Action for Frequency Separation

[jonathan_k]

Hey All

I want to create a basic Photoshop CS5 action to speed up frequency separation.

Basically, for frequency separation, what I do is make two copies of the image, the lower one of which contains the blurred tonal/colour section of the image, and upper one which contains the details of the image. In addition to the gaussian blur function, the apply image function is used.

Now, the level of the gaussian blur is always variable. Sometimes the blurred layer needs to be very blurred, sometimes not. All depends on the original image.

What I want is to start running the action, but have Photoshop let me select the level of blur, and then have it continue running the action.

Any ideas?

Any help would be much appreciated.

Thank you
Jon

[4personnen]

Hi
While recording your action, at the "moment" you want to place your blur menu, go to the drop down menu of the action palette and "insert menu item" choose your blur filter and click "OK".

[jonathan_k]

Thank you!

Will try when I get home later.

So simple.

Have a great day.

Jon

[mistermonday]

Jon, RP member Chain has already written such an action which I have attached below. There are actually two types of Frequency Separation actions included. The 1st is the standard one you referred to in your OP. The 2nd splits performs a frequency separation in which there is absolutely no color on the high frequency layer. All of the color is on the LF layer.
Just select the either of the two and click your Play action button.
Regards, Murray

[jonathan_k]

Perfecto. Thank you. I suppose thats basically the ideal solution, and the funny thing is that I didn't even know there would be colour contained in the HF layer at all. Or that it would an issue.

Double save Murray, thanks.

[pixeltek]

jonathan, if you set the layer as a smart object you can then apply the blur during your action ... when you play the action back it won't stop and you'll be able to adjust the blur once it's finished.

i hope that makes sense.

[jonathan_k]

Yeah, makes perfect sense, will probably have to mess about with it though. Thanks yo..

[4personnen]

The fact that using a smart object for LowF would be an annoyance as the brush/healing brush job will only be permitted once the layer opened as a document on its own (and you'll have to save to go back to your main document). Or maybe you're just working on the HF layer. Or maybe it changed with CS5
My2cents...
I am not sure I make myself clear

[pixeltek]

true about cloning/healing, etc., on the smart object but my workflow is such that i typically have all of that done before i do the hf/lf and am just using it as an overall evening out process.

[jonathan_k]

I have a follow up question if you all happen to have the time.

It would appear from discussion and from the actions posted that it is from from ideal to do a frequency separation with 8 bit images. My question then is this;

Is there any major downside to converting an image from 8-bit to 16-bit?

I get the advantages of the 16-bit workflow, but yeah. Any downsides that you can think of?

Thanks again

Jon

[mistermonday]

Originally Posted by jonathan_k

Perfecto. ... and the funny thing is that I didn't even know there would be colour contained in the HF layer at all. Or that it would an issue.

Double save Murray, thanks.

A High Pass or High Freq layer consists of edges, at and below the radius you have specified, whose contrasts have been increased, and the rest of that layer is grayscale ~128/128/128. You can verify this by dragging the Saturation slider all the way to the right. The GB or LF layer has had a reduction in contrast on those same edges as the LF layer. Individually those layers appear blurred and sharpened respectively. When you mathematically add the two layers together, you get the orig image. GB+HP=Orig. Having those edges in the HF maintain color is not necessarily a bad thing depending on what you are healing / cloning. The asymetric split where there is no color in the HF layer can be beneficial but the effect is different as you will see when you play with it. There will be situations where you do not want the texture to be too prominent.

Is there any major downside to converting an image from 8-bit to 16-bit?

There is no value in converting from 8 to 16 bit, in fact it is counter-productive because the data that was there in the original 16 bit file has already been destroyed. An analogy would be as follows:
You started with two number: 62.7854 and 23.5692. You rounded off the numbers to 62 and 23. Next you decided to perform some multiplication and because you wanted a great deal of accuracy, you made the numbers 6 decimal places > 62.000000 and 23.000000. Guess what - zero x zero = zero.
When you convert an 8 bit value to 16 bit you are just doubling your file size and not adding any value. My analogy is somewhat oversimplified but you get the point.
If you have an orig 16 bit file, it is always better to work with it from beginning to end in 16 bit and then spin off an 8 bit version for print or other 8 bit output.
Regards, Murray

[4personnen]

Be also sure that If you use the "apply image" method, to check the values of scale, offset and the blending mode which are not the same for 16 bits and 8 bits images...

[jonathan_k]

So. I've recorded an awesome 8-bit action and its all working swimmingly. And if I want to, I could do the same for the 16-bit. Thank you all for the help. Much appreciated.

[Chennae]

Probably annoying now that you've already done it, but if you look at this thread - http://www.modelmayhem.com/po.php?th...=636374&page=1 - and read through how to use the action, it's a real time saver!