New GM testing images at half-pixel device coords with nearest filtering
Remove image_scale_aligned and pixel snap GMs. The former tested similar things but is hard to triage, particularly on GPU where we don't control sampling. It also tests various filter modes but only nearest seems interesting in this context. We already have broader testing of image filtering modes. The latter tests how rasterization works for different primitives but is really more informational about how different GPUs behave than it is a test of Skia.
The new GM only runs on CPU. We have no control over how snapping is
handled on GPUs and it varies device to device and we just live with the result.
Bug: skia:13752
Change-Id: I7bcb9164154c5c82f351e551c73e5a9160779d2d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/599349
Reviewed-by: Florin Malita <fmalita@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
5 files changed