_PARTY_NOTICES.

Variopinta is the feminine form of the Spanish variopinto: “varied in color or appearance,” from Italian variopinto, “varied” and “painted.” — RAE

Project links

Data verified by PyPI on Sep 7, 2026
Data provided by the project maintainers, verified at the time the release was uploaded to PyPI.

Key dates

PyPI data
Data sourced directly from PyPI's database.
  • Released:
Latest release

1 maintainer

PyPI data
Data sourced directly from PyPI's database.
Avatar for claverru from gravatar.com claverru

Credits

Author: claverru

GitHub Statistics

Data verified by PyPI on Sep 7, 2026
The GitHub source repository was provided by the project maintainers and verified by PyPI at the time of upload. Stars, forks, and open issues/PRs are derived from that repository and have not been independently verified.

License expression

Apache-2.0
View SPDX License List

Requires

Python <3.14, >=3.10

Tags

augmentation computer-vision image-processing machine-learning rust

Classifiers

Report project as malware

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

variopinta-0.5.0.tar.gz (126.1 kB view details)

Uploaded Source

Built Distributions

If you're not sure about the file name format, learn more about wheel file names.

variopinta-0.5.0-cp310-abi3-manylinux_2_34_x86_64.whl (1.6 MB view details)

Uploaded CPython 3.10+manylinux: glibc 2.34+ x86-64

variopinta-0.5.0-cp310-abi3-macosx_11_0_arm64.whl (1.4 MB view details)

Uploaded CPython 3.10+macOS 11.0+ ARM64

File details

Details for the file variopinta-0.5.0.tar.gz.

File metadata

  • Download URL: variopinta-0.5.0.tar.gz
  • Upload date:
  • Size: 126.1 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/7.0.0 CPython/3.13.14

File hashes

Hashes for variopinta-0.5.0.tar.gz
Algorithm Hash digest
SHA256 df33a1ee5a3a26c60533ec73b3f0970a440a78eea3adb80fc78b3f4f8092e1e1
MD5 8b073abc09d0a9e22027d153dcdfb3b6
BLAKE2b-256 85db4e35eea5c8de34016a25dd33c29b7449b35b539b095db8fea9173e1a07a0

See more details on using hashes here.

Provenance

The following attestation bundles were made for variopinta-0.5.0.tar.gz:

Publisher: release.yml on claverru/variopinta

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file variopinta-0.5.0-cp310-abi3-manylinux_2_34_x86_64.whl.

File metadata

File hashes

Hashes for variopinta-0.5.0-cp310-abi3-manylinux_2_34_x86_64.whl
Algorithm Hash digest
SHA256 da713a30657854c0f18dfe0f7dbfbcb3df3cbf7e0619114968708f11c16e0a02
MD5 5690ffe3caff621b82a43484800a239e
BLAKE2b-256 639fa06a16f1b219226d65b59aa74e21167f3c26ee0f2c699a0500659046a39b

See more details on using hashes here.

Provenance

The following attestation bundles were made for variopinta-0.5.0-cp310-abi3-manylinux_2_34_x86_64.whl:

Publisher: release.yml on claverru/variopinta

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file variopinta-0.5.0-cp310-abi3-macosx_11_0_arm64.whl.

File metadata

File hashes

Hashes for variopinta-0.5.0-cp310-abi3-macosx_11_0_arm64.whl
Algorithm Hash digest
SHA256 e5302cbffcb579bec1f41c0980bdf9b322fd023e667bf180f188b26c44fdf62a
MD5 f5130349b1ebe9e5f2e3e6cb2c2a4440
BLAKE2b-256 b66eb76ad02fea47e94948d2faa0b793236b8470e3648c696cf579ea9afa56ce

See more details on using hashes here.

Provenance

The following attestation bundles were made for variopinta-0.5.0-cp310-abi3-macosx_11_0_arm64.whl:

Publisher: release.yml on claverru/variopinta

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Release history Release notifications | RSS feed

This release

0.5.0 This release

3 files

0.4.4

3 files

0.4.3

3 files

0.4.2

3 files

0.4.1

3 files

0.4.0

3 files

0.3.1

3 files

0.3.0

3 files

0.2.0

3 files

0.1.0

2 files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page