Color correction - face shot
djsilver666666
Registered Users Posts: 77 Big grins
I'm still pretty new with using curves, and the only trick I use on a regular basis is clipping the 0-15 and 245-255 since it isn't printed. But I've tried a couple of more advanced manipulations, such as balancing RGB to get a pleasing skin tone. The results were acceptable at best
Well, today I took 2 photos for a roster page of a soccer team I help coach. Unfortunately it was cloudy and this was the only venue I could take the picture at, so I will have to make-do with what I have.
Since it's late, my edits are completed, so they will be up tomorrow. Here are the originals:
There is certainly a green hue cast by all the foliage...
Well, today I took 2 photos for a roster page of a soccer team I help coach. Unfortunately it was cloudy and this was the only venue I could take the picture at, so I will have to make-do with what I have.
Since it's late, my edits are completed, so they will be up tomorrow. Here are the originals:
There is certainly a green hue cast by all the foliage...
0
Comments
Here is my edit of the second picture:
And if anybody wants to know, here are the changes I made:
(pic one)
(pic two)
Any feedback?...
- Curves for color correction (more red, less blue)
- Selective color to remove a little cyan from the reds and tone down the greens (add black to the greens)
- A slight curve for contrast
- A pinch of shadow/highlights masked to the eye area to remove some of the shadows around the eyes
- A little sharpening
- A tighter crop
I think mine looks a little oversaturated in my non-color-managed browser (that will vary from computer to computer), but I like the way it looks in Photoshop when it's color managed (which is presumably how it would print). Both edits look good. You might want to consider a closer crop.Here's mine:
Your edit:
Your original:
Homepage • Popular
JFriend's javascript customizations • Secrets for getting fast answers on Dgrin
Always include a link to your site when posting a question
Could you post a screenshot of your curves edit? I don't know how to do some of them:
- Selective color to remove a little cyan from the reds and tone down the greens (add black to the greens) {is that something other than the individual curve channels?}
- A slight curve for contrast {is that using the RGB curve?}
- A pinch of shadow/highlights masked to the eye area to remove some of the shadows around the eyes {using what? I don't know what you mean by 'masked'}
What did you use for the sharpening?BTW I do like your edit better, I thought the second photo was harder to edit.
Sorry if some of that are things I should probably know... I've gone through the "Moderator's Choice--The Best of Finishing School" thread and tried to pick up some of it.
Color correction curve. When I looked at color samples of the skin at several different brightness levels, I found this adjusted all of them better than just pulling the middle of the curve down. Pulling the middle down, didn't correct the brighter parts of the skin as much as I thought it needed. The reds were void of data in the highlights (you can see that in the red histogram) so this didn't cause any clipping.
Selective color on the reds:
Selective Color on the Greens:
Curve for contrast (this is just an overall RGB curve, with the adjustment layer set to luminosity blend mode so it won't affect color):
Homepage • Popular
JFriend's javascript customizations • Secrets for getting fast answers on Dgrin
Always include a link to your site when posting a question
You can literally just drag the adjustment layers from the first photo over to the second photo and apply the exact same changes to it. I found that worked pretty well. The only thing that's really different about the second image is that it has more highlights because of the white t-shirt. The colors seem to be happy with the exact same treatment as the first.
In something like Adobe Camera RAW or Lightroom, you could probably just apply the same white balance corrections to all your images from the shoot.
FYI, a small amount of fill flash at shoot time might have really helped with the shadows in the eye area.
Homepage • Popular
JFriend's javascript customizations • Secrets for getting fast answers on Dgrin
Always include a link to your site when posting a question
Wow, that does look pretty good when you just drag the curves layer from pic one (and saves a lot of work). I suppose that is probably because it was the same setting and they are similar in skin tone.
By 'fill flash' do you mean 'flash exposure compensation'? (I don't know a lot of photography lingo, so I tried googling it) I don't know when to make judgment calls such as using fill flash, is that for shadows on the face?
Fill flash means you are using your flash to supplement the existing lighting. The main exposure still comes from the ambient lighting, but you are using some light from the flash, either to illuminate a subject that isn't as bright as the background or to help fill in shadows. In this case, it would be to fill in eye shadows caused by the predominantly overhead lighting.
The general idea is that you set your flash to TTL-BL which tells the camera and flash that you want the camera to adjust exposure for the main image and background and the flash will adjust it's power to supplement the lighting at your focus distance and aperture. I then typically set my flash to a small negative EV (-1/3 or -2/3) so it doesn't overpower anything and doesn't look like I used a flash.
I will almost always use some type of supplemental lighting for outdoor portraits. I find it absolutely required in direct sun and useful in many other circumstances. On camera fill flash is the simplest form of supplemental lighting.
Homepage • Popular
JFriend's javascript customizations • Secrets for getting fast answers on Dgrin
Always include a link to your site when posting a question
Normally I shoot in Av mode, and did so when I was taking these photos. If you didn't gather this from the photo meta-data, I have a Canon XTi, and used F5.0, ISO400.
When using fill flash, should I just pop the flash up and use a different EV depending on circumstances?
Remember the EV setting for fill flash needs to be the flash EV, not the camera EV. You will have to find out how to change the EV on the pop-up flash on your Canon - I presume it's in the menus somewhere. I shoot with the SB-800 on my Nikon and it's a press of a button on the flash to adjust the flash EV on that (I don't even have a pop-up flash on my D2Xs).
For something like your recent shoot, you can probably find time for one or two test shots before-hand with flash where you examine both the histogram and the image on the camera screen so tweak the flash setting and any other exposure settings.
Is there any reason that you didn't set your camera's white balance for this shoot so the images would come out of the camera much closer to correct?
Homepage • Popular
JFriend's javascript customizations • Secrets for getting fast answers on Dgrin
Always include a link to your site when posting a question
I'll have to remember to take practice shots beforehand, though I'm not very good at reading histograms...
Yes, there is a reason... I don't really shoot portaits often, and I don't think of things like changing white balance. But that's why I'm here, to learn more! So I appreciate you explaining some of this to me. I've bought "Skin" by Lee Varis, but due to school I haven't gotten to start reading it until now.
FYI reading in Skin about using curves for color correction is how I even knew what to do. These two photos are the 3rd and 4th portait shots I've tried editing with curves, so I'm still pretty much a newbie at this...
Try, ask, read, learn, repeat. I'm no expert, but that's how I got to where I am. You're off to a good start.
Here's a good reference on histograms. I find them very useful for gauging exposure. Histograms will tell you if you have an exposure error that is causing you to lose data (clip shadows or clip highlights) and they will tell you if you are capturing the best possible shadows without clipping highlights. They won't tell you if your image is properly exposed because how the histogram "should" look for a proper exposure is very dependent upon the scene you are shooting and it's distribution of tonal values. Contrary to popular guesses, there is no ideal histogram shape. It's entirely scene-dependent.
Homepage • Popular
JFriend's javascript customizations • Secrets for getting fast answers on Dgrin
Always include a link to your site when posting a question
unless you set it off a gray card there is so much green that most cameras will create a poor color balance. Shoot in RAW and shoot a test shot with a gray card held next to the person's head.
For a quick color correction pull a white balance off of the near brightest white of a shirt.
Cheers,
David
Normally I am a sports photographer, and my card would not have enough room to shoot them all in RAW. But would it be helpful for the sport shots if I took one pic in RAW with a gray card and then applied the changes to the JPEGs?
BTW I did use the brightest white part of the first pic's shirt. Learned that from the 'Skin' book!
Yes, when in doubt, looking for an area that would be perceived as white or black is a good place to start for inspecting neutrals (although sometimes one may have to account for tones and casts which are not the darkest or the lightest part of the image and are more highlight/quartertone or whatever).
The girl with the black shirt, the shirt reads as an average -10A in Lab info palette readings (presuming that the file is sRGB), which is quite green (0 zero being dead neutral, negative values cool and positive warm). The girl in the white shirt, whites are negative AB Lab values. So it is "obvious" that areas that a human observer would perceive as being more neutral, the camera does not. This is also the case throughout the images.
Anyway, my basic steps were:
1. Using the black and white areas of both images, to come up with separate channel curves that roughly made the shadows and highlight areas more neutral. This may vary from shot to shot, but would probably be similar if the lighting was fairly consistent. The adjustment layer was set to COLOR blend mode, so as to try not to overly affect tone.
This helps reduce some of the green cast...and improves the skintone and overall colour quite a bit.
2. I also used a channel mixer adjustment laye set to color blend...red channel I added about 5% green channel, green channel I reduced to 95% green and the blue channel had the green set to +5 and the blue to 85 from the defualt 100. Channel mixer can be tricky for colour corrections, although in this case the image lends itself well to such minor corrections over and above the minor curving.
3. You should notice green reflections on their faces near the chin from the bibs...not good! I added a blank layer set to color blend mode and used the clone tool to copy good skin variation over the green reflected areas.
4. An optional step could be to do some channel blending, of say the red channel in luminosity mode over the entire image, using layer option blend if sliders to reduce the effect in the shadows and midtones, while adding some tone to the sponsor on the bib.
5. One may also perform some minor, subtle lightening/whitening of eyes/teeth.
Anyway, the first four points are illustrated below with the left hand side being the original image.
Best of luck!
Stephen Marsh
http://members.ozemail.com.au/~binaryfx/
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/
Here's a bunch to choose from. You should get one that's specifically made for setting white balance.
Homepage • Popular
JFriend's javascript customizations • Secrets for getting fast answers on Dgrin
Always include a link to your site when posting a question
A good non-specular gray card can really help - One I have used is the BalanceSmarter reflector. I have a thread about its use, among other tools, here. The Expodisk and the SpectraSnap filters can also be used to color balance in the camera jpgs.
It takes a brief extra step to shoot a custom color balance frame, but it is really nice, later in front of your computer, to be greeted with great color coming straight from 'out of the camera' jpgs.
Moderator of the Technique Forum and Finishing School on Dgrin
I've been checking back periodically because people have kept adding, which I appreciate. BinaryFx's post seems more complicated, still trying to digest what he did.
Since I'm still no expert in picture finishing, I use this is a quick-guide reminder when editing new pictures. That said, there is another picture that I'm editing, will post the original and my best attempt with what steps were done sometime tomorrow!
Edit: couldn't resist giving it a try tonight!
Original:
Edit:
Edit2, sky change:
For the edit, this is what I did:
1. Used neat image to auto profile and reduce noise.
2. Used 'auto contrast' (Alt+Shift+Ctrl+L) and applied this contrast curve, set to luminosity:
3. Adjusted the color, just end-point changes except on the blue - In:131, Out:121
4. Using the channel mixer in the red channel, increased the red by 5% to add warmth
5. (This is for the Edit2) Just messing around, and I liked the change to the sky so I kept it. There is probably a better way of doing this, but here's what I did:
Final list of layers, and my point values:
If you have the download speed and the time to putter around, here is my edited psd file: http://www.filefloat.com/temp/Picture 074.psd (32.3 MB)
Comments?
It's probably just me nitpicking, but I noticed in the first photo the background is kind of blurred, while in the second the girl is almost standing in the branches, so they're in focus. I would recommend a long lens, and have them stand a good ways away from the background, to increase the DOF difference. It also makes the subject stand out more.
Like I said, just me nitpicking
I agree, but this isn't even my picture, and was taken with a point-and-shoot camera (Sony DSC-W80). So that wasn't really an option, though certainly a good suggestion.
That reminded me of a question I meant to ask, should I leave the tree in? blur it some? leave it alone? I'm not really sure...
It kind of suprised me that no one has really comented yet, hopefully that means I did a good job
If it were my image, unless there was a specific reason for keeping the greenery, I would clone it out, as I prefer a simpler background.
Moderator of the Technique Forum and Finishing School on Dgrin
(sorry this isnt very topic friendly)
thanks!
Unbrok3n, I learned a lot by reading through some of the topics listed in http://www.dgrin.com/showthread.php?t=56678 but experimentation (since every picture is different) will truly help you the most.
Consider buying a good book too, I would suggest 'Skin' by Lee Varis. You will learn a strong base in how to make adjustments with curves, and the basics on how to remove color hues (Skin is the book I used to get started).
So does experimentation work even if you dont know all the numbers and stuff?
Yes, experimentation does work. I don't know how you tend to be, but I like to try things and figure out their properties by what the effect is upon the photo. While I do like to experiment, it is good to have a rudamentory knowledge of how things work, that way you aren't trying things completly in the dark I really do recommend getting a book if you are serious about learning.