Bifurcator wrote:
Yup, I think you're right. There's many many times where I will stop the script, undo the last sharpening step, and just select what I want sharpened - then continue the script from where it left off. I guess that describes about 25% of the images I edit.
Why not do sharpening on a duplicate layer and convert your selection to a layer mask (or just paint on the layer mask)? That way you can change it more easily if desired, at any time you want.
With a layer mask you can see in real-time how your selective sharpening looks, and where it needs more touching up. After every brush stroke you see how it changes, but with just the sharpening applied to a selection you can only go back and forth with undo/redo.
By the way, I think you're confusing the words "script" and "action". Actions are simply a recording of the operations you perform through the interface. PS also supports a kind of javascript (which you'd write in a text editor) that can do much more complicated things.
That'll work. Nik Sharpener Pro 3 has that built in for you too - which saves a bit of time - especially if you're doing many like that. The Asiva Sharpen+Soften plugin is also supposed to be like that as well but I've not tried that one yet.
Topaz DeNoise and Detail try to sharpen in a selective way automatically as well but it's on the micro-contrast level and doesn't affect edges significantly. Vivacity is Topaz's pro sharpener but I've not used it either.
My problem with selective sharpening or any other manual measure is that I typically need multiple resolutions of an image. The minimum is usually two versions - a 1024px (width) for posting here and a 1440px version for my own web gallery. Usually however I want to share some images with family and friends and then I need a 1920px version. If I intend to show stuff on my laptop i need a 1280px version. I also on occasion make photo books in A3 format which again needs different sharpening and processing.
So on occasion I need to make five or so different versions of the same image. So it is simply too much work to do manually. I therefor also like to do any manual processing in Lightroom and not Photoshop so that it is all at full resolution and reproducible.
1. resize image to 1,666 times the target size (ie if the target size is 1000pix, resize to 1666)
2. sharpen (that basic photoshop sharpen)
3. resize to target size
4. copy layer
5. add sharpen to top layer
6. add mask and paint the areas you want to be sharpest in the area, like eyes on portraits
I dont do much other pp except fiddle sometimes with recovery and fill light sliders in lightroom 3 for that "poor mans hdr". If I need that grunge look for some heavy metal band photos, I use topaz adjust.
Bifurcator wrote:
Masks resize with the image. I guess you know that already tho?
Yes, I know but any editing except for the final resize is problematic if I do it outside of Lightroom. LR is not just my RAW developer but also my main image library software. The organization is simple - I have a RAW file and any editing I do is stored in a separate XMP metadata file. If I do some sort of permanent editing in Photoshop I have to reimport the file as PSD or TIFF that doesn't have a history stack of all operations and I end up with duplicate images. So I try to avoid it as much as possible. PS is fine for me as a final output stage but not as an intermediate step.
Well, all I can say is that this is definitely proof that software does not a photographer make - because your images rock and by my testings LR is pretty low on the ladder of dedicated photo-editors - quality-wise.
I hear you on the PS thing being too permanent tho. Although if you open the images from ACR as "Smart Objects" then it's non-destructive all the way through. For me it goes like:
1) RAW acquisition (The Original)
2) ACR (The non-destructive interim edits - XMP files)
--- And/or CaptureOne if IQ is critical (completely non-destructive all the way to final)
3) If ACR then: PS (Permanent Edits for targeted output - .PSD)
4) Scaled Output (usually JPG but sometimes GIF or PNG)
1 = always kept.
2 = always kept. C1 is used if it's a "job" or I care much about the IQ personally.
3 = only kept if the target is hardcopy print
4 = I don't seem to care much. They're on the web for the web.
So I typically end up with just three files per image. The RAW, the XMP Sidecar file, and either the PSD file or whatever C1Pro creates.
But there's many roads to Roam and I'm not saying that my way is the best way for anyone other than myself.
Thank you! Yes, I definitely agree that Lightroom is weak as far as photo editing goes. It's basically just a glorified ACR wrapper . Still, I do like it as a photo organizer. I like the concept of keeping the all the RAW development as well as the organizing in one application and then just render for different targets.
It works OK, my two biggest problems being the multi-resolution output as I do need to go into PS on occasion (especially with images shot with Leica X1 shots) and the fact that I have a gigantic backlog of images that I need to tag and rate.
It's my intention, come say - November - when the days go darker & cloudy - to finally spend the time and practice a work flow and correct PP. I have and will have many hundreds if not thousands of non-processed images to practice with. Thanks for all who have contributed on this and other FM Alt threads on this subject.
denoir wrote:
My problem with selective sharpening or any other manual measure is that I typically need multiple resolutions of an image. The minimum is usually two versions - a 1024px (width) for posting here and a 1440px version for my own web gallery. Usually however I want to share some images with family and friends and then I need a 1920px version. If I intend to show stuff on my laptop i need a 1280px version. I also on occasion make photo books in A3 format which again needs different sharpening and processing.
So on occasion I need to make five or so different versions of the same image. So it is simply too much work to do manually. I therefor also like to do any manual processing in Lightroom and not Photoshop so that it is all at full resolution and reproducible. ...Show more →
Like Bifurgator says the masks resize. Also masks are much easier to draw when you have fullsize image, just enlarge 100% and draw, you have much more room for error.
For important images (e.g. ones sold to customer) which I have done PhotoShop work (e.g. selective sharpening) I always save to PSD file to Aperture (I would assume Lightroom is able to do same) and I save it with layers on fullsize. Then later doing resing is pretty easy to use layer mask; just dublicate layer before resize/sharpening and only to the top layer, then apply mask to it and the other layer below shows on "bokeh areas", and if you wish you can even blur the background layer.
However I try to avoid the PhotoShop as much as I could and have managed to keep >99% of my images Aperture edit only and just resize in PhotoShop (mostly just via Finder, never actually using PhotoShop).
That sounds like a pretty good way. One thing I notice is that my A4 Wacom tablet gets used a lot when doing methods similar to what you're describing. It really makes the drawing part fun and fast. So, that might be a tip for people looking into these practices for the first time - think about getting a tablet maybe - (for me the bigger the better YMMV).
Bifurgator, good point, didn't remember to mention that since after first time I draw a mask with Wacom there just has not been any other alternative... later I have find it out useful also for some edits in Aperture (in newest version adjustments can be painted in, similar to having adjustment layer in PS and applying effect only to some parts of the image)
The special sauce sharpening routines have been promoted for as long as digital has been around, to be joined later by High Pass and Smart Sharpen and so on - all have their enthusiastic proponents. There are some fine apps made by Nik and others as well. Probably all work well if used judiciously. I use PKS.
I use RCs for the job they must do (demosaicing) and little else - due to the variable quality of tools and hidden adjustments (and assumptions) they often make. I only do WB and overall tone levels in the RC.
Layers in PS do not damage the image, and retain full flexibility with much more control than many grafted on tools with fancy names, for example 'clarity', 'vibrance' etc. The marketing guys are in full swing! Maybe we will soon see 'beauty' as a tool name.
Separating luminosity from colour is vital to preserve colour relationships in editing. Rather than using LAB (the Margulis way) I use a set of luminosity masks within special RGB color spaces and variant profiles developed by Joseph Holmes (DCam3 for digital) which are great containers of typical image gamuts.
I use Blend Modes in PS (in empty curve layers) for contrast and tone control as they use actual pixel values to apply the edits: Screen, Multiply, Soft Light and Hard Light. Much more accurate than hand drawn curve adjustments, and you get the percentage slider for fine tuning. Using these with luminosity masks preserves color relationships and retains a realistic appearance. I work from global to local in all edits.
I fine-tune colours for images that do not respond to careful WB using Selective Colour, a very effective tool.
Very informative thread, thanks to all that contributed.
I have a problem, that i cant pull out all the detail from original into the "for web" image. Here is one example...I sharpened it as denoir suggested. To my eyes it looks a little bit oversharpened. And also the very small details seem to got lost trough resizing.
Any suggestions maybe? I would love to send the original photo to denoir or samuli to see how would they sharpen it with their scripts. Anyone of you willing to do this?
To keep all the information of the original image you have to display it in it's original size - there's no getting around that. It's why a lot of folks here like to post large-ish images thus losing as little as possible.
There was a poll thread here awhile back about preferred size for this site and the majority said that the upper end of 1200 to 2000 pixels was best for landscape mode. Most said that smaller than 1200 was too small. That's what I like too but many sites frown on or don't allow images that large.
Anyway, I wouldn't fret the loss of some detail. It's the price we pay for scaling. Maybe in the future there will be an image format that losslessly scales to a wide range of sizes and is suitable for the web. There's already been a few attempts at this - I'm sure more will follow.
@Philip: Your post raises lots of questions for me and others too I'd think.
RC = raw converter I presume
PKS = Photokit Sharpener? (after a google search)
I use a set of luminosity masks within special RGB color spaces and variant profiles developed by Joseph Holmes (DCam3 for digital) which are great containers of typical image gamuts
Could you explain in more practical terms how/why you use this DCam3 profile and how it makes your PP easier? I've taken a brief look at http://www.josephholmes.com/profiles.html but the different pages are almost a wall of text. I'll read it in due time, but it would be nice if you could give an example that shows why it's worth my while.
I would appreciate it if you could also elaborate on your usage of blend modes (and selective colour) by giving some example(s). The way you're not explaining it now doesn't really teach me anything.
sodjdobe: As Bifurcator says, details will be lost when you resize as you don't have the same amount of information. Your image does indeed look quite oversharpened. Feel free to send me the original and I'll run my script.
I think there are two possible explanations - the source image may be too small (my script is suitable for an 21 Mpixel image) or another possibility is that you used bicubic sharper resize instead of regular bicubic in the intermediate steps. If you post the full image I'll take a look at it.
rdst wrote:
A good method is to look at itunes under the podcast section. There are few very good podcasts concerning CS4/5 or Lightroom, for example Rich Harrington or other. You will get 1 or 2 6 to 10 min long podcast per week, thats not too much and learning by doing is the best way.
Podcasts and/or videos take too much time. Why can't they write it out? I can read something in a minute what it takes them to ramble on in a video/podcast for 10 minutes. I can read it completely in the time it just takes for the video to load and get past any stupid advertising. Do they do it that way just to hear themselves talk or do they just like to waste peoples time?
