John
Hard Mix produces either 255 or zero color values in each RGB channel, depending on the relative values in that channel of the top and bottom layers.
George
John,
Playing further, Hard Mix seems to be even simpler than I thought. Looks like you get 255 in a given color channel when the layer and substrate values in that channel add up to 255 or more. Else, you get zero. This blend seems to be aptly named. There are no gradations. There is an abrupt discontinuous jump from 255 to 0 or vice versa wherever the sum of channel layer and substrate values crosses 255.
Here’s an interesting twist: Fill a square with 127/128/129 (a mid-gray for all practical purposes). Duplicate the layer and set the duplicate (top) to Hard Mix. The result is 0/255/255. So hard-mixing that essentially mid-gray patch with itself produces cyan! Miracle of miracles!
George
Thanks George, I’ll experiment along these lines. It’s an odd but interesting mode – I’ve been experimenting with the same image copied into multiple layers, then changing the layers’ blending modes, and this hard mix can add some amazing distortions to an image. I’d thought it was a combination of dissolve and hard light. It seems left out of the documentation and out of CS’s scripting model.
John
According to <
http://www.tommyblogs.com/PermaLink,guid,a8fbf3d5-f077-4686- a9ff-6c88abd043fa.aspx>
this mode produces a posterized picture consisting of up to eight colors. The writer is interested in this mode’s use for sharpening.
He quotes Julieanne Kost of Adobe.
<
http://www.adobeevangelists.com/pdfs/photoshop/ButWaitPS_CS. pdf>
"An additional blend mode has been added to Photoshop CS called Hard Mix. In general, the results when blending two layers together are: lighter colors lighten the result, darker colors darken the result, lowering the fill opacity creates less posterization/thresholding. Hard Mix-blending a blurred version of an image with itself (for example) will sharpen the image with Fill opacity (on the Layers palette) controlling the sharpening strength."
John
Last time I checked, SHE had blonde hair.
Very tall as well.
John,
"…this mode produces a posterized picture consisting of up to eight colors…"
And the eight possible colors would then have to be black, white, and fully bright and saturated red, green, blue, cyan, magenta. and yellow, all of which are comprised of channel values which are either 255 or zero and nothing else.
I have found another proviso, namely, that when the layer and substrate values in a given channel add up to exactly 255, you’re right on the 255/0 edge. In that case, you get 255 if the substrate value is greater than the layer value and zero if it’s less.
"…I’d thought it was a combination of dissolve and hard light…"
I had the same thought when I got a result that looked like Dissolve blending—I got a field of randomly scattered pixels that were either 255 or 0. I can’t remember the circumstances and have tried everything I can think of to reproduce the effect but to no avail. I’m thinking again that it would have to be a scenario involving that 255/0 edge where substrate and layer add to 255.
George
John,
Ah, now I see how I got a dissolve-like result. It WAS a case of the sum of values hovering at 255, going randomly maybe a point above or below in adjacent pixels to yield a result equal to 255 or 0.
I made a left-to-right gradient going from 0/0/0 to 255/0/0. I superimposed a right-to-left identical gradient on a top layer and set its blend mode to Hard Mix. Theoretically, the sum should have been 255/0/0 at every point, since one layer dropped in value by the same amount the other gained in moving left-to-right.
I said "theoretically" because I used the PS gradient tool to construct the gradients and it is NONLINEAR. So the sums were not exactly 255 at all points. In fact,the incidence of spots increased toward the end where the gradient is least linear.
When I used Filter Factory to establish the gradients, the spotting ceased cold turkey becuase it produces a PERFECTLY LINEAR gradient and all sums were 255.
It is little wonder that Hard Mix produces weird results. I am hard put (no pun) to come up with a good use for it.
George
I also find it hard to see much use but have been experimenting with effects. In areas with more contrast and detail Hard Mix produces a gritty, noisy effect. But smooth tones are evened out and become blocks of colour. I’ve been able to produce some surprising results and would like to be able to achieve them less accidentally! If you’re interested I’ll put a couple online. Thanks for your help George.
John,
I have looked a lot harder <G> at Hard Mix and see that my discription applies for layer and fill opacities of 100%, so I have not done justice to this blend mode. With both opacities at 100%, the effects are gross. When opacity controls are used, you get tonal gradations and the blend becomes a lot more useful.
Fill and layer opacities act quite differently—fill opacity creating posterization and layer opacity creating edges where they don’t exist.
When SELF-BLENDING (one image with its duplicate), Hard Mix acts like a snow plow right down the center of the histogram. The width of the plowed channel at the histogram’s center is proportional to the layer opacity. At 100% opacity, the width is 255—you get pixel values only at 0 and 255. At 10% the channel is 25 values wide and no values exist between 114 and 140. Pixels are pushed into two zones (0-114 and 140-255). Likewise, at opacities of 20,40,…you get plowed openings in the histogram’s center of 50, 75,…The action is like the parting of the red sea as depicted in movies.
Clearly, more homework is needed to fully appreciate this tool. I’m working on it. But we shouldn’t have to. Surely, somewhere, this must be all laid out…all we have to do is find it! 🙂
George
I’m messing around with inverted self blending too. Another interesting aspect is to self blend with a slightly blurred layer – it’s a gradient map, but not as we know.
John
john,
I have worked out the algorithm to include layer opacity. Haven’t looked at fill opacity’s effect other than to establish that it is decidedly different from layer opacity’s effect.
George
