Rather than post my end result, I'll write down the steps that I take (or would take) so that other can see and critique more than just the final output. IamSam has taken a more visual route to showing step by step, but I am not diligent enough to do such a fine job of illustrating the work. Please feel free to critique the process so that I and other can learn.
This is what I would do:
Open file (jpg)
Save file as PSD
If the image did not have a border, I would choose Image->Canvas Size to increase the size by about an inch with the picture centered. I do this to prepare to use the Lens Correction tool which can lop off part of the image otherwise.
Then I straighten the image. In this case the original picture borders look good enough. In other cases, the original image was not straight or would benefit some perspective correction.
Filter->Lens Correction
- Use the line tool to straighten the image.
- click ok
Then I crop out the original border. It is too much of a mess and a new board is easy enough to create later. I've also noticed that the border can through off other auto correction tools.
Choose selection tool (rectangular marque tool)
Make a selection approximately to the original image content.
Recently, I learned a new trick to look at the channels. Someone else did that here. One of the channels can be clean of the damage, especially the orange on the shirt.
Duplicate the layer if using all channels, or create a new layer and copy a channel to it if that proves best.
On the new layer, fix the flaws in the image using healing brush and content aware. I've noticed that others are correcting the color and tone first. Is there a reason for this. I like to fix the flaws first since it has the original colors that may help in fixing. I also like having a repaired image to then play with multiple choices that I may make later. If you correct the color first, then you are stuck with that choice. (Unless you do that with a mask, which I suspect I should be doing but never do.)
For large flaws, I use content aware fill. I've notice others I using the patch tool. I never can get the hang of that tool.
For larger flaws such as the one on the left part of the shirt, use the magic wand to select the defect. Play with the tolerance to get the desired selected area.
Select->Modify->Expand to expand the select by 5 or 6 pixels. I do this to avoid the line left by content aware fill
Choose rectangular marque tool
Right click on selection and choose content aware fill.
If it is not acceptable, then undo. Perhaps repeat with a smaller selection, or use a different method.
For smaller flaws, I use the healing brush tool. For large damage, and when appropriate, I will replace the pixels, but more often I try to stay with normal mode.
Select healing brush tool
Adjust the brush to be about twice the size of the defect.
Alt click on the place to sample from. Usually this is a nearby place.
If trying to heal an edge, alt click right on the edge in a good place and then heal the flaw on the edge.
(If someone could explain why the healing tool results in dark blurs around edges even beyond the brush or transparent pixels near the edges, I would appreciate that.)
Sometimes it is easier to erase and start over in a section. In this picture, the orange area on the left background is easier to erase and then fill in with content aware fill than it is to fix the orange, in my opinion. (This is assuming that using a single channel in the beginning didn't fix the problem). This only work in areas in which there aren't important details. But the background is easy enough to fix.
I almost never use spot healing brush tool. The outcome has felt too random compared to the healing brush tool.
I've noticed that others have fix the background to the point of being completely smooth. How did you judge if this was the original look of the photo. For instance, the dark triangle shape on the upper right looks like it may have been part of a background style or may some back lighting and shadows. In my opinion, when posing for a portrait, the backgrounds have some design or light to making them interesting. With a smooth background, it starts to look like an ID photo. That is just my opinion, but how did others determine if the background ageing caused flaws or if the originally looked like that?
A favorite tool of mine is the Polaroid Dust and Scratch Removal Tool. It is really old. It was also free, so an even better plugin to have. Despite the age, I haven't found a better tool though to help with removing the tedium out of healing. Use it repeatedly on different areas made from a selection. This allows you to adjust the amount of feathering and defect removal that it does.
Keep healing the flaws until all that remains is the orange on the shift on the right. Sometimes I will heal these too. This is likely what I would do there. Sometime, there is useful information there, but in the wrong color. In which case, save these until later.
Duplicate the layer. You now have the original and a healed layer to go back to if needed.
On the new layer, auto tone, auto color, auto contrast. See what each does, and undo if you don't like the change. I don't always use all three. It just depends on the picture. I'm sure that the pros use adjustment layers here, I've not been comfortable with that (being a noob). I found that the auto tuning many times does a good enough job. I would guess the pros use the curves. I wish I was more comfortable with that interface.
For images that are too dark, or have almost blown highlights, I use Image->adjustments->Shadow Highlights to get more detail out of dark areas or tone down light areas. Adjust to taste, but not too much. For Shadows, I tend to use about 35% for amount on a 5% tonal width with radius between 6 and 30. For Highlights, I often use 0%, but sometimes about 20% amount, on a 5% tonal width with 6-30 radius.
I suspect the pros use curves to tune more. I am not comfortable with that interface. I tend to make small adjustments with Image->adjustments->levels and Image->adjustments->exposure instead.
Duplicate the layer.
I now try to tackle the orange on the shirt if the other methods above haven't worked.
I often hope the problem will go away if I convert to B&W. I use a Filter plugin, but a simple one can be found at Image->Adjustment->Black and White. If so, play with hue or apply a photo filter and be done. Usually, things are not that simple.
Duplicate the layer
Run your favorite noise reduction filter. Since the options vary, I'll just say that it is better to not over apply it. Focus on what it does to the eyes and lips. Too much blur elsewhere can be forgiven, but overly bad eyes and lips detracts from the image. Everyone has their own favorite. I think Nik Dfine is free now, so even if you haven't purchased one, this one is available to you.
If the orange on the shirt is still a problem, using the magic wand tool to select the color, or use Select->Color Range to choose the color and create a mask. Then adjust the color on that mask to match. I struggle with getting this right; it usually takes several tries.
At this point, the image should be pretty good, but may need more punch. I usually will create a mask to separate the person from the background. Many times, I will will create a second mask for the clothes. I wish I was better at making masks. How did others create one for this image? I usually just use the quick selection tool.
Now, I apply more blurring or noise reduction to background. I think a "crunchy" background is distracting, but it shouldn't be completely smooth either. I also play with hue and saturation just little a bit.
On the person, I'll sharpen with the unsharp tool just a little bit. Again, I might adjust the tone a bit or maybe add a slight photo filter to avoid looking too gray. A little extra exposure might help. Select use of the burn and dodge tool might help
I would appreciate guidance as to when a adjustment layer should have been used, or a different tool. I know the pros use more layers where I tend to make changes to just the top copy.
I hope the helps to start discussion and/or critique.
