p.1 #1 · Sharpening for web resize: multi-step script vs LR plugins
While browsing through some of the photos in the Zeiss Photos thread and others, I started to research how some of my favorites were processed. I found that some guys are using a multi-step resize / sharpening workflow using Photoshop and scripts, for example:
There were also some discussions in FM about this type of script.
I am convinced now its the way to go, but want to explore some possible alternatives.
Is this technique still considered "the best" for resizing for web output?
Which version of the script is best?
More interesting to me, are there Lightroom plugins (or standalone applications) that will do as good or even better?
Keep in mind, I am only talking about resize / sharpening for web output, not the initial stage or selective sharpening.
In my opinion, this is a topic that warrants deeper exploration, especially for those of us who will have the web as their main target for output.
People discuss and argue endlessly about the sharpest lenses, the best handheld technique, image stabilization, etc.... when often one of the biggest impacts (assuming good technique) will be in PP / resizing / sharpening.
p.1 #2 · Sharpening for web resize: multi-step script vs LR plugins
What your writing about is an action not a script. In his link the author stated that he hope that you would be able to see the difference between the image sharpened where he used a particular USM setting compared to the results his sharpening action produced. You may or may not be able to. For the resolution of computer displays vary quite a bit. Most displays these days are LCD displays which look best when run at there native resolution. I'm not referring to the number of Pixels here rather their pixel sizes their density. Pixels Per Inch PPI or DPI. LCD I have used range in resolution from 84DPI to 326DPI. Small devices like the ipod and iPhone have LCD with a 326DPI resolution. Displays do not have the same sharpness and images display on a low dpi resolution will be larger in physical size then the same image displayed on a high dpi resolution display. Web images do not look the same on all displays my laptop LCD has a 133DPI resolution and my desktop displays have 84DPI and 100DPI resolutions my iPod 326DPI.
His image is only 705x470 pixels will fit on your mobile device can you see a difference?
705/3=205 each image slice width is 235 Pixels wide and the height less label ares 426 Pixels.
On an 84DPI display an image slice width is 2.79" x 5.04"
On an 100DPI display an image slice width is 2.35" x 4.26"
On an 133DPI display an image slice width is 1.76" x 3.20"
On an 326DPI display an image slice width is 0.72" x 1.30" http://www.alexnail.com/blog/wp-content/uploads/2011/10/sharpen-comparison.jpg
p.1 #4 · Sharpening for web resize: multi-step script vs LR plugins
What's that, Mr Mouse, no 72dpi monitor ?!
If images are being sharpened correctly for print then they'll look crappy on most screens (perhaps not the 326 dpi, but they probably won't fit on it anyway). If they are being sharpened for web display then the simple reality is that most browsers will stuff up the colours and make things look crappy, and most screens are not calibrated or profiled and that too will make it look crappy - so why bother with such an intricate sharpening process ? Even if they could see the difference between sharpening techniques they probably won't be seeing the overall image the way you intended.
p.1 #5 · Sharpening for web resize: multi-step script vs LR plugins
Thanks all for the comments.
Mr. Mouse,
Thanks for clarifying that its an action rather than script.
Also, I am mostly interested in sharpening for 24" to 27" displays with HD formats (1920x1080), using browsers like FireFox and Safari that support builtin profiles.
So DPI is around 100.
Newhaven,
I am interested in how to sharpen images for resizing specifically.
I am not sure the problem of resizing / sharpening is mitigated by the new Bicubic routines of photoshop.
Alen321,
Firefox and Safari both handle colors and profiles well, and I myself and many other photographers do have calibrated monitors. Since viewing on web browsers is the main way of sharing and critiquing / learning about photography, for my own personal interests its very important to handle resizing / preparation for web well.
Here is a key quote from that discussion: I personally like to down sample in two steps to trigger one of Photoshop's 'undocumented' features. First, I determine my final output size, say 900px on the long edge. I then resize the original image to 1.66666666 x the desired output size (for this example 900px * 1.666666 = 1500px) using the Bicubic normal method. Next, I run the Filter->Sharpen->Sharpen command. Depending on the image, I run the sharpen filter once or twice. Finally, I resize the image to the final output size using Bicubic Sharper method. Can't remember where, but I read about this two step method about two years ago and the original author gave the rationale that this series of steps has the same effect as applying an ultra-thin USM to the low-res image. Since the results are so good (most images in web-size come out razor sharp) I recorded it as an action and use it as my default 'export to web' output action....Show more →
p.1 #7 · Sharpening for web resize: multi-step script vs LR plugins
And there is another thread of discussion with several people mentioning downsizing to web output in 2 steps, again with the number 1.6666 being mentioned:
Here is a script mentioned by Denoir, with a direct comparison of bicubic sharpening to his own script results. To my eyes, the script does a vastly superior job:
p.1 #8 · Sharpening for web resize: multi-step script vs LR plugins
You seem to be sharpening for yourself and intend to use color profile other the sRGB. Sharpen your images to the way they look best to you. Your Your own judge sharpening is very personal. So please yourself sharpen your images to your personal taste for your displays. Perhaps too much has been written about sharpening when to do it different methods is it worth the time and trouble to sharpen to the point you need to show people how well you did sharpen that you need to carry around your display and a magnifying glass or lope,
p.1 #9 · Sharpening for web resize: multi-step script vs LR plugins
I've started to run some comparisons. So far, Image Size... with Bicubic (best for smooth gradients), using multiple 10% reductions produces much better results than a single step.
I'm saving the result as a tiff file and sharpening in camera raw to take advantage of the clarity slider and the detail and masking features in the sharpening tab. Using camera raw for sharpening also eliminates the need for luminosity blending or going in to LAB.
p.1 #10 · Sharpening for web resize: multi-step script vs LR plugins
Mr. Mouse,
We spend vast amounts of money trying to get sharp images by analyzing and purchasing the best lenses, using tripods, etc... Yet, I don't think anyone has yet answered the original questions about getting the best sharpening for web output. Since it is indeed an issue of personal preference, and often that is what is mostly interesting to discuss here, I have refined my questions a little.
newhaven,
Sounds like an interesting test. Looking forward to hear your results.
Also, I would like to do a little reset and focus the discussion. I have confirmed that some of the posters on FM whose work I most admire, are indeed still using the "Multi-step resizing/sharpening with 1.666 factor". That is the "look" that I like, so now the question is how to best achieve it?
Does anyone else use the multi-step resize approach?
Any idea which approach is best (I linked to one web site and 2 different FM discussions about this topic, with actions generously donated by denoir and others)?
Has anyone discovered a way to achieve the same or better result with a Lightroom plugin?
p.1 #13 · Sharpening for web resize: multi-step script vs LR plugins
There a lot of different variables that will affect the amount of sharpening an image needs: detail and contrast in content, sharpness at capture, method of output (screen image vs print), size of output, and anticipated viewing distance.
I still use a method learned from a Dan Margulis magazine article back about 10 years ago for my small screen images:
1) Apply USM 500, .2, 0
then
2) Edit > Fade USM (luminosity)
Step one applies very aggressive but narrowly applied USM overall.
Step two allows me to use the opacity slider in the Fade step to modulate how much of it I apply from 0% to 100% by eye as I see the full size image in the edit window change. I slide from 0% (none) to 100% (too much) and find the happy place in between, usually in the 50-70% range. Again that depends on the content and other factors.
Changing the mode of the fade step from normal to luminosity has the same net effect switching the file to Lab mode and sharpening the L channel. Back in the days of shooting in JPG on 3 MP sensors changing to luminosity would help prevent funky sharpening artifact along contrasting color boundaries like yellow/blue. But with today's resolutions and 14-bit capture / 16-bit editing the difference it makes, while still there, is more difficult to see.
In most files I edit I don't apply sharpening globally. I will make a dupe layer, sharpen it, then blend it back into the softer original with a mask creating as sharpness gradient in the photo that subliminally tells the brain of the viewer to go to the sharper content at the focal point(s).
For portraits I have an action that creates duplicate surface blur, high pass filter, and USM 500,.2,0 layers with masks over the original. I will open the masks selectively to selectively blur the flat skin areas to hide blemishes to the point of Barbie Doll plasticity with the surface blur then add the skin texture back in with the high pass layer. By manipulating the masks and opacity of the layers I can subtract 20 years then add about 10 back to come up with a more flattering yet still naturally plausible appearance similar to a well applied makeup. I use the USM layer selectively on the eyes and mouth to exaggerate their sharpness a bit and make the contrast more from the rest of the face and clothing I've "knocked the sharp edges off" with the blur/high-pass.
If you look for a simple one-size-fits-all automatic solution you'll find that works about as well for photos as it does for pants.
p.1 #14 · Sharpening for web resize: multi-step script vs LR plugins
Thanks everyone for the feedback.
There are some very interesting ideas here, I will explore them.
In the mean time, I have decided to go with my own Photoshop action, which I will base on the advice of Denoir, CarstenW, Makten and others. They are all very similar, using multiple steps in the resize process.
It will take some time to set this up, but once I do I hope to post my experiences.
If anyone knows of a commercial plugin to LR that can do the same as what is achieved with the multi-step resizing, I would be very interested to learn about it.
Or, or anyone has any advice about the multi-step action, that would also be immensely helpful.
I guess the good part of this, is that it will force me to learn Photoshop!
p.1 #15 · Sharpening for web resize: multi-step script vs LR plugins
Perhaps I'm just not to that level, but personally I find LR's built in export sharpening tool to be more than sufficient for my web upload needs. I just use "sharpen for screen" as the method and "standard" as the strength.
Keep in mind, however, I usually only upload at 800px longest dimension and use variable jpeg compression for a final filesize of <250kb. If you upload to different dimensions/filesizes, different settings may be more optimal.
If you are looking for a "clean" lightroom plugin and have photoshop, don't forget that you can define a "droplet" or "action" to be applied to each image you export in Lightroom. I used to do this, using a custom action I made that would add a white border with a faint drop shadow, and "save for web" at 250kb or less. I would export my pics in LR as TIFF and the post-export action would automatically add the border, shadow, and save at a small filesize.
However, I have since gone to using LR2/Mogrify which can do essentially the same thing as my photoshop action (only no drop shadow, but I'm okay with that) and no need to load photoshop or have TIFF intermediaries in the process. I then use the FTP upload plugin to load them directly to some web storage I have for hosting. The whole thing saved as an export preset, it's super easy to upload pics for web viewing. Just use the export preset, type in my FTP password, and off it goes! Here are some quick examples:
p.1 #17 · Sharpening for web resize: multi-step script vs LR plugins
Multi-step resizing is not really necessary these days since Photoshop improved their resizing algorithms; certainly not for downsizing to web size, and not for ordinary resizing.
I resized an image from about 3800 pixels on the longest side to 6840 px. That's an 80 % increase in size, done in one step. It was in a juried show and the gallery owner was pleased to see it.
