All versions of Chrome are affected since as early as v23.

Chrome's implementation of matrix() is incorrect in case scaling traverses the 0 point.
The matrix in the test case is basically scaleX(1) -> scaleX(-1) and Chrome handles direct scaling correctly:

    g {
        transform: scaleX(1) translate(100px,150px);
        transition: all 1s ease;
        transform-origin: 250px 200px;
    }

    .move {
        fill: red;
        transform: scaleX(-1) translate(100px,150px);
    }

The only other workaround is to split the transformation matrix at zero point, which is very inconvenient since 0 is not animatable and it would have to be 1 to 0.00000001 and -0.0000001 to -1.

P.S. "transform-origin" in #0 has invalid syntax, but that wasn't what caused the problem here.

Oops, yep I forgot to get rid of the transform-origin. I was trying to fix the issue but forgot to remove it before filing the report.

Safari and Firefox [and Edge for non-SVG elements] have common behavior.

Confirmed that Chrome behaves differently, and that this isn't a regression.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/e08eb1644a55fb33111b0734d783e20b7017b413

commit e08eb1644a55fb33111b0734d783e20b7017b413
Author: Kevin Ellis <kevers@chromium.org>
Date: Thu Nov 15 17:38:00 2018

Improve robustness of transform-post-multiplication test.

The transform-post-multiplication testis very sensitive to changes in 2D matrix interpolation. This patch addresses the following:

* Use a prefix that guarantees that matrix and pairwise interpolation product different transforms.
  Previously relying on subtle numerical precision discrepancies.
  With the new prefix, pairwise will scale and transform will rotate.
* Fix matrix comparison to consider the full floating point value and not just the mantissa.
  Previously small non-zero values would be perceived as a matrix mismatch.

Bug:  797472 

Change-Id: I9120f2f92195cf2559ff2eb4294e1341195c277e
Reviewed-on: https://chromium-review.googlesource.com/c/1337639
Reviewed-by: Stephen McGruer <smcgruer@chromium.org>
Commit-Queue: Kevin Ellis <kevers@chromium.org>
Cr-Commit-Position: refs/heads/master@{#608417}
[modify] https://crrev.com/e08eb1644a55fb33111b0734d783e20b7017b413/third_party/WebKit/LayoutTests/animations/transform-post-multiplication.html

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/0688b8a4b05ab68b0fd77b33b10e64e19d767a12

commit 0688b8a4b05ab68b0fd77b33b10e64e19d767a12
Author: Kevin Ellis <kevers@chromium.org>
Date: Thu Nov 15 19:06:58 2018

Improve robustness of wpt animation interpolation tests.

This patch fixes the problem where subtle changes to numerical values in matrix interpolation cause mismatches in the expected results.

To address the issue:
* Styles are checked to see if they are of the form matrix(...) or matrix3d(...).
* If the style is a matrix, then arguments are rounded and the style is reconstructed.

Bug:  797472 

Change-Id: I44d6e0ed13e24dc2ecfeeacbcd4809d6dcdb6465
Reviewed-on: https://chromium-review.googlesource.com/c/1337640
Reviewed-by: Xida Chen <xidachen@chromium.org>
Commit-Queue: Kevin Ellis <kevers@chromium.org>
Cr-Commit-Position: refs/heads/master@{#608466}
[modify] https://crrev.com/0688b8a4b05ab68b0fd77b33b10e64e19d767a12/third_party/WebKit/LayoutTests/external/wpt/css/css-transforms/animation/list-interpolation-expected.txt
[modify] https://crrev.com/0688b8a4b05ab68b0fd77b33b10e64e19d767a12/third_party/WebKit/LayoutTests/external/wpt/css/css-transforms/animation/matrix-interpolation-expected.txt
[modify] https://crrev.com/0688b8a4b05ab68b0fd77b33b10e64e19d767a12/third_party/WebKit/LayoutTests/external/wpt/css/css-transforms/animation/resources/interpolation-testcommon.js

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/cab4d63402cb35f0080387a9ba5ed654ec3ab57b

commit cab4d63402cb35f0080387a9ba5ed654ec3ab57b
Author: Kevin Ellis <kevers@chromium.org>
Date: Mon Nov 19 15:11:13 2018

Update tests of 2-D interpolation.

The patch makes it possible to sanity check the interpolation paths for 2-D matrix transformations.

The decomposition of a 2-D or a 3-D matrix is not unique, and can therefore affect the interpolation path. For example, negative scaling and rotation can be interchanged in some combinations. By restricting the tests to transforms produced by a restricted set of 2-D geometric operations, we can highlight cases where there is a preferred decomposition, and more easily flag regressions.

Bug:  797472 
Change-Id: Ic7d240ae6978e25aded183ea811c76a179f1e16a
Reviewed-on: https://chromium-review.googlesource.com/c/1340941
Reviewed-by: Stephen McGruer <smcgruer@chromium.org>
Commit-Queue: Kevin Ellis <kevers@chromium.org>
Cr-Commit-Position: refs/heads/master@{#609293}
[modify] https://crrev.com/cab4d63402cb35f0080387a9ba5ed654ec3ab57b/third_party/WebKit/LayoutTests/animations/interpolation/transform-interpolation-005.html
[modify] https://crrev.com/cab4d63402cb35f0080387a9ba5ed654ec3ab57b/third_party/WebKit/LayoutTests/animations/interpolation/webkit-transform-interpolation-005.html

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/18d38f290c89dd430c6273a46f4ea21d0a9b7737

commit 18d38f290c89dd430c6273a46f4ea21d0a9b7737
Author: Kevin Ellis <kevers@chromium.org>
Date: Mon Dec 03 19:25:50 2018

Simplify interpolation of 2-D matrix transforms.

The decomposition of a transformation matrix into translations, rotation, scale and skew transforms is not unique. In some cases, the generalized 3-D decomposition does not align with the working draft for CSS transforms (https://drafts.csswg.org/css-transforms/).



In the special case where the transforms being interpolated are both 2-D, a simplified model provides more restricted set of decomposition transforms with less computational overhead.

Bug:  797472 
Change-Id: I2b8ba99fe02c2eef878d94f5dfaea55c39652759
Reviewed-on: https://chromium-review.googlesource.com/c/1332253
Commit-Queue: Kevin Ellis <kevers@chromium.org>
Commit-Queue: Ian Vollick <vollick@chromium.org>
Reviewed-by: Ian Vollick <vollick@chromium.org>
Cr-Commit-Position: refs/heads/master@{#613191}
[modify] https://crrev.com/18d38f290c89dd430c6273a46f4ea21d0a9b7737/third_party/blink/renderer/platform/transforms/DEPS
[modify] https://crrev.com/18d38f290c89dd430c6273a46f4ea21d0a9b7737/third_party/blink/renderer/platform/transforms/transformation_matrix.cc
[modify] https://crrev.com/18d38f290c89dd430c6273a46f4ea21d0a9b7737/third_party/blink/renderer/platform/transforms/transformation_matrix.h
[modify] https://crrev.com/18d38f290c89dd430c6273a46f4ea21d0a9b7737/third_party/blink/renderer/platform/transforms/transformation_matrix_test.cc
[modify] https://crrev.com/18d38f290c89dd430c6273a46f4ea21d0a9b7737/third_party/blink/web_tests/external/wpt/web-animations/animation-model/animation-types/accumulation-per-property-expected.txt
[modify] https://crrev.com/18d38f290c89dd430c6273a46f4ea21d0a9b7737/third_party/blink/web_tests/external/wpt/web-animations/animation-model/animation-types/interpolation-per-property-expected.txt
[modify] https://crrev.com/18d38f290c89dd430c6273a46f4ea21d0a9b7737/third_party/blink/web_tests/external/wpt/web-animations/animation-model/animation-types/property-list.js