Normalize gradient stops #1046

eqrion · 2017-03-31T18:22:14Z

Gradients can be defined with stops outside the range of [0, 1].
When this happens the gradient needs to be normalized by adjusting
the gradient stops and gradient line into an equivalent gradient
with stops in the range [0, 1].

This allows for proper repeating behavior for gradients because the
range from [first_stop, last_stop] must be repeated, not the range
[0, 1]. This also allows for proper rendering of gradients with stops
beyond the range [0, 1] because the GradientData struct only works
for stops in the range [0, 1].

LinearGradients and RadialGradients are updated to be normalized by
default. The DisplayListBuilder method for RadialGradient is updated
to be simpler and more in line with the CSS representation of a
radial-gradient(). The old method is kept for possible SVG uses in the
future, and replay capabilities.

This change is


        Normalize gradient stops

Gradients can be defined with stops outside the range of [0, 1]. When this happens the gradient needs to be normalized by adjusting the gradient stops and gradient line into an equivalent gradient with stops in the range [0, 1]. This allows for proper repeating behavior for gradients because the range from [first_stop, last_stop] must be repeated, not the range [0, 1]. This also allows for proper rendering of gradients with stops beyond the range [0, 1] because the GradientData struct only works for stops in the range [0, 1]. LinearGradients and RadialGradients are updated to be normalized by default. The DisplayListBuilder method for RadialGradient is updated to be simpler and more in line with the CSS representation of a radial-gradient(). The old method is kept for possible SVG uses in the future, and replay capabilities.


        Don't use an aligned gradient when the gradient line doesn't fill rect

The aligned gradient fast path doesn't correctly fill all of its display item's rect if it's gradient line doesn't intersect the whole rect. This is because extra geometry is not added to fill in that region. This can be fixed, but for correctness now we should use the angle gradient shader for these cases.


        Add gradient reftests


        Bump version

eqrion · 2017-03-31T18:22:59Z

r? @kvark @glennw
I believe this needs a version bump because of the interface change for RadialGradient.

eqrion · 2017-03-31T18:26:46Z

An example for repeating gradients,

root:
  items:
    - type: stacking-context
      bounds: 50 50 300 300
      items:
        - type: gradient
          bounds: 0 0 300 300
          start: 0 150
          end: 300 150
          stops: [0.1, blue, 0.2, blue, 0.2, red, 0.3, red]
          repeat: true

After,

Before,

An example for clamped,

---
root:
  items:
    - type: stacking-context
      bounds: 50 50 300 300
      items:
        - type: radial-gradient
          bounds: 0 0 300 300
          center: 150 150
          radius: 150 150
          stops: [0.0, red,
                  0.1, red,
                  0.1, blue,
                  0.2, blue,
                  0.2, red,
                  0.3, red,
                  0.3, blue,
                  0.4, blue,
                  0.4, red,
                  0.5, red,
                  0.5, blue,
                  0.6, blue,
                  0.6, red,
                  0.7, red,
                  0.7, blue,
                  0.8, blue,
                  0.8, red,
                  0.9, red,
                  0.9, blue,
                  1.0, blue,
                  1.0, red,
                  1.1, red,
                  1.1, blue,
                  1.2, blue,
                  1.2, red,
                  1.3, red,
                  1.3, blue,
                  1.4, blue]

After,

Before,
GradientEntries would only hold entries up to 1.0, and the rest would be clamped to blue.

kvark

Looks good overall, just one question about the tests.

kvark · 2017-03-31T19:31:17Z

webrender_traits/src/display_list.rs

+    // of the gradient line to where stop[0] and the end of the gradient line
+    // to stop[n-1]. this function adjusts the stops in place, and returns
+    // the amount to adjust the gradient line start and stop
+    fn normalize_stops(stops: &mut Vec<GradientStop>,


should use &mut [GradientStop] instead of Vec

Yes, that is better. I will update with this.

Ah, wait normalize_stops can in the bad case clear() and add() to the list. That's why it's a Vec.

kvark · 2017-03-31T19:31:17Z

webrender_traits/src/display_list.rs

+    // the amount to adjust the gradient line start and stop
+    fn normalize_stops(stops: &mut Vec<GradientStop>,
+                       extend_mode: ExtendMode) -> (f32, f32) {
+        assert!(stops.len() >= 2);


nit: will technically work for stops.len()==1 too, and will crash on first() on zero, so perhaps this assert can be removed?

Yes, I suppose it could. The motivation was mostly because CSS gradients always have at least two stops so it's nice to enforce that. Not sure on SVG. Regardless, it's not really necessary.

kvark · 2017-03-31T19:31:17Z

wrench/reftests/gradient/reftest.list

+
+# fuzzy because of differences from normalization
+# this might be able to be improved
+fuzzy(255,1200) == repeat-linear.yaml repeat-linear-ref.yaml


Why using the difference of 255 here?

Normalizing the stops for these tests is causing some differences in stop positions. You can see the results of running without fuzzing by pasting below into here. It is kind of significant, so I am unsure whether this is a problem or unavoidable. We are normalizing the stops with the same math that Gecko uses, if that helps.

REFTEST TEST-UNEXPECTED-FAIL | reftests/gradient/repeat-linear.yaml == reftests/gradient/repeat-linear-ref.yaml | image comparison, max difference: 255, number of differing pixels: 1200 REFTEST IMAGE 1 (TEST): data:image/png;base64,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 REFTEST IMAGE 2 (REFERENCE): data:image/png;base64,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 REFTEST TEST-END | reftests/gradient/repeat-linear.yaml == reftests/gradient/repeat-linear-ref.yaml

If the new code path matches gecko, why don't we update the reftests to match it?

kvark · 2017-03-31T20:43:40Z

according to IRC discussion:

test "repeat-linear" failure is weird
need to investigate with regards to Gecko tests
unlikely to break Servo tests, since using non-normalized gradient stops didn't work before anyway
thus, should be safe to merge now, then follow-up

@bors-servo r+

bors-servo · 2017-03-31T20:43:41Z

📌 Commit 398db7c has been approved by kvark

bors-servo · 2017-03-31T20:43:43Z

⌛ Testing commit 398db7c with merge 6d5f80f...


        Auto merge of #1046 - rlhunt:gradients-2, r=kvark

Normalize gradient stops Gradients can be defined with stops outside the range of [0, 1]. When this happens the gradient needs to be normalized by adjusting the gradient stops and gradient line into an equivalent gradient with stops in the range [0, 1]. This allows for proper repeating behavior for gradients because the range from [first_stop, last_stop] must be repeated, not the range [0, 1]. This also allows for proper rendering of gradients with stops beyond the range [0, 1] because the GradientData struct only works for stops in the range [0, 1]. LinearGradients and RadialGradients are updated to be normalized by default. The DisplayListBuilder method for RadialGradient is updated to be simpler and more in line with the CSS representation of a radial-gradient(). The old method is kept for possible SVG uses in the future, and replay capabilities.  --- This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/webrender/1046)

bors-servo · 2017-03-31T21:10:18Z

☀️ Test successful - status-travis
Approved by: kvark
Pushing 6d5f80f to master...

eqrion added 4 commits Mar 29, 2017

Add gradient reftests

e3a12af

Bump version

Loading status checks…

398db7c

kvark approved these changes Mar 31, 2017

View changes

bors-servo merged commit 398db7c into servo:master Mar 31, 2017
3 checks passed

3 checks passed

continuous-integration/appveyor/pr AppVeyor build succeeded
Details

continuous-integration/travis-ci/pr The Travis CI build passed
Details

homu Test successful
Details

bors-servo mentioned this pull request Mar 31, 2017

Proper dither re-enabled #1045

Merged

glennw mentioned this pull request Apr 3, 2017

Gradient reftest regression #1051

Closed

eqrion deleted the eqrion:gradients-2 branch Apr 6, 2017

servo / webrender

Normalize gradient stops #1046

Normalize gradient stops #1046

eqrion commented Mar 31, 2017 •

edited by larsbergstrom

eqrion commented Mar 31, 2017

eqrion commented Mar 31, 2017

kvark left a comment

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

kvark commented Mar 31, 2017

bors-servo commented Mar 31, 2017

bors-servo commented Mar 31, 2017

bors-servo commented Mar 31, 2017

servo / webrender

Join GitHub today

Normalize gradient stops #1046

Normalize gradient stops #1046

Conversation

eqrion commented Mar 31, 2017 • edited by larsbergstrom

eqrion commented Mar 31, 2017

eqrion commented Mar 31, 2017

kvark left a comment

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

kvark commented Mar 31, 2017

bors-servo commented Mar 31, 2017

bors-servo commented Mar 31, 2017

bors-servo commented Mar 31, 2017

eqrion commented Mar 31, 2017 •

edited by larsbergstrom