shoenberg3: Not really, but it should probably be less aggressive as the initial resolution is smaller.
Heinz: I have LR configured to send a 16 bit PSD in ProPhoto RGB to Photoshop. After doing the PS postprocessing & resizing I use the "Save for web & devices" function. These are my settings:
No, actually, I don't add it to the LR catalog. If I do some significant PP in Photoshop, i.e other than resize & sharpen, I save the image before resizing and it's added as a PSD to the LR catalog.
The resized jpegs I only save to disk and upload to the web - I don't reimport them into LR.
DocsPics wrote:
My 2 cents from the lower end of the learning curve:
4. Sharpen last (always) (selctive sharpening can really add "pop")
5. Resize for web using bicubic sharpen
Using Bicubic Sharper isn't a very good idea because it's a rather blunt tool and can introduce stair steps or other artifacts. It's better to give more attention to sharpening after resizing (this is called output sharpening) than before it.
Gunzorro wrote:
Next question: what is the simplest way to pull 100% crops from images in LR?
You could make a virtual copy of your image, crop it and then export without resizing it. 100% crop means export without resizing. Then you could discard the virtual copy. You can also simply crop your image, export and then undo the cropping.
shoenberg3 wrote:
Denoir:
Do you have recommendation for sharpening workflow for a 5D classic? Currently, I am doing few resize steps, with just sharpens in between, finished by a mild USM.
I tried Luka's script ~6 months ago and found it was too agressive for my 5D's files. With some testing, I found that only one intermediate resize is enough to preserve detail and prevent stair steps. Two or more resizing steps offered no perceptible improvement.
I also found that the initial sharpens had a huge effect on the bokeh (making it very harsh) in some shots, so you really have to use a layer mask to not sharpen the bokeh. I haven't done exhaustive testing to find optimal settings, but nowadays I just resize to 1.5-2x final size, sharpen, resize to final size and then smart sharpen twice at 0.1 radius and strength 30-50. Then fix oversharpening artifacts with layer mask. I'm sure this can be improved, but I need to take the time to experiment with it more some time.
6. For printing: good luck...lots of patience....lots of "how did that happen"...etc
LR does a pretty good job with sharpening for print. PhotoKit Sharpener is also great for print sharpening using various algorithms based on your print media and technology.
Aham -- Thanks for the method. Suppose you want to do test comparisons (panels, or two adjoining images), having all shots exactly the same size and magnifcation? Is there a simple way to do that, or is each demo unique?
I'm not sure if LR can do particular magnifications unless you calculate the number of pixels. For precise crops it's easier to use PS(E). I don't know if there are any plugins for LR that make this easier, but I just use PS for such things.
You can get exactly the same crop if you simply copy the crop settings from one shot to the next though.
@robinlee: Did you have Photobucket resize it, or did you upload it as 1023px wide? Make sure you choose the advanced option to only limit the filesize (1MB I believe), not the dimensions. If Photobucket (or some other image host) resizes your uploaded image, some of the sharpening is lost again.
On the Photobucket upload page, there is a link at the bottom that says "Customize your upload options". Click the link and in the popup with the resizing/quality options choose: "Best possible quality (while staying 1mb or less)". The other options recompress your files, so you will lose quality.
Flickr is a bit better, but limits you to a 1024px maximum dimension. I use Picasa and it doesn't recompress if you don't let it resize your pictures.
AhamB wrote:
On the Photobucket upload page, there is a link at the bottom that says "Customize your upload options". Click the link and in the popup with the resizing/quality options choose: "Best possible quality (while staying 1mb or less)". The other options recompress your files, so you will lose quality.
Flickr is a bit better, but limits you to a 1024px maximum dimension. I use Picasa and it doesn't recompress if you don't let it resize your pictures.
It has come up a couple of times, so I thought I'd post a short tutorial on white balancing an image that has different tones in the highlights / shadows.
To illustrate the problem, we'll start with an image that was shot during the 'golden hour' with soft yellow sunlight.
We can see that the building that is supposed to be white is either yellowish in the parts in sunlight while the shadows have a blue cast. Trying to fix it with a simple white balance adjustment in Lightroom or similar software won't work. If we pick the grey point to be in the sunlit area we get this: http://peltarion.eu/img/comp/labWB/1-highl.jpg
If we on the other hand pick the area in the shadows as a reference, we get this:
We can however solve the problem by using LAB mode curves adjustments in Photoshop. LAB space is an alternative to RGB space. In RGB we have three channels, one for red, one for green and one for blue. In LAB mode we have also three channels and they are:
L = lightness/luminosity
a = green/magenta
b = blue/yellow
Something that for instance instance has a negative b value will be more blue than yellow.
The most important tool for color correction is the info box in photoshop:
It will show you the color values under the cursor. The first panel shows by default the actual color space while the second one shows CYMK values. I've reconfigured mine to show LAB space (click on the eyedropper icon and you can choose what it should show).
In the sunlit area we can see L = 82, a=2, b= 8. We're primarily concerned with b here (blue/yellow). A positive value of 8 tells us that there is a yellow cast. In the shadow area we have L = 36, a=-7, b=-20. A b value of -20 tells us we're well into blue territory. Neutral is 0. So we want to correct both sides of the building so that b is around 0.
This is how it's done.
1) Go to Lab mode (Image->Mode->Lab Mode)
2) Add a curves adjustment layer
It allows us to see where we are on the curve and add points on that position by clicking. It also allows us to change the curve by clicking and dragging the mouse up or down.
3) Go to b channel. We have a blue and yellow cast that we wish to remove, so it's the b channel we want to look at primarily.
4) We want to start by adding some anchors to the curve - i.e points we don't want to change. So we find one or two spots where we're happy with the color. I chose the blue sky and the yellow lower part of the building, but other options are of course possible. To add the anchors just click on the spots.
5) We now correct the blue/yellow balance at chosen adjustment points. In our case we want to eliminate the yellow cast on the sunlit part of the building and the blue cast in the shadows. So any two points on those surfaces will do. You just click and drag to change the curve (up=more yellow, down = more blue). Look at the info box while doing it and stop when you see that you are around 0 on the b channel.
The blue/yellow cast is gone, but there is definitely a greenish tinge to the whole thing. Green/magenta is the domain of the a channel, so we go to the a channel and repeat the process with the anchors and adjustment points. I just used one anchor here - the sky again.
Thanks for sharing, I have to try this some time on some of my images.
The left version still looks a bit more natural to me, perhaps because it looks like the sun was low in that shot, so I'd expect the light to be warm.
