System information (version)

• OpenCV => 4.2
• Operating System / Platform => Windows 64 Bit
• Compiler => Visual Studio Code 1.56.2

Detailed description

When running cv2.seamlessClone() the error is a bit misleading when the incorrect image path is supplied. It doesn't make it obvious that the problem is in the path

Steps to reproduce

import cv2
# Read i

Similar to jpeg / jpg we would appreciate adding alternate tif delegating to tiff format in formats map which is used in toFormat method.

https://github.com/lovell/sharp/blob/7467fa8b50d280780dfa9fe9a7248deaa499c7a1/lib/output.js#L14

Right now we put image to the center and pad it to the desired size.

It would be great to add an option to put it to the top left, top right, bottom left, bottom right positions.

Prerequisites

• I have written a descriptive issue title
• I have verified that I am running the latest version of ImageSharp
• I have verified if the problem exist in both DEBUG and RELEASE mode
• I have searched open and closed issues to ensur

Description

Missed this in reviewing #5397. We prefer to release wheels before source because otherwise there is a small window during which users who pip install scikit-image will require compilation. In general this will be very small but in fact it could be big if e.g. there is a network failure in the VM during release.

The relevant lines are here:

https://github.com/scikit-image

In #1031 several tests were identified that rely on a higher than default tolerance starting from torch==1.9.0:

test/augmentation/test_augmentation_mix.py::TestRandomCutMix::test_random_mixup_beta0
test/augmentation/test_random_generator.py::TestRandomPerspectiveGen::test_random_gen
test/augmentation/test_random_generator.py::TestRandomPerspectiveGen::test_same_on_batch
test/augmentat