I would like to have a gradient changing smoothly the color in the x direction and the saturation y direction, in order to have a nice, highly saturated spectrum on the left part and something almost uniformely gray on the right.
I tried to edit a circular gradient, but when I add and modify a stop point, let's say by selecting one on the x-axis, it changes the colors of the y-axis stop point

Any clue?

And also, is it possible to have orthoradial gradients? color depending of an angle, not a distance...

Sorry for my bad english, I need to paractice a bit more

didier973 wrote:I would like to have a gradient changing smoothly the color in the x direction and the saturation y direction, in order to have a nice, highly saturated spectrum on the left part and something almost uniformely gray on the right.

This sounds like a linear gradient and should be easy to setup as one. In the next line though you mention a circular (radial?) gradient:

I tried to edit a circular gradient, but when I add and modify a stop point, let's say by selecting one on the x-axis, it changes the colors of the y-axis stop point

The x and y axes in a radial grad have the same gradient colours but can be scaled independently.

And also, is it possible to have orthoradial gradients? color depending of an angle, not a distance...

Conical or sweep gradients are not yet possible in Inkscape but I am sure someone is working on it. You can try using interpolation as a substitute but it might get a little processor intensive.

The best method of implementing Freeform gradients (beyond linear and radial) is to use Gaussian Blur. Some artist consider linear and radial gradients as outdated and should only be used as a last resort. Using Gaussian Blur as your default method of implementing gradual color mixing, will definitely enhance your artwork.
see the manual for more information
http://tavmjong.free.fr/INKSCAPE/MANUAL ... nBlur.html
At times using Gaussian Blur with Interpolation provides an easy and quick method of applying a freeform gradient.
http://inkscape.org/doc/interpolate/tut ... te.en.html

One linear gradient is ok, using the editor I could obtain whatever gradient I tried... Here, I would like to have different gradients depending from direction. If you go from top to bottom, color should change, with the same hue and saturation (like in a "straight rainbow"), from left to right, red should remain red, but saturation should decrease, from a "pure" red down to a reddish gray.
The nice, intuitive way in my opinion is to allow to edit independently all the "stop" nodes of a circular gradient. I think this is not (yet?) possible.
But it may be possible to combine 2 different gradient that i didn't discover "add/multiply/fusion"... of colors from differnet layers/objects, if I could do it now, it would be nice!

Blend modes via the Layer feature

http://tavmjong.free.fr/INKSCAPE/MANUAL ... ters-Blend

http://fedev.blogspot.com/2008/04/filte ... blend.html

Here, I would like to have different gradients depending from direction. If you go from top to bottom, color should change, with the same hue and saturation (like in a "straight rainbow"), from left to right, red should remain red, but saturation should decrease, from a "pure" red down to a reddish gray.

I am not sure without knowing exactly what your purpose is but you can probably get this effect by overlaying two linear gradients perpendicular to each other and masking the top gradient with a gradient filled mask. Complicated. You can add to the complication by putting the masked object in its own layer with a different blend mode.

The nice, intuitive way in my opinion is to allow to edit independently all the "stop" nodes of a circular gradient.

I see what you are thinking of - x-axis one set of colors and y-axis one set of colors, but then what about along the line at a 45 degree angle? How would the color of a point on that line be determined? The stop nodes of a radial gradient are not two independent series of nodes - they could be just as easily only shown on one axis with the other axis represented only by the final point. That is to say, in a radial gradient each node represents the division of the ellipse/circle into concentric ellipse/circles.
I totally agree that conical gradients and gradient meshes (which are really blurs..) would be very welcome in Inkscape. They are probably one of the most requested features. I am sure they are coming soon... (hint to developers)

druban wrote:I see what you are thinking of - x-axis one set of colors and y-axis one set of colors, but then what about along the line at a 45 degree angle? How would the color of a point on that line be determined?

With an orthoradial gradient between the corresponding stop points of the x and y axis... One example of what i would like to get is the filling you have in the different kind of "color picker", but maybe even more flexible: colors are defined with 3 independent parameter+ transparency, in a color picker you usually set transparency to 1, and intercepted the 3 dimensionnal space remaining by a plane (R remain constant, saturation remain constant...). In a generalization of this, you should have the color you get by intersecting the 4 dimension-space by a 2D surface

I almost succeeded for my illustration thanks to the interpolation, in order to decrease saturation, I'm mixing more and more almost transparent layers. The bad thing is that the resulting colors depends on the layer order...