p.1 #1 · Exposing to the right, processing the Raws
just want to make sure that I,ve got this straight as it seems counter intuitive
I,m exposing to the right with my 7D normally plus 2/3 until just before the histogram hits the right hand side , then move the exposure back in ACR
I understand that its to reduce noise
I get that bit its the post processing that I'm not sure about
Ive got CS5 and use ACR to process the raws
This butterfly is an example I overexposed by plus 2/3 then in ACR moved the exposure slider back by 0.25 until it looked right then moved the recovery slider until the small amount of overexposure blinkies were gone
is that the best way of doing it? http://img.photobucket.com/albums/v490/lcpete/IMG_5739.jpg
p.1 #3 · Exposing to the right, processing the Raws
James_N wrote:
What happens if you press the Auto button in Adobe Camera Raw rather than making the adjustment manually?
just tried it and it came out a bit dark , underexposed
p.1 #6 · Exposing to the right, processing the Raws
Well, there are lots of approaches you could possibly use. The Auto tool isn't always perfect but in most instances it sets the correct black and white points of the image. Then you could use the Fill Light and Brightness sliders to adjust the image to taste.
p.1 #7 · Exposing to the right, processing the Raws
What makes photographs look "seen by eye" normal is when they have the same amount of detail seen by eye. The Catch-22 is that a digital sensor can't record the full range the eye sees with most outdoor lighting, overcast days and open shade being a couple exceptions.
So exposing to the right is only 1/2 the solution to correct exposure — not blowing the highlight detail— and the better tool for determining when the highlights are getting blown out is the clipping warning rather than the histogram. The clipping warning makes every pixel on the playback a spot metering zone for the highlights. You'll see when and where small highlights clip well before they will register a blip on the right side of the histogram.
The clipping warning is based on the JPG generated by the camera and the RAW will have more headroom. But if at some point at the end of your workflow you convert RAW to 8-bit sRGB JPG you'll need to allow room for the highlights to change during PP. I find that by exposing per the clipping warning in the camera I don't have clipping in my JPGs. If I expose the RAW to the bleeding edge of clipping at capture my JPGs wind up with lost detail and clipping red channel in skin highlights.
Once the highlights are exposed optimally via the clipping warning looking at the left side of the histogram will tell you whether or not the scene exceeds the sensor. If the histogram is piled up on the left bracketing with shutter or aperture until the left side is barely touching will tell you how much the scene exceeds sensor.
If the scene exceeds sensor you can change camera angle — flat lit scenes have lower contrast than cross- or backlit ones — or use flash to change the foreground range to fit sensor range.
In PP you can move the midtones in the file independently of the end points. For example if you open a file in Levels and move the middle slider left you can pull much more detail out of the shadows...
The problem is that at some point in the shadows all that is amplified is the noise. Doing similar adjustments in ACR on the RAW file produce better results.. http://super.nova.org/MP/HSS/_MG_5034_ACRedit.jpg
What I do to avoid the shadow noise problem is expose the ambient for the sunny highlights below clipping the use a pair of flashes in front to match foreground to sensor: http://super.nova.org/MP/HSS/_MG_5037.jpg
So half of the exposure solution is "expose to the right" with the clipping warning. and the other half is "fill to move the shadows to the right" to the point the sensor can record detail in the shadows. The way the photographic process is engineered when you can manage to get detail in the highlights and the shadows both at the same time everything in-between winds up looking seen-by-eye normal.
p.1 #8 · Exposing to the right, processing the Raws
In principle, if you exposed to the right then you probably need to process to the left. Discrepancies arise between how far to the right and how far to the left, in part because in-camera settings such as contrast affect the histogram you see on the camera, and your software has its own different settings and gives you a different result.
There are many ways to achieve this and no one way is best for every image.
Personally, I preferred to reduce exposure and increase brightness rather than use much recovery, but the newest Lr and ACR might change that. Reducing exposure brought the highlights down, as well as mid tones, and increasing brightness restored the mid tones but not the brightest highlights. Recovery did relatively weird things to too many contrast edges.
p.1 #10 · Exposing to the right, processing the Raws
James_N wrote:
Well, there are lots of approaches you could possibly use. The Auto tool isn't always perfect but in most instances it sets the correct black and white points of the image. Then you could use the Fill Light and Brightness sliders to adjust the image to taste.
Just tried the auto tool on a different shot and it got the shot pretty much spot on very similar to what I achieved by setting the black and white points in levels
see what you was getting at now
I have now got a way of setting the black and white points in raw thanks James
p.1 #12 · Exposing to the right, processing the Raws
cgardner wrote:
What makes photographs look "seen by eye" normal is when they have the same amount of detail seen by eye. The Catch-22 is that a digital sensor can't record the full range the eye sees with most outdoor lighting, overcast days and open shade being a couple exceptions.
So exposing to the right is only 1/2 the solution to correct exposure — not blowing the highlight detail— and the better tool for determining when the highlights are getting blown out is the clipping warning rather than the histogram. The clipping warning makes every pixel on the playback a spot metering zone for the highlights. You'll see when and where small highlights clip well before they will register a blip on the right side of the histogram.
The clipping warning is based on the JPG generated by the camera and the RAW will have more headroom. But if at some point at the end of your workflow you convert RAW to 8-bit sRGB JPG you'll need to allow room for the highlights to change during PP. I find that by exposing per the clipping warning in the camera I don't have clipping in my JPGs. If I expose the RAW to the bleeding edge of clipping at capture my JPGs wind up with lost detail and clipping red channel in skin highlights.
Once the highlights are exposed optimally via the clipping warning looking at the left side of the histogram will tell you whether or not the scene exceeds the sensor. If the histogram is piled up on the left bracketing with shutter or aperture until the left side is barely touching will tell you how much the scene exceeds sensor.
If the scene exceeds sensor you can change camera angle — flat lit scenes have lower contrast than cross- or backlit ones — or use flash to change the foreground range to fit sensor range.
In PP you can move the midtones in the file independently of the end points. For example if you open a file in Levels and move the middle slider left you can pull much more detail out of the shadows...
The problem is that at some point in the shadows all that is amplified is the noise. Doing similar adjustments in ACR on the RAW file produce better results.. http://super.nova.org/MP/HSS/_MG_5034_ACRedit.jpg
What I do to avoid the shadow noise problem is expose the ambient for the sunny highlights below clipping the use a pair of flashes in front to match foreground to sensor: http://super.nova.org/MP/HSS/_MG_5037.jpg
So half of the exposure solution is "expose to the right" with the clipping warning. and the other half is "fill to move the shadows to the right" to the point the sensor can record detail in the shadows. The way the photographic process is engineered when you can manage to get detail in the highlights and the shadows both at the same time everything in-between winds up looking seen-by-eye normal.
thanks for that thats fantastic a lot to think about
have now enabled the highlight alert on the 7D
I do see what you mean, yesterday I was shooting a white butterfly, the histogram looked OK but it was a bit blown out at plus 2/3
luckily I bracketed and the plus 1/3 exposure was spot on
I now realise that I should have been exposing to the right on my previous camera as well (550D) but its much quicker to adjust on the 7D
I see that I should think about how the camera is recording light
I try to shoot when the light is "right" but see that I could get results when the light is not so good by for by example using fill flash
(I mainly shoot wildlife)
The midtones adjustment in ACR looks much better than in levels guess theres more information available in the raw to work with
p.1 #13 · Exposing to the right, processing the Raws
Alan321 wrote:
In principle, if you exposed to the right then you probably need to process to the left. Discrepancies arise between how far to the right and how far to the left, in part because in-camera settings such as contrast affect the histogram you see on the camera, and your software has its own different settings and gives you a different result.
There are many ways to achieve this and no one way is best for every image.
Personally, I preferred to reduce exposure and increase brightness rather than use much recovery, but the newest Lr and ACR might change that. Reducing exposure brought the highlights down, as well as mid tones, and increasing brightness restored the mid tones but not the brightest highlights. Recovery did relatively weird things to too many contrast edges.
- Alan
...Show more →
Thanks Alan I always forget that the camera does not show what Ive actually taken but a jpeg
Just tried reducing exposure then adjusting brightness on the butterfly shot
I see whay you mean
Im using CS5 but have noticed that if I have to use too much recovery the image doesnt look "right"
p.1 #14 · Exposing to the right, processing the Raws
In sunny conditions the problem when exposing for the highlights is loss of shadow detail, but that's less of a problem if the shadows in aren't visible. So whenever I'm shooting nature or action shots kept the sun over my shoulder when possible. It's not "flat" lighting because it's not level with the lens axis, it hit at a downward angle creating 3D modeling, but casts the shadows where there is no detail down behind the highlighted front of the content of the scene.
Direct sun actually works better than indirect diffuse light for animals, creating the specular highlights off the flat facets of feathers and hair shafts which give fur and feathers their sparkle. That sparkle, the sharp contrast between specular highlight and surrounding area, is what creates the illusion of 3D shape and texture. With bugs the specular highlights glinting off the hard body parts also are important clues to shape in the 2D rendering.
But the highlight clues only work effectively to create the illusion of 3D when they fall naturally on the top of objects:
They fall in the top hemisphere of objects because the sun travels overhead in an arc. When the light source gets lower the highlights move down on the object clues move down on the object and won't look natural. The light will produce the same 3D modeling, it just won't look "seen by eye" normal.
That's not a problem with natural light, but when you add flash from on camera or the hot shoe, especially in portrait mode, what happens? Because the flash is close to the lens axis the telltale highlighting depicting 3D shape in the 2D photo wind up unnaturally low and make the use of flash obvious. The flash near axis will also overpower and cancel out most of the subtle 3D clues the skylight creates on the shaded side the flash is "filling".
"Fill flash" is a misnomer in most situations because what the flash does is put it's highlight pattern over the top of the skylight which had been modeling the shaded side. Flash also doesn't work effectively to change contrast on a subject half in shade and half in sun.
The shaded side of a object in sun will be 3 stops darker. The camera has a range of with detail of about 6-7 stops, which is why in my test shots of the white towel the shaded side winds up middle gray in the middle of the camera sensor range when exposed for highlight detail.
A 3-stop difference is a 8:1 incident ratio between sunny and shady areas. On a cross lit subject like a human face in the sun if you expose for the highlights on the cheek the shaded eyes look like dark voids in the photo. What happens if flash 1/4 power of the sun (i.e. 2 units) is added?
H:S
8:1 Incident ratio of the direct sun on face
2:2 Flat near axis flash added 1/4 strength of sun
==
10:3 = 3.3: 1 ratio
What happens with fill 1/2 the strength of the sun?
H:S
8:1
4:4
==
12:5 = 2.4:1
Full strength?
H:S
8:1
8:8
===
16:9 = 1.8:1
So the more flash you add to a cross-lit object to "normalize" the shadows the flatter the overall lighting in the foreground becomes. Lighter shadows are also a clue the brain uses to evaluate 3D shape. The same pattern with light and dark shadows will seem "softer" and "harder" respectively. It's not the light that's hard or soft it's how your brain interprets the clues. With on-camera flash over a subject half in sun the foreground winds up looking washed out with highlights placed abnormally low. It looks unnormal — fake and artificial.
The solution to that problem when adding flash? It has two parts:
1) Keep the sun off the front of the object entirely
2) Use two flashes in a key over fill configuration.
Keeping the sun off the front allows the flash to lift the front side independently. Here's a single flash shot from the test series above:
The flat target is poor choice for evaluating 3D modeling but if you look at the towel you'll see the flash while correcting the exposure problem overpowered the skylight and killed the modeling hidden in the shadows of the ambient only shot...
So what I do in that situation is first "fill" evenly with the flash on my camera bracket as in the single shot above, then place my second flash off axis to recreate the frontal skylight modeling the first flash canceled, placing it as close as possible to the natural downward angle of the skylight. The second off axis flash compements the natual lighting vs. trying to completely overpower it...
But the two flashes can only "normalize" (reveal detail everywhere) in the flash lit foreground. The background remains "exposed to the right" and loses detail in the background beyond the range of the flash. The solution for that problem? One is to crop out the background so the lost of detail really isn't noticed — similar to the strategy of putting the sun at me back and shadows out of sight behind things.
The brain evaluates contrast by comparison. That's the same wide shot, but with with the underexposed background cropped out it creates a different, more "normal" impression and it becomes more difficult to tell whether it is naturally or flash lit.
The other solution is to find a light background...
The shot above was taken during a walk along a river, a candid not a planned portrait session. I put her back to the sun to address the contrast problem, raised the flash on a bracket with diffuser to match the angle of the flash to the downward angle of the natural light and put her on the river bank 3 feet below me to get the skylight into her eyes. The background looks "normal" but it's actually the sun glaring off the water underexposed by 2-3 stops. The trick there was eliminating anything in the shot that was lit only by the ambient light and was dark in tone.
The common denominators in all these examples is they are all solutions to work around the limited range of the camera sensor and have the goals of recording a full range of "seen by eye" detail that looks "normal" with highlight and shadow clues which look "natural". The stategies are:
1) When using flash isn't an option keep the sun behind me to avoid the important content of the scene winding up in the shadows. The camera DR will adequately record most of the scene and the loss if detail in small insignificant shadows will not be noticed or seem abnormal because the same loss occurs by eye in wide views. It's only when the eye shifts between shaded foreground and sunny background the pupils of the eye shift the exposure and create the impression of a wider tonal range that the eye can actually record.
2) When I can compose a shot for something interesting in the foreground within 10 ft or so I will face the sun, with front side of subject completely in the shade and illuminate the front with flash, either one flash centered over the camera (which hides most of the shadows the flash creates similar to the frontal sunlight strategy) or two in a key over fill configuration when I want sideways modeling on the foreground.
3) I don't as a rule add flash to cross-lit scenes if I can avoid it because I know it makes the lighting ratio abnormally flat as power is increases. I use that knowledge to control the ratio with flash power relative to the sun. I always have my flash on a bracket so the highlight clues fall more natually on the upper half of the cheekbones and other surfaces.
4) In additon to the lighting strategies I use a variety of PP techniques. In the RAW stage I'll manipulate brightness, fill and blacks in ACR but only to the point where I'm not making the shadow noise obvious (my test examples are exaggerated). Then in Photoshop I use a screen adjustment layer to selectively lighten the shadows in areas where I want the viewer to focus and will sometimes darker the areas I want them to ignore even more...
What I work towards, to the extent possible, is creating an appearance in the photo which matches my impression in person, not just in the wider initial impression (which will lack shadow detail as the eyes adjust to the brighter parts) but what my brain remembers seeing after scanning around and focusing on detail hidden in the shadows. It's that scanning and adjustment of the puplis of the eyes in person which create the impression there is detail everywhere you look in the scene.
p.1 #15 · Exposing to the right, processing the Raws
Very interesting about the having the sun directly behind you
I nornally just try not to shoot facing the sun and often am with the sun at the side
I see now why I get very mixed results when shooting in direct sunlight
Will try the technique of facing the sun and illuminating the front with flash have an idea that it will work really well for butterflies
I will be using just one flash but do have a bracket so will be able to light from higher up
Thanks very much for going to the trouble to explain it so well
Am really glad I posted have learnt a lot today!
Pete
p.1 #16 · Exposing to the right, processing the Raws
Peter, if you want the highlight warnings in the camera to give a better idea of what is likely to be really burnt out in the raw data then set the picture style to neutral for less saturation, and set the contrast to the very minimum. The preview will look crappy but any blinkies are at least very close to true complete burnout.
You can instead go the other way. Set contrast to maximum and you'll get blinkies all over the place, giving you a warning of areas that are getting close to burning out but unfortunately you can't tell them apart from what actually has burnt out. This is more of an early warning approach whereas the low contrast approach is closer to an actual raw data burnout indicator, but of course neither looks good as a preview and more corrections will be needed in software to extract a good image from the raw data.
Another approach is to also use uniWB, a white balance setting that better reflects the actual raw data performance of the sensor but at the expense of looking almost unusable on the previews with a terrible green cast. It allows the jpg histogram to best approximate raw data but it can be very disconcerting when you try to review your images.
None of these approaches are good if you actually want to capture jpg files but they are all better than standard histograms and blinkies for assessing how good your raw data capture has been, at least in terms of exposure.
p.1 #18 · Exposing to the right, processing the Raws
LCPete wrote:
Thanks very much everyone brilliant
I will go and have a go converting some raws and reply properly later
One thing is that many of default profiles are twisted so as you shift exposure they can also shift color tints. If you may your own color profile for the camera it won't be twisted.
p.1 #19 · Exposing to the right, processing the Raws
Alan321 wrote:
Peter, if you want the highlight warnings in the camera to give a better idea of what is likely to be really burnt out in the raw data then set the picture style to neutral for less saturation, and set the contrast to the very minimum. The preview will look crappy but any blinkies are at least very close to true complete burnout.
You can instead go the other way. Set contrast to maximum and you'll get blinkies all over the place, giving you a warning of areas that are getting close to burning out but unfortunately you can't tell them apart from what actually has burnt out. This is more of an early warning approach whereas the low contrast approach is closer to an actual raw data burnout indicator, but of course neither looks good as a preview and more corrections will be needed in software to extract a good image from the raw data.
Another approach is to also use uniWB, a white balance setting that better reflects the actual raw data performance of the sensor but at the expense of looking almost unusable on the previews with a terrible green cast. It allows the jpg histogram to best approximate raw data but it can be very disconcerting when you try to review your images.
None of these approaches are good if you actually want to capture jpg files but they are all better than standard histograms and blinkies for assessing how good your raw data capture has been, at least in terms of exposure.
also set the jpg to adobergb instead of srgb to expland the gamut a bit so it won't blink intense red and green as quickly
May 18, 2012 at 02:13 AM
Guest
Guest
p.1 #20 · Exposing to the right, processing the Raws
LCPete wrote:
I only shoot in raw anyway
Thanks
Pete
this does not matter either. its still applies. the image you see and the data from the historgam is derived from your jpeg settings. it is not a raw data histogram.
