Printable View
The new Image Adjustment Lab is a powerful tool. The masking here was done with the wand and brush selection tool.
I did a lot of refinement along the edges of the rope with the clone tool.
I usually put the original on the left, however, the natural color of the rope here seems so much like it should have been the original. The one on the right is really the original. Honest.
That looks like a lot of work... How long did it take?
It required making two masks, one for the blue legs of the saw horse and its top, the other for the rope.
Because I wanted as few errors in making the masks, I used the path tool, then converted the path to selection and made a clip mask. This gives you a mask you can edit with black or white or shades of gray.
I did the same thing with the rope.
Then I entered the Image Adjustment Lab and changed the hues on both the legs and the rope. The rope was more difficult but the adjustment was made at one shot for both.
Then it was just a matter of touching up a few edges that didn't look quite right on the rope, this was done with the clone tool and the smear tool. And I was done.
Took about 45 minutes all together.
It's interesting you've posted this excercise Sally. I've just recieved Knockout 2 this morning. Haven't been able to give it a go yet, but it will be interesting to see if it is more efficient than the method you describe, which is basically how I've been working in Photoshop.
Photoshops Extract tool is very poor and no quicker than manually tracing and editing the mask. Hopefully Knockout will be particularly good with more difficult subjects, like hair etc. When I get some time I will give it a thorough test.
Sark
This picture is bigger than what usually gets posted here so I zipped it.
You may wish to do an actual comparison, Sark, so I am posting the picture I used, but zipped.
Hi Sally.
Will give it a go. Already I suspect it will be very easy. Only had a quick play with Knockout, but I'm already very impressed. The Dragonfly below took minutes (The top is original). It's on hair that it really shines. Below is a link to a .psd file that shows one of the included samples after I extracted and replaced it's background colour. You make two selections, one which shows what you want to keep in the foreground and another that you want removed from the background. The .psd shows how crudely I made these extractions (bottom layer) yet it still retained the fine detail in the hair. And again it took minutes.
http://www.yousendit.com/transfer.ph...8DFC976334FBE8
Sark
Wow, that is really quite impressive.
So is the girls hair a challenge, I haven't bought KnockOut yet, but would like to. Too many expenses.
But it sure tempting.
Hi Sally.
At first I thought your challenge would be quite straightforward, then I realised the original was the blue wood and rope version, not the other way round. Extracting the blue rope from the blue wood was always going to be a challenge (for any extracting tool). As it turned out it wasn’t too bad. First I made two separate extractions. One of just the rope, this needed a little reconstructing on some edges in PS. This was followed by an easier extraction of both the rope and wood together. Then I just used Hue and Sat in PS.
This all took about 15 minutes. The main problem is that the image, on close inspection, has some jpeg artefacts around the edges of the wood and rope (exactly where Knockout processes information). Also, my Wacom has died and I’m making selections with a mouse. Using Knockout’s Polygon selection mode helped here, but the closer you can get to the edges, the better the results, and a pen and tablet would speed things up here.
After the above method I tried a different approach. In PS I just used a Hue and Sat adjustment layer above the image, and using its mask, painted in the colour of the rope. I timed myself and this took just 4 minutes. I placed a duplicate of the original image above this and extracted its background using Knockout. A Hue and Sat adjustment layer was used on this duplicate. Its mask was controlled with an inverted selection from my rope layers Hue and Sat mask. This hid the rope. I could now alter the colour of just the wood. This took just 4 minutes again, both to extract and colour. So, 8 minutes total for the image shown. I’ve included the .psd to show more clearly what I did.
Just got to save up for a new Wacom now.
http://www.yousendit.com/transfer.ph...98A80472C7390A
Sark
The extraction is very good. Sorry to hear about your tablet.
There are differences between our two renderings, as though the sun was stronger in mine, yours makes me feel it is going to rain or storm as the color temperatures are different. This is fine, they don't have to be the same.
When light is more diffused, edge contrast is normally less. With .jpegs there can be artifacts which can create an intermediate color which, depending up on the color change, may make a difference.
On mine when I was done I sampled the color and stroked the outside of the saw horse with a 2 pixel line and faded the color in so it looked right. This completely obscures the artifacts left by the original .jpeg.
It standardly isn't in this type of tutorial. It is just an eyeball kind of thing. My son, Nathan is quite the art critic, he won't let me get away with anything. If it doesn't look natural to him, I have more work to do. But then it is good, he is more critical than the customers at work.
The other thing which can happen in shadows with a hue shift is the shadows can look a little unnatural. When building a shadow originally, you'd apply Multiply mode to wed the darkened pixels with the color. When you don't know and are working with how it should look in you imagination, I look for similar pictures with similar coloring and lighting until I know exactly what color I need.
In this picture there is more than one light source, the rear lighting is different between the two of ours, but the forground lighting as well as the rear agree on both or ours, which is cool.
Your rope looks like it got wet and darkened in the process. Mine was used by workmen who only used gloves and therefore never got it dirty. Yours is more realistic.
The colour differences are not so critical here. You have a lot more freedom with this image to create some quite wacky variations. As long as the Saturation was not too high, I found I could have virtually any colour combination for the wood and rope.
It's an interesting point though, because some subjects are not so tolerant. The girls hair for example, only allowed light colours to replace the original white background. Darker colours created undesirable effects on many areas. This is just the nature of how hair will photograph with different backgrounds and lighting situations. Knockout leaves transparency in some areas for a more natural blend with the host image, but this won't work well in all situations. Ideally if you were taking the original photo yourself you would shoot with a background and lighting that simulated as closely as possible the host image lighting.
Still playing with Knockout. No doubt I've still a lot to learn to get this most out of this app. Still, very impressed so far.
Sark