Posted by Steve Sieren on August 16, 2009 at 4:16 PM
Hello LFer's after seeing Preston's Lodge Pole Pine detail shot I became curious to how you guys sharpen your images for the web. Do any of you all care to share your technique on that or at least just the basics of it?
When I first saw the title of your post, I thought, perhaps, that you were going to come over to the dark slide.
Onward...
I use two techniques in CS2, each about 50% of the time.
The first is quick and easy:
Make sure image is the size you want
1. Convert to sRGB.
2. Layer>Layer via Copy...(of the background layer
3. Zoom the image so that the ruler is accurate.
4. Filter>Smart Sharpen or USM; Radius 0.3, Threshold 4
5. Apply to taste.
6. Save
Another method:
sharpening technique...
1. Open your sized RGB image.
2. Choose Layer>Flatten
3. Choose Image>Mode>Lab Color
4. In the Channels Pallet, choose the "Lightness" channel.
When you choose the 'Lightness' channel your image will revert to a very light B&W image. Don't worry.
5. Zomm so that ruler is accurate.
6. Choose Filter>Unsharp Mask
Settings for USM: Amount=85%, Radius=1, Threshold=4. Preview--Checked
If you need to apply more USM, press Ctrl-F to apply USM again. Repeat this step as necessary. You can also expermint with the radius setiing if 1px results in jaggies.
7. Choose Image>Mode>RGB to switch back to RGB color.
8. Convert to sRGB and save.
Steve, I use a mixture of Preston's processes along with Marc's technique. Some of it gets into having to watch as you go along as I'm sure your aware of already.
1) Image is scanned at auto sharp feature in Silverfast at 4 scans per copy.
2) Worked and processed in PS as Adobe RGB.
3) Choose Image>Mode>Lab Color
4) In the Channels Pallet, choose the "Lightness" channel.
5) In B&W mode now, magnify to Actual Pixels.
6) Filter>Unsharp> Amount 400% - Radius .4 - Threshold 4
7) Mode>RGB
8) Mode>16bit to 8bit
9) Save as jpeg
10) Size jpeg from exisiting pixel size to 1600 long side for 1200 finished jpeg. 1200 for finished 720 jpeg.
11) Filter>Sharpen (once or twice according Marc's approach) For me once is enough.
12) Size jpeg from 1600 to 1200 for large mounted jpeg or 1200 to 720 for small jpeg.
13) Edit>Convert to Profile sRGB
14) Save for web as Optimized.
Note: you can achieve the same effect of sharpening in lab, without converting to and from lab, by doing your USM sharpening on a separate layer and setting the blend mode of that layer to luminosity. Or, after applying USM, go to edit>fade, then choose luminosity.
Personally, for printing I've had great results with my recently discovered high contrast high pass filter - after uprezzing to say 20x30in at 360 ppi from a 5D2 file I'll duplicate the flattened image, run high pass at 0.6 or so pixels, remove all color, and then apply a very steep curves adjustment to that layer only. Then set to overlay mode. It's astounding what little details it picks out! And no edge halos at all!
I'm sure methods are different for film vs digital; have you secretly gotten into some LF?
Posted by Lon Overacker on August 16, 2009 at 8:09 PM
Yeah Steve, making plans on shooting some film??
Up until recently, I had been using Marc's technique posted on the PNW regional forum. I'm sure you're familiar with that so I won't repeat all the steps. But I pretty much had followed what Paul was describing except my size settings are a bit different; 1200>600 because I also add the frame that takes it back up. But actually only to 684 because that's the size of images on my website. For the larger version it's usually 2000>1000 or 1050. I've yet to have an image that I could save at 1200 and keep under 500k.
Also, keep in mind that most of my scanned images may be in the 8000-10000px size after a scan. It's amazing that we/I can retain detail after several resamplings.
Currently, I'm using TK's web sharpening actions and I love them. He has two sharpening actions called "less Halos" and "more contrast." According to Tony, the actions perform very much the same way as Marc's Technique and also one that Floris had written about (forgot where.) The things that I really like about the actions are: 1) it's an action and makes the workflow more efficient, 2) creates adjustment layers so you can mask out any sharpening (like a blank sky,) 3) adj layer also allows control with opacity to reduce or increase the effect and 4) he throws in a curves layer in case you want to add a tad bit of contrast at the end. Oh, and 5) the actions pause with the image-size dialogue box to let you set your own sizes. Both of these actions are pretty aggressive, but as I mentioned, the adjustment layer allows you to mask to taste.
It's very interesting how many different approaches there are for sharpening, preparing for web, and printing.
But to answer the question, my methods are not different for digital captures or film scans. Just the strength of the application of them may vary.
Posted by Steve Sieren on August 18, 2009 at 1:11 AM
Sorry guys no medim format yet, just curious since I always hear that there are different methods for the scanning. Seems like the sharpening must be more difficult with the large files. Thanks for the info hope you don't mind if I play around with the techniques if I get a chance.
Actually, Steve, when working on the larger format scans, the pixel density is higher, thus resulting in sharper originals, especially after reduction to print or web sizes. It is not more difficult at all. I've had numerous 4x5 scans that did not require any sharpening.
Discussion
8 repliesWhen I first saw the title of your post, I thought, perhaps, that you were going to come over to the dark slide.
Onward...
I use two techniques in CS2, each about 50% of the time.
The first is quick and easy:
Make sure image is the size you want
1. Convert to sRGB.
2. Layer>Layer via Copy...(of the background layer
3. Zoom the image so that the ruler is accurate.
4. Filter>Smart Sharpen or USM; Radius 0.3, Threshold 4
5. Apply to taste.
6. Save
Another method:
sharpening technique...
1. Open your sized RGB image.
2. Choose Layer>Flatten
3. Choose Image>Mode>Lab Color
4. In the Channels Pallet, choose the "Lightness" channel.
When you choose the 'Lightness' channel your image will revert to a very light B&W image. Don't worry.
5. Zomm so that ruler is accurate.
6. Choose Filter>Unsharp Mask
Settings for USM: Amount=85%, Radius=1, Threshold=4. Preview--Checked
If you need to apply more USM, press Ctrl-F to apply USM again. Repeat this step as necessary. You can also expermint with the radius setiing if 1px results in jaggies.
7. Choose Image>Mode>RGB to switch back to RGB color.
8. Convert to sRGB and save.
Note that I use the second method for printing.
-P
.
1) Image is scanned at auto sharp feature in Silverfast at 4 scans per copy.
2) Worked and processed in PS as Adobe RGB.
3) Choose Image>Mode>Lab Color
4) In the Channels Pallet, choose the "Lightness" channel.
5) In B&W mode now, magnify to Actual Pixels.
6) Filter>Unsharp> Amount 400% - Radius .4 - Threshold 4
7) Mode>RGB
8) Mode>16bit to 8bit
9) Save as jpeg
10) Size jpeg from exisiting pixel size to 1600 long side for 1200 finished jpeg. 1200 for finished 720 jpeg.
11) Filter>Sharpen (once or twice according Marc's approach) For me once is enough.
12) Size jpeg from 1600 to 1200 for large mounted jpeg or 1200 to 720 for small jpeg.
13) Edit>Convert to Profile sRGB
14) Save for web as Optimized.
For printing, there's a number of good techniques that were discussed here:
http://www.naturephotographers.net/imagecritique/bbs.cgi?a=vm&mr=44348&CGISESSID=b63d39eb1651a044ff8d15dfcb634a01&u=23831
Personally, for printing I've had great results with my recently discovered high contrast high pass filter - after uprezzing to say 20x30in at 360 ppi from a 5D2 file I'll duplicate the flattened image, run high pass at 0.6 or so pixels, remove all color, and then apply a very steep curves adjustment to that layer only. Then set to overlay mode. It's astounding what little details it picks out! And no edge halos at all!
I'm sure methods are different for film vs digital; have you secretly gotten into some LF?
Up until recently, I had been using Marc's technique posted on the PNW regional forum. I'm sure you're familiar with that so I won't repeat all the steps. But I pretty much had followed what Paul was describing except my size settings are a bit different; 1200>600 because I also add the frame that takes it back up. But actually only to 684 because that's the size of images on my website. For the larger version it's usually 2000>1000 or 1050. I've yet to have an image that I could save at 1200 and keep under 500k.
Also, keep in mind that most of my scanned images may be in the 8000-10000px size after a scan. It's amazing that we/I can retain detail after several resamplings.
Currently, I'm using TK's web sharpening actions and I love them. He has two sharpening actions called "less Halos" and "more contrast." According to Tony, the actions perform very much the same way as Marc's Technique and also one that Floris had written about (forgot where.) The things that I really like about the actions are: 1) it's an action and makes the workflow more efficient, 2) creates adjustment layers so you can mask out any sharpening (like a blank sky,) 3) adj layer also allows control with opacity to reduce or increase the effect and 4) he throws in a curves layer in case you want to add a tad bit of contrast at the end. Oh, and 5) the actions pause with the image-size dialogue box to let you set your own sizes. Both of these actions are pretty aggressive, but as I mentioned, the adjustment layer allows you to mask to taste.
Hope this helps.
Lon
But to answer the question, my methods are not different for digital captures or film scans. Just the strength of the application of them may vary.