ElDeveloper
commented
4 years ago Thanks for tracking this down. Yes, it would be very helpful if you can put together a PR.
On (Oct-13-20|20:03), Marcus Fedarko wrote:
The actual coloring of samples is fine, I think, but the SVG gradients produced sometimes cover only ~99% of the domain, not 100%. This means that the end colors shown in the gradients are very slightly incorrect.
It's pretty confusing, because right now there are two tests that test the gradient SVG creation:
In one of these tests this error is present. The returned Viridis gradient, which is used as the reference for verifying the result, ends at 99% (with
#f8e725), rather than at 100% (#fee825, which is the actual end color for Viridis).In the other one of these tests, this error does not seem to be present: the color map being tested is Blues, and the minimum and maximum colors (
#f7fbffand#08306b) match up with the extrema for this color map (going off whatchroma.scale('blues')(0)andchroma.scale('blues')(1)return).I think the problematic part of code is within this block:
The first for loop seems suspicious to me. I'm not completely sure yet what's wrong with this code, but it looks like there might be precision issues with the
s <= maxcheck: since the data used for the test that failed had astepof(4 - 0) / 100 = 0.04, while the data used for the test that succeeded had a relativelystepof(50 - 0) / 100 = 0.5.I think it would make the most sense to just remove some of this code and use
chroma.scale.colors()to do the work for us:var stopColors = interpolator.colors(101); // Now we can construct gradientSVG just like we did before, with the final color (corresponding // to the maximum value) being placed at 100%Can put together a PR for this tomorrow if you'd like.
This impacts Empress as well, since the gradient SVG code there was adapted from this code.
-- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://urldefense.com/v3/__https://github.com/biocore/emperor/issues/788__;!!Mih3wA!VeKZkpvzEILHoclhWPbUFi5mPuFp3ezfo9QqGvYAozXbwgUZKc2TQB04BH9AxT4$
The actual coloring of samples is fine, I think, but the SVG gradients produced sometimes cover only ~99% of the domain, not 100%. This means that the end colors shown in the gradients are very slightly incorrect.
It's pretty confusing, because right now there are two tests that test the gradient SVG creation:
In one of these tests this error is present. The returned Viridis gradient, which is used as the reference for verifying the result, ends at 99% (with
#f8e725), rather than at 100% (#fee825, which is the actual end color for Viridis).In the other one of these tests, this error does not seem to be present: the color map being tested is Blues, and the minimum and maximum colors (
#f7fbffand#08306b) match up with the extrema for this color map (going off whatchroma.scale('blues')(0)andchroma.scale('blues')(1)return).I think the problematic part of code is within this block:
https://github.com/biocore/emperor/blob/b625a5403ce9ad0ef3a84c6bc0b68b33e97888a4/emperor/support_files/js/color-view-controller.js#L426-L439
The first for loop seems suspicious to me. I'm not completely sure yet what's wrong with this code, but it looks like there might be precision issues with the
s <= maxcheck: since the data used for the test that failed had astepof(4 - 0) / 100 = 0.04, while the data used for the test that succeeded had a relatively largerstepof(50 - 0) / 100 = 0.5.I think it would make the most sense to just remove some of this code and use
chroma.scale.colors()to do the work for us:Can put together a PR for this tomorrow if you'd like.
This impacts Empress as well, since the gradient SVG code there was adapted from this code.