Home · Register · Search · View Winners · Software · Hosting · Software · Join Upload & Sell

Moderated by: Fred Miranda
Username   Password

  New fredmiranda.com Mobile Site
  

FM Forums | Lighting & Studio Techniques | Join Upload & Sell

  

Archive 2013 · Fill, Ratios and Sensor Range
  
 
cgardner
Offline
• • • • •
Upload & Sell: Off
p.1 #1 · Fill, Ratios and Sensor Range


The traditional approach to portrait lighting ratios such as 3:1 is focused primarily on the appearance of the shadows on the face. As the ratio changes from 2:1 to 3:1 and greater exposure is typically adjusted to keep the highlights the same and the change seen in the photos is the shadows getting darker.

Note: edited to correct typo....

It's not apparent but the ratio convention used for portraits isn't f/stops or incident differences. A 2:1 ratio is means 2x more light (1/stop brighter) light is reflecting off the face in the HIGHLIGHTS for every 1 unit reflecting from the SHADOWS. The lowest integer ratio under the convention is a 2:1 ratio created by overlapping equal incident strength off axis "key" over centered fill:

H:S
1:1 Fill even on entire face
1:0 Key light overlaps creating highlights
===
2:1 Highlights reflect 2x more light.

There's no rule that says fill must be centered, but look what happens with equal key and fill are placed on opposite sides:

H:S
1:0 Fill on left
0:1 Key on right
==
1:1 The net effect is mostly flat lighting with unfilled shadows where neither light reaches. It's the same incident ratio where the light hit but the direction of the fill changes the results.

It is also possible to get a 2:1 ratio like this with crossed key and fill:
H:S
2:0 Key
0:1 Fill
==
2:1 But the results will look different than centered fill because the fill falls off sideways vs front>back on the face and the ratio will be different and lighter on the side closer to the fill source and darker on the side opposite. Also where key and fill shadow cross each other or don't reach there is no fill and very dark shadows.


YMMV, but I prefer the appearance of centered chin level fill for portraits because it creates very few visible shadows on a face eliminating unfilled voids. While it might seem the fill light is "flat" because it creates no sideways modeling clues the front>back fall off on a face creates an exposure gradient in the shadows which makes the closer nose shadow lighter in tone than sides of the face and ears.

Making the shadows the constant in the highlight:1 notation may be because the very first ratio meters consisted of a card with two holes held in front of one eye to judge the ratio. One hole was clear and held over the shadow side. The other hole had gradient of neutral density on a strip or wheel. The ND filters where moved until the appearance of both sides matched. The density reading of the ND filter reveals how much brighter the highlight side is by comparison: .30 = 2x, .45 = 3x, .60 = 4x, etc. So with that primitive meter if you wanted a 3:1 ratio you'd move a .45 filter over the highlight side and adjust key light until the two sides matched. It's still a very accurate way to judge ratios visually.

A digital sensor has a fixed range that varies with sensor size and other design factors. You can determine your camera range with a gray card and fast lens. Open the lens as wide as possible and adjust shutter slower until the gray card reproduced is just below clipping with eyedropper = 250. Then close down the lens (and adjust shutter faster in full stop increments) until the card is reproduced at around 30 units, the threshold where detail above 0 black can be seen in an image. The difference in stops it takes to reproduce the card 250 and 30 eye dropper units is the practical range of the camera and typically between 6 and 8 f/stops of visible highlight:shadow detail.

If your FF camera has a range of 8 stop and my crop camera has a range of 6 stops will the same lighting ratio produce the same results? No.

I first realized that myself when ugrading cameras and seeing the shadow detail change when I exposed the highlights the same using the same ratios. I was setting ratios via identical power on identical key and fill flashes via distance so only variable that could explain the change in shadow detail I saw was the difference in range of the cameras. It made me realize the ratio I really needed as a baseline for "normal" looking results in photos wasn't 3:1 or 2:1 but the one which would reproduce a full range of tone in the subject on the camera sensor range I was using.







Something I realized since the early days learning the Zone System and making halftones in the National Geographic photo lab is that when the shadows and highlights are reproduced accurately, per what you'd expect to see by eye in the same context, all the tones in between the ends of a gray scale line up like ducks in a row in a linear progression. That's to say if the ducks ranged in even .30 reflective densities from black > grays > white by eye they'd wind up looking similarly linear in the photo, automatically. That's because the process on film/print paper and digital is engineered to work that way.

What is the lighting ratio in that shot above numerically? Between 2:1 where key= fill incident strength and 3:1 where key is a stop higher than fill if measured with an incident meter.

What happens if more or less fill is used from that "fit the sensor" ratio baseline? Since I only had lighting that matched the sensor range for that shot I've simulated the over / under fill with Photoshop.

If fill was reduced to make the shadows on the face darker and the key intensity raised so the highlights in the white shirt are exposed the same at the same f/stop the overall scene range will exceed the sensor's and both detail in the black suit and separation with the background would be lost...







If the fill was increase more and key reduced to keep the highlights exposed the same in the white shirt? The shadows which were reproduced as seen by eye (black) would begin to look washed out and the overall image tonal range flat:







Conclusion? Only one ratio will fit the full range to the sensor. But that creates a problem. What if I want the shadows on the face to be darker than the "fit the range" sensor response reproduces them?

Since any overall reduction of fill would cause a loss of detail in the shadow the options at capture would be to adjust the fill so it darker on the face than on the suit. That could be done by "feathering" the fill source, aiming away from the face.

But what I find easier in the age of Photoshop is to record the subject range to fit the sensor in the camera then modify the mid-tone values in post processing. That can be done many different ways in ACR at the RAW stager or in Photoshop after the RAW adjustments are made. My preference is to use adjustment layers in Photoshop as shown below:







I'll typically start with a slightly overfilled capture on the RAW, especially with black content because when dark shadow in the subject are recorded as seen on the threshold of the sensor's range they are noisy. Overfilling at capture eliminates much of the noise from capture. When I darken the over filled areas with the Multiply layer by opening the mask I can get the same black tone, but with more detail and less noise. I can also apply the changes selectively to do things like put more detail in the shadow of the book (a gift from grandma) or more separation in the lapels than would have been in a SOOC shot exposed with "normal" seen by eye shadow detail.

it's a case of an old dog learning a new trick as the tools evolve. But in concept its no different that what I did years ago with the zone system. Negatives where adjusted to scene range so a straight no adjustments ZS print I made would have a full tonal range and detail in all the shadows that weren't void areas, then I'd burn and dodge where I wanted to reveal or de-emphasize detail when making the print.

Once I realized how easy it was to just set lights based on shadow and highlight detail make adjustments to the ratio on the face mid-tone in Photoshop I changed the way I set-up studio from metering ratios by number to simply adjusting fill for detail on a draped black towel and key and rim light for detail and rim/key ambient on using a white one to record a full range of detail as my starting baseline "negative" as with the zone system. That produces a "normal" looking result SOOC works fine for most subjects because I don't as a rule do moody character studies and the like using very dark shadows on faces.

Nowadays my incident meter mostly sits on the shelf next to the film cameras, CD player, cassette recorder, and other really nice but functionally obsolete gear.



Edited on Jan 18, 2013 at 04:04 PM · View previous versions



Jan 17, 2013 at 08:23 PM
BrianO
Offline
• • • • •
Upload & Sell: On
p.1 #2 · Fill, Ratios and Sensor Range


cgardner wrote:
...A 2:1 ratio is means 2x more light (1/stop brighter) light is reflecting off the face in the shadows for every 1 unit reflecting from the highlight.


That's exactly backwards from how everyone else that I know of would describe ratios.



Jan 18, 2013 at 05:31 AM
RustyBug
Offline
• • • • • •
Upload & Sell: On
p.1 #3 · Fill, Ratios and Sensor Range


Kinda hard for a shadow to be twice as bright as the highlight.

BrianO wrote:

+1 @ words might have gotten a bit tangled

That's exactly backwards from how everyone else that I know of would describe ratios.
Including Chuck

cgardner wrote:
H:S
1:1 Fill even on entire face
1:0 Key light overlaps creating highlights
===
2:1 Highlights reflect 2x more light.







Jan 18, 2013 at 11:07 AM
cgardner
Offline
• • • • •
Upload & Sell: Off
p.1 #4 · Fill, Ratios and Sensor Range


BrianO wrote:
That's exactly backwards from how everyone else that I know of would describe ratios.


Opps... That was a typo / brain fart... Thanks for the catch. It's been fixed.

A kinder person would have simply said "I think there's a typo here.... "



Jan 18, 2013 at 04:02 PM
thumphrey
Offline

Upload & Sell: Off
p.1 #5 · Fill, Ratios and Sensor Range


Good info! Thanks



Jan 19, 2013 at 06:10 PM
 

Search in Used Dept. 



erichard
Offline
• • •
Upload & Sell: Off
p.1 #6 · Fill, Ratios and Sensor Range


I have a hard time following it all. Maybe a short summary would help.

On the photos, the first one looks best. The last one is in my opinion not as great.

To separate the black suit from the background, why not pull him further away from the background and use grids, etc. to focus the light on him and not the background (I assume you did use grids). From there, just adjust the light such that highlights are not blown, and get the fill ratios you like with modeling lights or experimentation. The background would be relatively black, even with a fair degree of fill due to inverse square law.

So far as I know, the main sensor limitation is blown highlights. Sure, some have more bit depth and therefore more levels of gray, etc., but the universal limitation is blown highlights. Max out the highlights to be just sub clipping, and the world is your oyster. From that point, you can manage the file best, blacks included. Curves might be the best way to do it, though lightroom can basically do curves with sliders now.

The problem with dodging and burning in this situation is that without masking, which would be problematic here, you will be spilling over into the background with your brushing.

To me, the last photo looks washed out somewhat (not as much as the one before it, but more so than the first one).



Jan 20, 2013 at 08:52 PM
cgardner
Offline
• • • • •
Upload & Sell: Off
p.1 #7 · Fill, Ratios and Sensor Range


erichard wrote:
I have a hard time following it all. Maybe a short summary would help.


Only one key:fill ratio (scene brightness range) will match sensor range (first photo) and create full range of detail in content of camera file (black suit / white shirt)..

The next two photos show what happens when key:fill ratio does not match sensor range and exposure is set for the same highlight detail:

Too much fill = shadow look grayer than normal
Too little fill = shadows loose detail and are noisy

The last photo shows how I avoid noise (rainbow speckling) in shadows on black objects. I find it's better to err on the side of over filling (grayer than normal shadows) at capture to minimize the noise then burn the back down in post processing,





Jan 21, 2013 at 04:23 AM
erichard
Offline
• • •
Upload & Sell: Off
p.1 #8 · Fill, Ratios and Sensor Range


Thanks. That is much more clear in my mind. I can see the logic of your idea. I'm not sure it's easy to execute it well, though, with respect to the post processing. Without a good mask of the subject, the dodging and burning spills too much into the background IMHO. You could do a mask, but it is time consuming and somewhat difficult with black on black.

I find that if you push the histogram to the right in your original capture, just sub clipping for relevant parts, the photo often is a bit bright at first blush, hot off the press. From that point, it is often just a matter of pulling back what they are now calling "exposure" in LR, while still keeping the whitest white at the far right (and blackest black at the far left). (In photoshop, "levels" works well to lock the far extremes of white and black where the histogram begins and ends.) Pulling back exposure, or what used to be called brightness I believe, usually tends to settle the blacks down where they need to be if you're aiming for what your eye sees, as you say. Not in all cases of course, but it does basically what you are attempting to do in avoiding noise while still fulfilling the zone system aims.



Jan 21, 2013 at 04:46 PM
cgardner
Offline
• • • • •
Upload & Sell: Off
p.1 #9 · Fill, Ratios and Sensor Range


erichard -

The layers adjustment example I posted is an extreme case and also bit of an exaggeration of how light in tone I make it with not a lot of careful attention to the masking. I actually didn't have a noise problem with that shot, it was just one few black background / black suit session I had to use as an example. Noisy shadows are more of a problem with higher ISO ambient and speedlight shots. Sorry for any confusion that created.

As for the masks? In practice I find it's pretty easy to adjust masks in Photoshop by hand if I first enlarge the monitor view to 400% or greater when working on border areas. I use a Kensington Orbit trackball that adjusts it's movement with acceleration and when it's moved slowly it is very precise.

Recording a full range with key:fill flash indoors is easy. More difficult are outdoor scenes in back- and cross-light. I cover those situations in my web site tutorials: http://photo.nova.org

Outdoors I expose sunny highlights on white clothing 1/3 stop below where it clips in the warning. I adjust the "key" flash on the foreground 1/3 stop under the sunny highlights (2/3 stop below clipping warning) and the "fill" flash on foreground for detail in the darkest shadows. That way the most important content in the foreground has a "normal" full range of detail preserving the ambience of the backlight without any clipping and loss of detail on the subject. Where there will be a loss of detail is in the shadows in the backlit background. They will be dark and noisy. The best way to avoid the noise in that situation is don't select very dark backgrounds that will generate noise when possible.



Jan 21, 2013 at 10:58 PM
erichard
Offline
• • •
Upload & Sell: Off
p.1 #10 · Fill, Ratios and Sensor Range


Very good. Thanks for taking the time to respond.


Jan 22, 2013 at 01:36 AM





FM Forums | Lighting & Studio Techniques | Join Upload & Sell

    
 

You are not logged in. Login or Register

Username   Password    Retrive password