|
Forums >
Digital Art and Retouching >
HighPass Sucks (+ solution)
Wow a thousand replies for this thread ! Kiwigrest wrote: What I've read seems overcomplicated and in the end he doesn't say how you may use the the slip in 3 frequencies... Jun 10 11 06:51 pm Link Sean Baker Photo wrote: Not sure if you ever heard back from Ben, but I wrote him about this and he replied: Hi Kurt, I do not use Aperture nor iPhoto (and not too keen on jumping into yet more image processing software!). Maybe someone who does can share anything notable. Jun 21 11 10:03 pm Link Sungoddess Studios wrote: Photoshop doesn't get corrupted from "bad actions". Jun 25 11 11:39 am Link But in a printed magazine, does this really shot? Jun 26 11 02:45 am Link Wow, I've just spent the last few hours reading this thread top to tail. I know I'm a couple of years late to the party but after a few inital tries I can see this is going to revolutionise my workflow! Thanks so much to those involved, particularly Photons 2 Pixels and Sean, who have both put so much great info and time into this thread. I can see you're both still patiently replying to those who can't grasp the basic principle, which takes the patience of a saint when it's been gone over 100 times on this thread! Again - thanks a million. This is the best retouching thread I've ever encountered on the net by a long chalk. Jun 28 11 01:19 pm Link Harry Duns wrote: If you have any questions, the best way to get me is to send me a message. I don't frequent the forums very often anymore. Jun 28 11 07:54 pm Link Photons 2 Pixels Images wrote: That's kind of you but I 'get it' and everything I need to know has been covered in the huge number of previous posts Jun 29 11 03:41 am Link Personally, I'm a fan of High Pass. When used correctly, it's very effective. Jul 01 11 01:35 pm Link Kurt Chambers wrote: Brilliant! Thanks for sharing that with us - I'd gotten a reply from him (he is an incredibly gracious guy), but with far fewer details than would have made for an informative post [it probably helps that you know him!]. It's crazy the way that paths can cross like that. Jul 04 11 12:33 pm Link David A Gentry wrote: And there you've found one of the main points of this thread Jul 04 11 12:33 pm Link The computation should be commutative, right? I.e. original - low pass = high pass, then the following should also be right: original - high pass = low pass Now, let's see how you get the high pass (say, original value is x, lp value is y, and hp value is z): (x-y)/2 + 128 => z (taking 8 bit mode as an example: range for all values is 0~255) Then, can you do the same, substract high pass from original, to get the low pass? (x-z)/2 + 128 => this is not equal to y ! The reason I've got this question: by intuition, I thought low-passed portion and high-passed portion should be "symmetric", the sum of these two parts is the orginal, then these two should be commutative. But the blending mode of "linear light" indicates that the order of those two layers matters. I just don't understand how come these two parts are not "symmetric". Why? BTW, really appreciate your post, really like to see what's really happening behind the "art" part. That's also a reason I'm having that question :-) Jul 20 11 05:42 pm Link This thread is gold. Picked up so much tips here ! Aug 03 11 05:09 pm Link ThePreset wrote: I couldn't agree more with you and here is the fruit of that http://jcdalmanphoto.tumblr.com/post/10 … d-outs-for Sep 27 11 07:14 am Link Somewhere in this thread, pretty sure it was in here and not one of the treads that branched off from it, was a discussion of separating into Luminosity, Color and Detail. I was playing with it for a while and then stopped and would like to play with it again as I have a few images with real shiny models who I think I could more reliably rebuild the skin color using that method... Anyone know where in this thread that is or if it was a different thread altogether? Sep 27 11 08:01 am Link Invaluable help and advice - Thank you so much to the thread starter - along withgetting a wacom tablet this is improving my work tenfold. Nov 10 11 01:15 pm Link Hi First of all, I would like to thank all the recurrent contributors to this thread. This seems to me to be the place with the biggest concentration of ready to use information on the subject. Once again thanks (all the way from the French West Indies) I however have a question. Once I take a year or two to wrap my head around the math involved and mastered the different layer styles and the calculations I might get it myself, but now, I am deeply in need of help. I've a couple of times an image to explain spatial frequencies where a square that looks like the cloud filter + some noise is separated in 6 frequencies. How can I do that with an image, in other words how can I separate an image in 3, 4 or more frequencies. I've tried to refer engineer it, but can't seem to find the proper Apply Image values or/and layer blend modes. Thank you mighty ones Apr 11 12 05:54 am Link Anyone, please ? Apr 29 12 09:06 am Link teddyvestris wrote: Anything in there? Apr 29 12 09:44 am Link teddyvestris wrote: split the image as described for a very narrow frequency; say, one pixel. Then split the low-frequency layer for a higher value, like 4 px. then split the low one again, for a higher value, like 10px. repeat as desired. Apr 29 12 12:12 pm Link Works Great!! Thanks a lot! May 23 12 03:22 pm Link As much as I admire the technical knowledge behind the math, I do not think its mandatory to know this info to achieve the result you are looking for. Photography and Sharpening are subjective. One can easily complicate anything they choose, or simplify a process to meet their needs. Some responses here are very valid and show users how to over-think what should or could be a very simple approach. FS is an often used term and technique, primarily used in cosmetic work to split frequencies. Its inventive to use it as a method to sharpening. In all honestly I have tried these techniques on files of all sizes and find no added value to them over the simplest method of HP for sharpening. If HP is done correctly with selective choosing, its as good as it gets. HP is really just an action of FS anyways. If you use it in moderation for the tones you seek to enhance, and eliminate it where its not needed, that serves my purposes. IE: stamp your image and apply a .5 HP and set to VL or LL and boom, a very nice way to enhance Specular HL's. Or apply a 2 HP and set to SL or OL and you get another range of tones being sharpened. *These settings are of course relative to the size of your file. May 24 12 08:48 am Link if u think this thread is just about sharpening images and the math behind it you have completely missed 99% of the gold that is in this thread.... May 24 12 09:51 am Link Julian Marsalis wrote: I dont and didnt. Its about FS. I get that. May 24 12 10:17 am Link Sony releases SpectraLayers - Spectral Waveform Editing Software (for audio) Divide audio files into discrete, user-defined layers that can be processed independently and reintegrated to achieve perfect master files. Remove unwanted noises and artifacts with ease while leaving surrounding audio data completely intact. http://www.kvraudio.com/news/sony-relea … ware-19586 Jul 20 12 12:37 am Link Have read all the 26 pages this afternoon........wow, some of it too technical for me but BIG respect for Sean. Is there an update to this amazing thread? Aug 22 12 10:58 am Link Marc Turnley wrote: Same here! Great tip! Dec 30 12 12:40 pm Link I am certain after 26 pages someone may have mentioned it and likely I missed it... But... just curious. I'm trying out the method now, ProPhoto 16-bit... and I'm finding my file size is well, getting massive. Because there are 2 methods, based on whether you are in 8-bit vs 16bit, what happens to your HP layer(s) if you started in 16-bit and converted to 8-bit? or vice versa? Sep 09 13 09:39 am Link THRobinson wrote: There are those that will tell you you're risking a life in hell for all the destruction that move will create in your files, and there are people like me, that will tell you the net effect is so small as to hardly be noticeable. About the only time I've seen a benefit to using 16 bit files is when banding in vignettes is an issue. You can also crank on a curve a bit harder in 16 bit before the file disintegrates, but most of my work doesn't require such heavy hands. Sep 09 13 09:47 am Link I've replaced my high pass sharpening and unsharp mask with a more advanced method Group: Overlay{ Inverted Layer -> Vivid Light -> Surface Blur (or any blur) Layer to be Sharpened -> Normal } Layer to be sharpened -> Normal This way all the blur tools become non destructive sharpen tools I've also broken it up sharpening tints and shades seperately by using the darken and lighten blending modes with the linear burn and linear dodge modes. Sep 09 13 10:11 am Link Robert Randall wrote: I would probably never see a change between 8 and 16-bit... not too worried about that. What I was wondering more was, would I have to recreate the 16-bit HF layer using the 8-bit method? or once the layer was made, converting down to 8-bit won't harm the layer and I can continue using the same HF layer like nothing changed. Sep 09 13 10:19 am Link THRobinson wrote: Like I said, unless you see visible artifact, which I doubt you will, there should be no down side. Again, there will be those that think I'm nuts. I assume most of those folks have 30X40 prints hanging at Fahey Klein next to the Herb Ritts display. Sep 09 13 10:23 am Link Ya my main fear was more with the HF layer, and how making the layer differs between an 8-bit image and a 16-bit image... wasn't sure if dropping the image down to 8-bit messes up that layer or once it's been made, it doesn't matter. I'm just practising with an image I got off the retouch section of the forums here, using a few new methods for the skin retouch and using the curves for dodge and burn.... but soon realized my file was 540mb and chugging along very very slowly. Switching to 8-bit dropped it down to 325mb. Older PC, kinda miss my MAC for this stuff. Sep 09 13 10:38 am Link THRobinson wrote: I sometimes have to work for art directors that insist, as a final product, that I supply a complete layered file in 16 bit. I'm working on one right now that is closing in on 4GB. Stupid! Sep 09 13 10:41 am Link Book marked for future reference! Thanks to the op for sharing. Sep 09 13 10:56 am Link . the high pass filter is good to me. Sep 24 13 09:19 pm Link I'm kind of concerned by some results I'm seeing in the 8-bit process. I wonder if I'm doing it wrong? The image below links to a crop of a layered PSD. The original layer is straight from lightroom. The image has been split twice, once for noise and once for detail (and then stacked up with linear light). I see basically zero detail in the (admittedly noisy) shadows.  By contrast, in 16bit, I detect no difference whatsoever, and have to really crank a curve on a difference layer to see any change at all. I know the 16 bit process is more "accurate" but are my results typical? Jul 04 14 02:21 am Link THRobinson wrote: I see a definite brightness shift when I convert a 16bit layered and separated file down to 8 bit. Jul 04 14 02:27 am Link |
Am now busy re-editing old pics with this concept - it's such a powerful technique!
