CSS gradient syntax: comparison of Mozilla and WebKit (Part 2)

Warning This article was written over six months ago, and may contain outdated information.

Update: I wrote this arti­cle in 2009. In ear­ly 2011 WebKit decid­ed to change their syn­tax to match that used in Fire­fox (and the W3C spec­i­fi­ca­tion). The syn­tax con­tained in these arti­cles will be main­tained for rea­sons of back­wards-com­pat­i­bil­i­ty, but you should use the new syn­tax for the future. I’ve writ­ten a post about the new radi­al gra­di­ent syn­tax.

In the first part of this post I gave a pot­ted his­to­ry of the dif­fer­ing syn­tax­es, and pro­vid­ed an overview of how that affect­ed lin­ear gra­di­ents. In this sec­ond part I’m going to look at radi­al gradients.

Here the syn­tax­es diverge slight­ly more, with WebKit requir­ing more val­ues than Mozil­la; while that adds some flex­i­bil­i­ty, it also increas­es the complexity.

I‚Äôll be using the same page of exam¬≠ples, which you‚Äôll need to view in Safari / Chrome (or oth¬≠er WebKit deriv¬≠a¬≠tive) or Fire¬≠fox 3.6 (beta 3+):

CSS Gra­di­ents com­par­i­son: Mozil­la & WebKit

Radial gradients

As with lin¬≠ear gra¬≠di¬≠ents, the key dif¬≠fer¬≠en¬≠ti¬≠a¬≠tion between the Mozil¬≠la syn¬≠tax and the WebKit syn¬≠tax is that the lat¬≠ter requires a start and end point, where¬≠as the for¬≠mer is con¬≠strained by the bound¬≠ing box.

You can see in this first exam­ple (Radi­al 1) that the Mozil­la syn­tax is much short­er when mak­ing a sim­ple two-colour radi­al gradient:

-moz-radial-gradient(green, yellow);
-webkit-gradient(radial, center center, 0, center center, 70.5, from(green), to(yellow));

In Mozil¬≠la‚Äôs case I set only start and end color-stops; for WebKit I must spec¬≠i¬≠fy a posi¬≠tion (50% 50%) and radius (0 ‚ÄĒ I‚Äôll explain that lat¬≠er) for the start point, and a posi¬≠tion (50% 50%) and radius (70.5) for the end point, as well as the color-stops.

As far as I can see the radius val¬≠ue has to be a val¬≠ue in pix¬≠els, so in order for the end radius to be the diag¬≠o¬≠nal of the square (where Mozil¬≠la defaults to) you need to use the cal¬≠cu¬≠la¬≠tion (side)(sqrt(2)) ‚ÄĒ or, do what I did and use this online cal¬≠cu¬≠la¬≠tor.

In the next exam¬≠ple (Radi¬≠al 2) I‚Äôve off¬≠set the cen¬≠ter of the gra¬≠di¬≠ent and set it to end at the fur¬≠thest edge of the con¬≠tain¬≠ing box:

-moz-radial-gradient(40% 40%, farthest-side, green, yellow);
-webkit-gradient(radial, 40% 40%, 0, 40% 40%, 60, from(green), to(yellow));

And then used the same cen­ter off­set but con­strained the radius to the dis­tance of the near­est wall (Radi­al 3):

-moz-radial-gradient(40% 40%, closest-side, green, yellow);
-webkit-gradient(radial, 40% 40%, 0, 40% 40%, 40, from(green), to(yellow));

As you can see, Mozil¬≠la takes a series of con¬≠stants using nat¬≠ur¬≠al lan¬≠guage ‚ÄĒ farthest-side, closest-side, contain, etc ‚ÄĒ to set the lim¬≠its of the gra¬≠di¬≠ent, where WebKit accepts only pix¬≠el val¬≠ues. The advan¬≠tage to the for¬≠mer approach is that the syn¬≠tax is sim¬≠pler and no cal¬≠cu¬≠la¬≠tions are required; the dis¬≠ad¬≠van¬≠tage is that if you want to cre¬≠ate a radi¬≠al gra¬≠di¬≠ent that is small¬≠er than the lim¬≠its of the con¬≠tain¬≠ing box, you have to com¬≠bine it with the background-size property.

Next (Radi­al 4) I’ve set the cen­ter of the gra­di­ent to the top-right cor­ner, using three colours equal­ly dis­trib­uted; here’s where you can start to see the WebKit syn­tax start to become real­ly unwieldy:

-moz-radial-gradient(right top, green, yellow, blue);
-webkit-gradient(radial, right top, 0, right top, 141, from(green), color-stop(50%, yellow), to(blue));

With WebKit I again have to spec­i­fy start and end points (with radii), and also spec­i­fy the stop posi­tion of the mid­dle colour. Both out­comes are the same, but the Mozil­la syn­tax is sig­nif­i­cant­ly easier.

One aspect of WebKit’s syn­tax which allows for more flex­i­bil­i­ty in a gra­di­ent is the start point and end point; by pro­vid­ing two sep­a­rate val­ues you can set the start gra­di­ent at a dif­fer­ent point to the end gra­di­ent, as well as pro­vid­ing dif­fer­ent radius val­ues, allow­ing for effects that Mozil­la can’t eas­i­ly repli­cate (Radi­al 5):

-moz-radial-gradient(60% 60%,circle contain,yellow,green 75%,rgba(255,255,255,0));
-webkit-gradient(radial,45% 45%,5,60% 60%,40,from(yellow),color-stop(75%, green),to(rgba(255,255,255,0)));

You can see how that is ren¬≠dered here (Mozil¬≠la on the left, WebKit on the right):

Comparison of CSS gradients

Achiev­ing the same effect is only pos­si­ble in Mozil­la by using mul­ti­ple val­ues on the background-image prop­er­ty, along with background-size (Update: This isn’t nec­es­sary; see Tab Atkins Jr’s com­ment, below).


While the (orig¬≠i¬≠nal) WebKit syn¬≠tax does allow for a few effects that the sim¬≠pler Mozil¬≠la imple¬≠men¬≠ta¬≠tion can‚Äôt eas¬≠i¬≠ly copy, I think these are real¬≠ly edge cas¬≠es and the sim¬≠plic¬≠i¬≠ty of the new¬≠er syn¬≠tax is more than ample com¬≠pen¬≠sa¬≠tion. It seems the CSS WG agree, which is why the sim¬≠ple syn¬≠tax is to become an offi¬≠cial pro¬≠pos¬≠al; I hope the WebKit team accept the pro¬≠pos¬≠al and imple¬≠ment it soon.

It was sug­gest­ed that I also com­pare these with Inter­net Explor­er’s Gra­di­ent fil­ter, but that’s a brows­er-spe­cif­ic imple­men­ta­tion that has no chance of becom­ing a stan­dard, so I did­n’t feel it was suit­able for this arti­cle; per­haps in a future extension.

Update: Just after I fin­ished this arti­cle, Mozil­la Hacks pub­lished an in-depth look at the new syn­tax.

17 comments on
“CSS gradient syntax: comparison of Mozilla and WebKit (Part 2)”

  1. You don’t need to use back­ground-size to get a gra­di­ent of a dif­fer­ent size in the cur­rent syn­tax. Just set your col­or-stops appro­pri­ate­ly, espe­cial­ly the last one. If your last col­or stop is at 50%, then the gra­di­ent will be half the size of normal.

    And yeah, I sort of like the skew effect you can get with the orig¬≠i¬≠nal Webkit syn¬≠tax, but it turns out to com¬≠pli¬≠cate the syn¬≠tax too much. I val¬≠ued the abil¬≠i¬≠ty to do gra¬≠di¬≠ents that depend¬≠ed on the box size more than I val¬≠ued the abil¬≠i¬≠ty to do a skew like that. The orig¬≠i¬≠nal drafts of the syn¬≠tax did have that abil¬≠i¬≠ty, it just grad¬≠u¬≠al¬≠ly dropped out as we hacked on it in the mail¬≠ing list.

  2. Thanks for cor¬≠rect¬≠ing me, Tab; I‚Äôve updat¬≠ed the post to reflect your feed¬≠back. Good work on sim¬≠pli¬≠fy¬≠ing the syn¬≠tax, by the way.

  3. Social com¬≠ments and ana¬≠lyt¬≠ics for this post‚Ķ

    This post was men­tioned on Twit­ter by angelus12: RT @stopsatgreen CSS gra­di­ent syn­tax: com­par­i­son of Mozil­la and WebKit (Part 2) | Bro­ken Links http://bit.ly/5jHxQt…

  4. Hi Peter,

    Unre­lat­ed to the arti­cles (which were very well explained by the way), but when I first tried access­ing your site in Webkit night­ly on a Mac, it crashed my brows­er after warn­ing me about down­load­ing Grablau Sans. Each time I tried to restart the brows­er it would crash again after try­ing to load the page. So I switched to Safari and got the same problem.

  5. Hel¬≠lo Peter, 

    Well done on these two arti¬≠cles. Now time to have fun with it.

  6. @ John ‚ÄĒ Thanks for let¬≠ting me know. I can‚Äôt repli¬≠cate the prob¬≠lem myself, and oth¬≠er Safari users are vis¬≠it¬≠ing the site with no prob¬≠lems. What ver¬≠sion are you using? I‚Äôll try to repli¬≠cate the prob¬≠lem myself.

    @ Jason ‚ÄĒ Thanks very much. Let me know if you cook up any cool stuff.

  7. Tab,

    Please explain how the Webkit syn¬≠tax does not have ‚Äúthe abil¬≠i¬≠ty to do gra¬≠di¬≠ents that depend¬≠ed on the box size.‚ÄĚ It cer¬≠tain seems to have that abil¬≠i¬≠ty to me.

  8. Tab, I see now, sor¬≠ry. In Part 1 it is made clear all the cool dif¬≠fer¬≠ent box-mod¬≠el-aware siz¬≠ing options in the new syn¬≠tax. Very cool.

  9. @John Faulds: After inves¬≠ti¬≠gat¬≠ing this fur¬≠ther, seems like Safari does¬≠n‚Äôt like the font-weight prop¬≠er¬≠ty inside @font-face; I need to inves¬≠ti¬≠gate this a lit¬≠tle more.

  10. [‚Ķ] CSS gra¬≠di¬≠ent syn¬≠tax: com¬≠par¬≠i¬≠son of Mozil¬≠la and WebKit (Part 2) [‚Ķ]

  11. [‚Ķ] There are two great arti¬≠cles on this top¬≠ic, delver deep¬≠er into the syn¬≠tax dif¬≠fer¬≠ences: CSS gra¬≠di¬≠ent syn¬≠tax: com¬≠par¬≠i¬≠son of Mozil¬≠la and WebKit and CSS gra¬≠di¬≠ent syn¬≠tax: com¬≠par¬≠i¬≠son of Mozil¬≠la and WebKit (Part 2). [‚Ķ]

  12. Hey these two parts are awesome.
    I real¬≠ly like it and read it more than 5 times to get every¬≠thing in my head.
    I total¬≠ly agree with you con¬≠clu¬≠sion, but there is one prob¬≠lem I can¬īt solve at the moment.
    You said, with a lit­tle help of TAB that fire­fox is able to show the right result with anoth­er gradient.
    But i am unable to put 2 gra­di­ents in the back­ground-image property.
    How would it look like!


  13. Hi, thanks for your kind com­ments. You can use mul­ti­ple back­ground images in Fire­fox 3.6; you just need to use com­ma-sep­a­rat­ed val­ues for the property:

    background-image: gradient, gradient;

    Just make sure your copy of Fire­fox is updat­ed to 3.6.

  14. While I cer¬≠tain¬≠ly¬≠think that Mozil¬≠las approach is eas¬≠i¬≠er, there‚Äôs one thing that Webkit can do but I havn‚Äôt seen or fig¬≠ured out to do in Fire¬≠Fox, and that‚Äôs ‚Äúcom¬≠pound‚ÄĚ gra¬≠di¬≠ents. Can I have mul¬≠ti¬≠ple gra¬≠di¬≠ents lay¬≠ered upon each oth¬≠er, like the WebKit exam¬≠ple imaged here (the top top): http://i2.sitepoint.com/g/nl/tt/cssgradients.jpg

    In WebKit this is done this way (where three gra¬≠di¬≠ents are lay¬≠ered on top of each other):
    back¬≠ground: ‚ÄĎwebkit-gradient(properties 1), ‚ÄĎwebkit-gradient(properties 2), ‚ÄĎwebkit-gradient(properties 3);

    Is there a way to do this in Firefox?

    Henrik Wannheden [May 25th, 2010, 23:30]

  15. You can cer­tain­ly do that in Mozil­la browsers:

    background: { -moz-radial-gradient(20% 50%,circle,white,black 95%,transparent), -moz-radial-gradient(50% 20%,circle,black,white 95%,transparent), -moz-radial-gradient(10% 10%,circle,white,black 95%,transparent); }

    The trick, as you can see, is to remem¬≠ber to set the last colour of each gra¬≠di¬≠ent to ‚Äėtrans¬≠par¬≠ent‚Äô; if you don‚Äôt, the last set colour will fill the rest of the box and oth¬≠er gra¬≠di¬≠ents will be hidden.

  16. […] (ref­er­ence: http://www.broken-links.com) […]

  17. This is one object where the W3C team has nev¬≠er real¬≠ly addressed the issue of gra¬≠di¬≠ent col¬≠ors. It would be great if they would devel¬≠op a stan¬≠dard as it would remove a ‚Äėlot‚Äô of graph¬≠ic work, improve page load times and be more in line with the HTML5-CSS3 stan¬≠dards. Guess we‚Äôll have to wait. :(