Less than impressed with Smart Sharpening

Discussion in 'Digital Darkroom' started by bill_tuthill, Dec 29, 2009.

  1. I have been reading about SS because GIMP now has a plug-in for it. I suspect high-res LCD monitors have changed
    the art of sharpening. Many of the online tutorials seem to make an image worse after sharpening, I find. Maybe
    the images are improved on a CRT monitor, but CRTs are rare now. Of course sharpening is useful for printing, but that is not what interests me now.

    It is stated in several places that Remove > Gaussian Blur is similar to USM, so people do not recommend it; they
    use Remove > Lens Blur instead. Do you agree? I really cannot see much difference between them, except that Gaussian Blur seems a bit stronger. That is, Gaussian Blur 90% = Lens Blur 100%.

    The Remove > Motion Blur option does not seem to work in CS2. Did it start working in later versions? (Over a
    year ago I posted about this and FocusMagic, which does work.)

    Does Smart Sharpen have a threshold function as in USM? I find threshold very useful, but I do not see an option
    for it in CS2.
     
  2. Bill -
    Ditto. I know some guys who really like Smart Sharpen, but after a few attempts to use it, I've pretty much given up on it, and now completely ignore it. I can do everything I need with Focus Magic (as you mentioned), USM, and occasionally Topaz Sharpen.
    Obviously, I'm not the guy to answer your questions about this tool.
    Tom M
    PS - In the interest of full disclosure, I haven't tried it in CS4. I'm still using CS3.
     
  3. I still prefer to use an edge mask then just apply the normal unsharp filter. As well, I'll use the history brush to either take some of it out or put it in as I see the need.
     
  4. http://www.espacephotos.com/photonet/TUTORIAL/smartsharpen_eng.pdf
    Try this. Patrick kindly provided this tutorial on Smart Sharpening which I now use for 90% of my capture and creative sharpening.
     
  5. Bill:
    If you're just wanting to sharpen for display on your monitor, enter a lower percent. Sharpening for your own monitor is easer than for print. If it doesn't look right, do it over. You have instantaneous feedback. :)
    Eric
     
  6. Bill, for me Smart Sharpen is the best thing Adobe put in Photoshop for serious user who are very dmeanding..like me.
    Heres a quick example of a difference you will achieve from your regular highpass / usm technique that require fidling around numbers and trial and error..and the simple 3 setting Smart Sharpen.
    You cant be disapoint with something you dont fully understand, can you? ; )
    tell me from this example that Samrt Sharpen is deceiving....
    by the way, nothing on the market i have seen give better result as fast and as simple. Of course many plugin are out there that with a push of a button you get a spectacular result, but its not free, and it wont be better than this... so since you alreayd paid a good amount of $ to get Photoshop..why dont use it correctly ; )
    As another user said, i have a quick free tutorial that will certainly change the way you use SS, and it is well explain why you also need 3 step in your sharpening process (if you go to print, if not, then just 2 step could be use)
    let me know.
    00VN1G-204873584.jpg
     
  7. Bill,
    As you can see, Patrick gets great results with Smart Sharpen. I’m personally a fan of high pass sharpening for lots of images. (In case you don’t know: duplicate the layer, set the mode to overlay, and run the High Pass filter with a radius of about 1 pixel. Use smart filters and it’s easy to tweak.) Some images do better with one or the other. And, in some cases, some parts of an image do better with one or the other (or none at all), in which case you wind up doing both and masking out selected parts.
    There isn’t any magic button you can press for this. You simply have to experiment with lots of options. After a while you get better and faster at it.
    Cheers,
    b&
     
  8. Ben, with all respect, i think that when you have find the correct settign in Smart Sharpen, witch is kind of always 225%, radius 1.1, lens blur and do not check the more precise button (dotn need ever to touch the more advanced tab by the way) you will get a perfect most of the time if not ALL the time a perfect result as a first sharpen.
    Then in need, what i suggest is another pass on part you want to enhence further more and use a mask to apply it...
    With respect to high pass that i use for 7 years on the 10 im doing this professionaly, is that Smart Sharpen is way more precise and more delicate vs high pass (as you can see on my example) High pass seem to keep the soft focus but enhance the edge making the file look more precise, but in fact its not. Have a look at the file i upload, a perfect example of what im talking about.
    The only time i still use the high pass is when after a interpolation i need to enhance the detail i have (smart sharpen give a too much natural look) or when i want to have a more crunchy look a la silver efex high structure ; )
     
  9. Patrick,
    It’s a vanilla v chocolate thing. The thing I don’t care for as much about Smart Sharpen is that it can still create halos, even if very slight ones. They’re just barely visible in your example above. Some people don’t mind them; some prefer them; some find them distracting. I find myself in all three camps, depending on the particular image.
    High pass sharpening is great for avoiding halos, as you’ll never get a halo when you use it. On the other hand, it has a tendency to step on fine details, as your example above demonstrates.
    That you prefer vanilla to chocolate and use it more often, and that I’m the opposite, is perfectly okay. Just because I eat more chocolate than vanilla doesn’t mean I don’t like vanilla — quite the contrary. And, often, the best desserts have both in suitable proportions, perhaps even with some other goodies mixed in as well.
    (For the record, I think your Smart Sharpen example above is spot-on and an excellent application of sharpening.)
    Cheers,
    b&
     
  10. I am also a Smart Sharpening fan in general, also using it about 80% of the time. Some people over-use the tool and the results are overly-obvious and un-natural. I generally sharpen to taste as a first step, then back off by about 60-70% (depending on the size of the file). I find that applying tonal adjustments, saturation, and other tweaks, can over-emphasize the apparent sharpness. After I'm good with the adjustments, and before converting the file from 16bit to 8bit at final size, I will give it a final sharpening.
     
  11. Hi Michael,
    i also reduce the first pass from 225% to around 150%, leaving me the extra 125% or so to be had later on with mask as the second pass of my sharpen method (it is better explain in my tutorial vs here ; )
    Since Smart Sharpen is the first thing i do when the file is open in Ps, i dont apply this filter over other effect so the risk of having a bad original is close to 0. I dont agree about SS creating halo IF use correctly, it is in fact what push me to use this filter vs high pass, where you can get easily halo if the effect is too strongly apply and depending of the choice of overlay / soft light you could also push the effect further in the halo creation IF not use correctly.
    I agree with you Ben that final result is a matter of taste, i personnaly like natural result so halo is not a option for me, in fact when i see some it ring a alarm in my head " my god, this guy applied a too strong sharpen to is file" kind of ; )
    thanks for the good word.
    In the end, nothing is beter than doing a test, so i invite the OP and any other user to use it as i describe and give me your oppinion after, with a image post that would be pretty cool; one with your method / one with my method like my example.
     
  12. Thanks Patrick for your comments, and Howard for pointing me at Patrick's tutorial. I remember downloading it some time ago, but could not find the PDF on my hard disk. Good to know that the More Accurate checkbox is inadvisable, because it is slower (one web comment says) because it is a two-stage process.
    In this 2006 thread, Robert Vine says "use 20% 1.0 on the highlight tab to remove haloing." Not sure whether he meant 20% Fade Amount or 20% Tonal Width. Probably the latter. With Radius=1, there is no difference than I can see in changing Fade Amount from default 50% to Robert's recommended 20%. It does make a slight difference with Radius=5, but nobody in their right mind would use such a large radius.
    My workflow does not allow for a Capture sharpening phase, because I start by Lanczos (or Bicubic) downsampling the out-of-camera JPEG. This greatly improves Bayer-pattern images,* so I suppose this is akin to capture sharpening. Images look dramatically better if I apply Smart Sharpen after downsampling rather than before. Using Patrick's settings (125), I will say that Photoshop Bicubic & Smart Sharpen is somewhat better than Irfanview Lanczos & Sharpen. Photoshop Bicubic Sharper is worse than Irfanview.
    USM threshold (~15) was extremely important for preventing further grain-up in blue skies, but instead of smoothing film grain, with digital cameras I often use the Gradient tool, or sometimes the sky actually appears as a color other than white (hurrah!) so threshold is not as important. Guidelines recommend against its use for film, but Smart Sharpen does seem to be an improvement for digital capture.
    * Foveon and Fuji-EXR images do not benefit as much from downsampling as Bayer-pattern images.
     
  13. Using a flash or lots of full spectrum light during capture is going to influence how far you go and the thickness of the halos when applying any sharpening to an image. I always get sharper images to start out with when using flash or any close to full spectrum lighting to the extent all is needed is a bit of USM and sometimes ACR's sharpening renders perfect sharpness without any halo's shooting Raw.


    Where halo's come about in my experience is when the image is too soft from lens diffraction or inaccurate focus and has a noticeable amount of low contrast to start out with. In those instances no sharpening tool will work satisfactorily.


    With no disrespect to Patrick I do think his continued use of the same professionally shot studio image to demonstrate how well HIS method works is not an accurate way to show how it will work on other images shot under different conditions outside a studio setting, use of a tripod and expensive light setup.


    Wish he'ld provide a demo of his technique applied to one of his own images he shot hand held outdoors like the rest of us photo shlubs have to deal with.
     
  14. Tim, I was just in the process of formulating a similar response to Patrick when your most recent post in this thread arrived in my inbox. I agree completely with you. When one is trying to make the best of a lower quality image, which is what most of us are forced to do, there is NO way anyone could get good results out of any single set of sharpening parameters or method. Based on Ben G's vanilla / chocolate analogy post urging flexibility, I suspect he probably also feels the same way.
    Patrick, your emphasis / insistence on SS with a radius of 1.1 px and method = lens blur strongly suggests to me, as it did to Tim, that you don't use the method and parameters you are suggesting on images in which (for whatever reason ... poor lens, camera movement, slight OOF, etc.) a point in the image gets mapped to anything other than a single, strongly peaked point-spread function.
    Attempts to deconvolve messy PSFs, especially when caused by blur at multiple length scales, clearly requires flexibility in one's approach, and likely involves multiple sharpening steps with different settings. I can guarantee you that providing users such flexibility is precisely the reason that Adobe provided sliders and not a single fixed setting of r=1.1, etc. in SS (and their other sharpening tools).
    Another example of why sharpening using a variety of methods is clearly necessary for most of photographers is the popularity of the recently introduced "clarity" slider. This is essentially a variant of classical sharpening using a radius in of several tens of pixels (on normal sized images). When Adobe introduced this tool, they didn't remove other sharpening tools such as USM (which is often used at much smaller radii). The reason is that the "clarity"slider helps with a low but broad PSF, while USM or SS at r~1 px helps with strongly peaked, symmetrical PSFs spread out over much smaller radii.
    It's obvious that SS works well on some images, eg, the studio shots that Tim suggested. It's equally obvious that it can't possibly be the be-all and end-all of sharpening methods for all images, which is, I think, the impression you tend to give.
    FYI, it is trivial to demonstrate the impossibility of universal applicability of any single sharpening method or set of parameters by using linear system theory and showing the uniqueness of the solution to most well-posed deconvolution problems.
    Tom M
     
  15. Tim, the continued use of the same image is because;
    1_it show the result perfectly.
    2_and i didtn think i need to do more images to illustrate something that work well
    But, you know me, ask and you shall receive ; ) You needed shot not in studio, not professionaly made, like the rest of us..where here they are LOL
    Heres a first example of a shot i have done during my vacation in Mexico with a Olympus E20 years ago.
    Im putting the original, the one with a highpass of 4 with softlight, and one with a smart sharpen of 250, 1.1, lens blur, box not checked. All the images i will post have the same exact setting of sharpen to show you that it could work. None have use any mask..i applied the effect all over the image, a bit strongly so you can see it.
    Feel free to download them, and put them together in the same PSD dociument so you can clearly see the original, the highpass and the smart sharpen result... i think it the least you could do to appreciate the effect ; )
    Of course, let me know what you think ; )
     
  16. MEXICO_real de catorce ORIGINAL
    00VNZA-205205584.jpg
     
  17. MEXICO_real de catorce HIGHPASS
    00VNZB-205205784.jpg
     
  18. MEXICO_real de catorce SMARTSHARPEN
    00VNZD-205205884.jpg
     
  19. TORONTO_from the CN tower ORIGINAL
    00VNZH-205205984.jpg
     
  20. TORONTO_from the CN tower HIGHPASS
    00VNZJ-205206084.jpg
     
  21. TORONTO_from the CN tower SMARTSHARPEN
    00VNZK-205206184.jpg
     
  22. PARIS_euro disney ORIGINAL
    00VNZO-205206284.jpg
     
  23. PARIS_euro disney HIGHPASS
    00VNZP-205206384.jpg
     
  24. PARIS_euro disney SMARTSHARPEN
    00VNZR-205206484.jpg
     
  25. PARIS_cimetiere pere lachaise ORIGINAL
    00VNZU-205207584.jpg
     
  26. PARIS_cimetiere pere lachaise HIGHPASS
    00VNZV-205207684.jpg
     
  27. PARIS_cimetiere pere lachaise SMARTSHARPEN
    00VNZY-205207884.jpg
     
  28. You should clearly see in all example that the Smart Sharpen result is by far more precise and delicate than is High Pass counter part...and far from the original... in a good way.
    let me know if there is something else you might want to see to get a better understanding of the SS plugin ; )
    ____
    This is what i would like to see from user who affirm that this or that is not as good as or better as something.. nothing is better to prove a point than visual example. SO thanks Tim for your demand, it push me further to demonstrate my point and i like that ; )
    It may or not suit you or your workflow, but you now know the method exist and give good result...For ME its a perfect setup to give my client maximum quality out of there raw images.
    ____
    On a side note, im in a process of doing my own personal web site with my own personal images.. so you would be able to see what i do during my trip, not just what i do for living... post a link when its time for those curious ; )
     
  29. Patrick, thank you for providing some examples. Of course, I couldn't resist playing with your original. I used my own mix of Focus Magic + multiple radii sharpening and obtained the attached image.
    Unfortunately, I had to begin processing your image with the version of the original already degraded by the photo.net upload process, but folks should still be able to easily see a clear improvement over SS. Pay particular attention to:
    (a) the reduced halos paralleling the closely spaced vertical lines in the stone column at the rear of the upper tomb;
    (b) the reduced halos around the squares that go around the top of the largest tomb;
    (c) the reduction in the bright halos on the leaves at the base of the red flowers closest to the camera; and,
    (d) the general improvement in texture, particularly, in the nearest two dark stone markers, the dirt, the flat stone surfaces, the trunk of the tree on the LH edge of the frame, and the leaves in the upper LH corner of the frame.
    I think that the improvement over a single step SS correction is obvious. OTOH, I will be the first to admit that my method involved about 4 steps (but all were easy and fast), whereas yours was a one-step operation, which is a big advantage to your method.
    Cheers and Happy New Year to you and yours !
    Tom M
    PS - I will likely be away from the computer all day and this evening, so my next response will likely be tomorrow.
    PPS - After I posted the image in this message, I now fully appreciate the importance of Patrick's suggestion to download all of the images into PS layers for comparison. It's quite difficult to see subtle changes like these by scrolling up and down the thread.
    00VNaz-205223584.jpg
     
  30. Tom, maybe is my limited english language but i didtn say or implied that 1 setting is good for all everytime, and i didtn say that you should applied this effect all over in one pass... have you take the time to read the free tutorial i post once in while? it is clearly explain that you should do a 3 step sharpening on all your images, with different setting and by applying mask at least in the second step.
    What i believe is that when you have found a amount that work well, and i discover than 225-250 was that number at least for my taste from a Canon G9 to a Phase One P45, and that if you use a radius of 1 or 1.1 you will not get halo and still keep the file as nautral as possible . Sure you can put 5 in the radius box and get a stronger effect a la dragan if this is you are after or if your file is really soft..but i like to start with good original not try to salvage all the time my poor shot.. and by learning how to properly shoot at first i achieve that most of the time.
    And whats the idea behind you dont use this or that on bad exposure file, out of focus one, or other problem file..why imply that because you are not a pro you have the *rigth* to shoot badly? i dont understand that point of view at all and ear it a lot from amateur... If you end up having bad exposed / out of focus / cheap lens or else, well just buy better equipment or learn to use what you have.. but please stop with the i shoot badly so your technique wont work..
    I didtn use USM for years because it simply sucks.. and to said what Bruce Fraser once said in is book " USM is like trying to cut a tree with a blunt tool, eventually you will get it down, but theres far more better tools to get the job done" or pretty close to this (dont have the book on sharpening with me).
    The best use of USM would be for me to make a local contrast enhancement, like what the simplier and better Clarity slider do in Lightroom..other than that i dont see the use of it when you are after quality in any kind of file, from a pro or not, well or not exposed files.
    As for the HighPass, it was THE tool choice by me to add sharpness to my images up to 2006, when i discover SS it change my life and the way i applied sharpening because as you can see and cant deny on the example above is that highpass seem to give a sharpness effect but in fact just emphazise edge, where the SS make all more precise for real.
    Why did Adobe let this filter there, or the brigthness / contrast for instance still active..its because of people who learn that way years ago and didtn like to learn new stuff.. they have created Layer with version 4, and still today a lot of people still work directly on there original.. they have create mask and still today people still dodge and burn directly on the original.. why? because many people are too lazy to learn new stuff or dont feel the need to learn this new stuff. with every version of Photoshop i get better and faster because they create better tool for us, that help me acquire better quality images in less time (stamp tool vs healing brush tool for example).
    I insist of using Smart Sharpen and i like to show the world that this filter work amazingly when you have some number to start to remove the burden of doing test all over the placer, i feel as strong with that filter as someone else would feel strong about PK Sharpener; for me they are doing the exact same thing when well execute, just depend of the knowledge you have and the money your are ready to spend to simplify your life in need.. i dont think that PK Sharpener will do more for me today vs the use of SS, and yes i use it and test it extensively before saying that in case someone ask.
    What you should retain from this post is simply that SS work very well when you have the correct setting, and that i dont feel anyone should or could be not impressed by it after seing the example i post (not to say that my image are the best a user have seen but the example i think are well done)..you have to agree that it seem to work well at least : )
    In the end, use whatever tool that make you happy and make you think its the best method; USM, HighPass, Smart Sharpen, PK Sharpener or else.. I feel comfortable using Smart Sharpen on 9 out of 10 of my or client images, and i still use HighPass for some for different reason or need.
    I dont say that if you dont use SS you dont know nothing about quality, but i could say that if you dont think after this post that its not a good alternative you dont know much about quality for sure.
    Tell you what, send me a shot that you will be proud to print (by proud i mean a basic file with quality, not out of foucs, not soft focus to the extreme..not a file to salvage, but a file you would like to show the world), and i will sharpen it for you for free.. that way you will see that SS work on anybodys image, not just mine. Dont do anyhting on it, if its a JPEG or a RAW i will do it if no sharpen have been applied first (thats is after all THE way of working wiht a file no?) I limited this offer to the first 3 email.. One of them being Bill the OP if he want ; )
    *hope this post doestn sound too arrogant.. im just passionate ; )
     
  31. *a quick edit to start, as i just post this and see Tom answer. As i said, i apply the effect in one pass witch i never do and a bit stronger so you can see the effect on a unedit image..no contrast or curve have been applied to any of them that could have also make a more interesting visual of course ; ) , but i agree with him that his version look more natural with a bit of manual work (and the filter seem to have applied a kind of contrast alos, witch is good) I will get the demo myself and start experiment with it.
     
  32. Tom,
    That's interesting. I'll have to check out Focus Magic.
    Smart Sharpen seems to be particularly brutal on skies, especially when applied repeatedly. Also, I find it incompatible with Lightroom's sharpening tools, e.g. when I edit a Lightroom image as a Smart Object in PS and then switch back to LR when I'm done. Now as a rule I never mix PS an LR sharpening, which seems odd since they both come from Adobe.
    Sharpening is such a broad and mysterious topic. Someone could probably write a Masters thesis on sharpening and not cover all of the options.
     
  33. Focus Magic is totally amazing for reducing shake-induced lens blur, or bad IS behavior as sometimes occurs. My friend Ted Marcus says it's good for sharpening also, but I have never had time to investigate. However in Tom's Pere LaChaise FM sharpening, I see strong black halos at 200% zoom. Tom sharpened a lot more than Patrick did in his SS example.

    HighPass seems quite poor as a technique. In all of Patrick's samples, except maybe EuroDisney, I think the original looks more natural on my LCD monitor. It seems to me that HighPass reduces the 3D illusion.

    I like to look at 200% zoom to judge image quality. Luminous-Landscape is a site that pushed high pass at one time, and it appears to me that most of the images posted there are mildly to extremely oversharpened when viewed at 200%.

    This is an interesting discussion, but the bottom line is that I don't have time for fiddling around with every image. I want something that works fairly well all the time, and SmartSharpen seems to be it. Thanks again Patrick.
     
  34. Patrick, those are excellent examples of good sharpening using SS. On my display they're a bit on the crispy side. Don't know if those are 100% crops. Just FYI I never used High Pass sharpening. USM has always worked for me in the past but I have to admit viewing on a pixel level SS has a lot of control but I still get halos from it but not thick ones like USM. I can't show any examples right now because my main computer went down about a week ago and I'm stuck with PS 7 on a 2000 Pismo PowerBook in OS 9. 2003 version of Mozilla is the only browser that works and I have to turn off JavaScript to be able type a response which accounts for the lack of line breaks so I'm keeping it short. Happy New Year everybody. Waiting for my power supply to arrive by FedEx so I can get my 2004 iMac working.
     
  35. It sounds like a lot of expert opinions above.
    I happen to really like smart sharpen as follows:
    Use lense blur and NOT more accurate mode for first pass. Play with settings. The larger the original file and more real info. the more the sharpening and the larger the pixel. Usually a bit less than 1 pixel and and near 100 as to amount.
    Sometimes I like to do a second pass at half or more the pixel size and half the amount as the previous pass using more accurate mode.
     
  36. "It seems to me that HighPass reduces the 3D illusion." I agree, particularly in the eurodisney image.
     
  37. It's not like chocolate and vanilla. It's like blaming the stove because the food keeps burning. While it could be the stove, we have to assume that if other people can get the stove to work it's most likely the stove operator's error. I rarely find tutorials that match my taste. They are for initial learning. Practice and study may still be required to figure how the technique or tools work best for me.
     
  38. some good tutorial are out there, and some are really not well done or show old technique, or worse show a good technique but wrongly applied..
    lynda.com is a good place to learn, photo.net is also a good place when people provide visual example of what they say, but like Matt say, blaming a technique because it doestn work for you but strangely work for others is too simple. While practice and study is good, you always need a start up manual or some sort of direction to be on the right track fast, and this is where tutorial are good.
    As for the thesis on sharpening that Dan ask, it exist and its call > Real World Image Sharpening with Adobe Photoshop, Camera Raw, and Lightroom, Second Edition
     
  39. I'm pretty new to digital processing and know that I'm deficient in experience and judgment when it comes to sharpening (along with a lot of other things). So here's what I've been doing lately as a learning technique: Put two copies of the image side by side in Photoshop. Run the applicable PK Sharpener routine on one. Then use Smart Sharpen on the other to try to tweak a little more sharpness out of it without significantly changing the overall look compared to the PK sharpened version.
     
  40. If sharpening full rez images later to be downsized for the web to like say 700 pixels wide, create a New Window set at 25% zoom while you sharpen the original at 100% view giving you live feedback as it will closely appear downsized for the web. This helps avoid the crispy look after downsizing. LCD's have gotten really sharp lately and combined with the way an OS draws pixels in web browsers (which isn't exactly the same as Photoshop BTW) can deliver unexpected results.
     
  41. As I discovered while researching another thread (ImageMagick sharpening, not of interest to Photoshoppers), JPEG parameters make a lot more difference than sharpening techniques for thumbnail-size images.
     
  42. One quick trick for reducing the amount of halos when sharpening in photoshop is to switch to LAB mode and use any of the sharpening tools on the Lightness channel. I find many thrid party sharpening tools better than those within photoshop.
     
  43. no need to go into LAB for that (or to use LAB at all for 99% of anything you do in Photoshop) just use the Luminosity blending mode..same effect, quicker and simple.
     
  44. Agree with Patrick with that LAB go round. The only thing it will do to halos is to prevent color fringing along edges, but there is a more powerful weapon for that using ACR's sharpening combined with Chromatic Abberation panel setting the drop down menu selection to All Edges (requiring a saturation boost afterward) or Highlight Edges with hit or miss on some edges. Viewing at 200% or above to where you see how the pixel spread looks along edges is the only way to see how powerful this combination really is. You're looking for clean stairstep edges and you can get this very quickly in ACR or LR. Not sure how it looks on jpegs but Raw really shines using these two tools.
     
  45. Hi Patrick - I'm sorry that I couldn't get back to this thread before now, but the holidays were very busy and I've been in and out of town.


    [PL]: "...Tom, maybe is my limited english language but i didtn say or implied that 1 setting is good for all everytime, ..."

    Patrick, in your post of Wed, Dec 30, 2009 at 12:49 PM, you state:
    [PL]: "...i think that when you have find the correct settign in
    Smart Sharpen, witch is kind of always 225%, radius 1.1, lens blur and do not
    check the more precise button (dotn need ever to touch the more advanced tab by
    the way) you will get a perfect most of the time if not ALL the time a perfect
    result as a first sharpen...."


    I think that most people would categorize the above statement as an extremely strong emphasis / insistence on using only Smart Sharpen, and using it only with a narrow range of radii (ie, around 1 pixel, not the tens or hundreds of pixel radii typically used in large-R USM). This is what I claimed in my earlier post in this thread.

    Going back over quite a few of your other postings on sharpening and re-reading your tutorial, I could not find one single instance where you recommended any other sharpening tool over Smart Sharpen (aka, SS) for any situation, and I certainly have not seen you even discuss SS's range of applicability in a logical and dispassionate manner. Repeatedly recommending SS to the exclusion of all other algorithms is limiting, illogical and impossibly optimistic. It's a "one-size-fits-all" approach. Every algorithm has its domain of optimal applicability, has its own unique set of adverse side effects, etc. and SS is no exception to this. While Smart Sharpen is certainly a reasonable choice for images of a certain type (ie, fairly high quality; a tight, isotropic point spread function, etc.), a truly knowledgable and professional user makes all processing decisions fully cognizant of the other methods available and the trade-offs involved in each.

    -----------


    [PL]:"...and i didtn say that you should applied this
    effect all over in one pass... have you take the time to read the free tutorial
    i post once in while? it is clearly explain that you should do a 3 step
    sharpening on all your images, with different setting and by applying mask at
    least in the second step...."


    I did read your tutorial, and I did see your statements on this. I have absolutely no disagreement with you on spreading the sharpening out over more than one step (as per Bruce Fraser's recommendations). Rather, my concern is about an undue emphasis on using only one algorithm (and to a lesser extent, a small range of parameters) for all photographic sharpening tasks, not on dividing the sharpening over several steps.

    ------------


    [PL]: " ... What i believe is that when you have found a amount that work well, and i discover than 225-250 was that number at least for my taste from a Canon G9 to a Phase One P45, and that if you use a radius of 1 or 1.1 you will not get halo and still keep the file as nautral as possible ..."

    You are absolutely correct that when some particular technique works for a person, as long as the nature of the work doesn't change, that person should stick with the tried-and-true method. My disagreement is that you seem to completely dismiss the legitimacy of people who, for some reason, are forced to work with lower quality images, and hence Smart Sharpen at a radius of around 1 pixel is very likely inappropriate or certainly inadequate. Examples of your statements in this direction include:


    [PL]:"..but i like to start with good original not try to
    salvage all the time my poor shot.. and by learning how to properly shoot at
    first i achieve that most of the time. ..."


    [PL]: "...And whats the idea behind you dont use this or that on bad exposure file, out of focus one, or other problem file..why imply that because you are not a pro you have the *rigth* to shoot badly? i dont understand that point of view at all and ear it a lot from amateur... If you end up having bad exposed / out of focus / cheap lens or else, well just buy better equipment or learn to use what you have.. but please stop with the i shoot badly so your technique wont work.."

    [PL]: "...You cant be disapoint with something you dont fully understand, can you? ; ) ..."



    Although I'm sure it was entirely unintentional, can you see how such statments might easily be taken as serious insults by the large numbers of professional photographers who, on a daily basis have to prepare important images for publication, but these images are far from technically perfect. Would you tell such pros to learn their craft and buy better equipment when even the best of them can't guarantee pin-sharp images? You only mention amateurs in this regard, but may I remind you that photojournalists, shooters who specialize in sports or street photography, microscopists, photo-restoration experts, underwater photographers, and many other photographic specialties often can not produce in-camera pin sharp pictures because of because of light scatter, abberations in their optical system that studio photographers would consider bizarre, abberations because the subject is near the edge of the frame, specimen vibration or motion (in microscopy), slight focus error in following the quarterback, etc.

    As a joke, I might ask that if SS is good for any problem, why did the Hubble Space Telescope team once spend months perfecting a sharpening algorithm and not just use SS? Oh, sorry, at that time, I guess the Hubble optical system wasn't at a pro-level sufficient for SS to be applicable. Sorry, couldn't resist the joke. ;-)

    Rather, one should teach specialists in these areas about tools which will help them overcome the problems they face in their individual specialties. Appropriate tools for such problems include large R USM, anisotropic blur reduction (e.g., focus magic), local contrast enhancement based on different length scales (ie, Topaz Details), luminosity dependent local contrast enhancement (ie, NIK Tonal Contrast), etc.

    Hopefully, I am wrong and you have given appropriate weight to these relatively new tools in some other forum, but because I have not seen this, I feel compelled to point out what seems to be a glaring omission from your many posts on sharpening.




    [PL]: "...I dont say that if you dont use SS you dont know nothing about quality, but i could say that if you dont think after this post that its not a good alternative you dont know much about quality for sure. ..."

    Of course Smart Sharpen is a very reasonable method for many images, BUT, it should be only one of many tools in the toolbox of a true, broadly educated and broadly experienced pro.

    [PL]: " ...hope this post doestn sound too arrogant.. im just passionate ; ) ..."

    Me, too!.!.!.


    Best regards,


    Tom M


    PS - I just noticed your message in which you said that you liked the way I sharpened your image using Focus Magic, and said that you would look into it as another tool to have in your toolbox for sharpening. Give it a try, I think you'll find it useful for some applications. For example, I find it to be particularly good for single step sharpening after bi-cubic downsizing a full rez dSLR image to a 700px wide image for photo.net. A setting of r=1 generally works quite well this task, but for sharpening of full rez images larger values of R are often useful. In addition, I'll often split the right (ie, bright) "blend if" slider to fade out the focusing effect on the brightest tones in the image.
     
  46. i agree with you, having differents tool for different need are the best avenue. I had use many of them, and for my workflow i end up using most of the time SS as i find it work well, or the sharpen function in Ligthroom when for my personal image i dont need any photoshop other than darkroom.
    I still think that SS give a more natural feeling vs HighPass or USM, and im agree that different setting can be use depending of the problems. Since i cant create a tutorial with all the different problem shot in the world i go with one good shot to show how 2 technique compare between them..what you have seen in my different post.
    Some find USM acceptable, i dont. Some find PK Sharpener the best plugin, i agree with them that its one of the simpliest one, and that it give good result, but nothing i cant do with a free included filter in Photoshop..And some use HighPass, i have use it and find many flaws, so again i decide to stick with one good alternative. Im not close to other filter (if they work on a mac) if they can give me a better result in less time or same time, thats why i will when i ca have a look at your plugin ; )
    I have the chance to work with pro and you are right by saying that even them dont give me super sharp pin point focus all the time, but they and i dont always use this as a excuse... on those problem case i use sometime high pass to emphasize some details more agresively or a stronger setting in SS... but this is on problem shot only. Also, i think that the digital have bring the pixel piping to a new level, because most of the time, even a not so sharp image look great at a regular size like 8x10, in a magazine for example.. if you remember it, not so long ago whe where use to see smoothness on film shot in the magazine, now everything need to be more than sharp..unaturally sharp.. im sure they will create a plugin soon that gave the smooth film look to digital file .. wait is here, its call Alien Skin Exposure ; )
     
  47. Below is an example of the type of stairstep pixel edge sharpness I look for when sharpening without kicking up halo's. The samples are from screenshots viewed at 400% in ACR and CS3.
    I've included the sharpening settings for ACR vs Smart Sharpening's setting applied to the ACR default sharpness version. As you can see I arrived at identical results using both methods. This is as sharp as I could get it without amplifying noise and halo's. ACR Luminance Smoothing was set at 20 and
    Color Smoothing set at 2.
    00VQDG-206831584.jpg
     
  48. Some find PK Sharpener the best plugin, i agree with them that its one of the simpliest one, and that it give good result, but nothing i cant do with a free included filter in Photoshop.​
    I like SS better than PK Sharpener for web sharpening. I think the SS sharpened web size jpegs look better. I think though that PK Sharpener is a great tool for sharpening for different types of print output.
     
  49. Smart Sharpen is an improvement of Adobe USM.
    But, a good USM can produce results that are similar to SS.

    In any case with a good USM or SS you get halos or too shining elements producing a "digital sharpness look".
    To have a "natural sharpness look", you need a sharpen tool that can adapt to the edge strength, preventing halos or too shining elements.
    00VUMF-209435584.jpg
     
  50. That's pretty darn good, jacopo. I compared it to Patrick's original version above and yours is noticeably sharper but without the halo's.
    What was your workflow and tool settings?
     
  51. Nothing special, I used PhotoResampling ( http://www.photoresampling.com/index_eng.php) Image->filters->Sharpen->GBSHN.
    Then you can set parameters to your taste.
     
  52. Boy, I wish I could go back and change the subject of this to "Very impressed with Smart Sharpening." I am investigating GIMP wavelet sharpening, and have new-found respect for Patrick's suggestions.

    BTW, comparing Jacopo's image just above, Patrick's Smart Sharpen is much sharper, but that could be just because the JPEG is Q94 1x1 instead of Q90 2x2. I think Tim had the images mixed up when he compared; the names are similar.
     
  53. I've read all of this as well. It's not all that explained up above.

    For example -- is Patrick recommending Smart Sharpen at 225% and 1.1 at full resolution (say 18MP from my 7D) before downsizing for Web use? (say reduced 700 to 1200 px wide). And when downsizing use Bicubic (not "Sharper") then after the SS was applied at full res.?
    Also, recently (Dec. 2010?), Patrick is using USM again -- he wrote he used "15-15-0" on nearly all his images.
    -Ken
     
  54. the usm is more for a local contrast enhencement a la clarity slider kind of. i use it only, and only for that purpose : )
     
  55. Ken Papai wrote:
    Also, recently (Dec. 2010?), Patrick is using USM again -- he wrote he used "15-15-0" on nearly all his images.
    Patrick Lavoie wrote:
    the usm is more for a local contrast enhencement a la clarity slider kind of. i use it only, and only for that purpose : )
    Hi Ken,
    Can't answer your other questions, but if you'd like a more elaborate explanation of Partick's use of USM, refer to the Mac Holbert's Mid-tone Contrast Enhancement action thread and check out Ron Bigelow's Localized Contrast in Photoshop tutorial where he painstakingly explains every nuance of this trick.
     
  56. wow.. i read this article quickly.. i like thing simply explain, less technical... for user that want a explanation to the core of the trick.. he is good. But a simple 15-15 or 20-20 in the amount and radius box do a good result.. without too much unneeded information for the rest of us ; )
    but thanks for the link, im sure many will appreciate.
     

Share This Page