cgardner
Offline
• • • • •
Upload & Sell: Off
|
ukphotographer wrote:
Because he wasn't a Luddite.....
In 1801 Joseph Marie Jacquard invented an automated loom which used punched card connected together is a belt to automatically. The Luddites were English textile artisans who protested against this automation often by destroying the mechanized looms.
The term "Luddite" became associated with anyone opposed to increased industrialization or new technology—like TTL control of flash. When I said "But in my view Hobby was a Luddite in his slavish devotion to old manual flashes exclusively." it was because he preached against TTL as if it were a false religion.
Back when his name first was mentioned I Googled him and came across an audio interview with Hobby a Strobist fanboy had posted. In it Hobby kept repeating the work "ethical" to describe his a approach. Both he and the interview disparaged TTL metering, but from the ways they described it's many shortcomings it was obvious they hadn't actually used it, at least not enough to master it's use. In particular being Nikon users they weren't familiar with how the Canon system actually works, making erroneous references such as the AF assist lamp being the source of the woefully inadequate IR signaling.
I give Hobby credit where I think credit is due, but the main reason I became an advocate of the Canon system was to get the type of Canon owner I described — a hobbyist with a single 430ex in the hot shot wondering what to do next — "deprogrammed" from Hobby's brainwashing that manual and radio trigger was the ONLY way they could successfully move their flash off camera. I did that by creating tutorials they could read and consider before they threw their money at a problem they didn't understand based on the advice of someone opposed to the use of TTL for "ethical" reasons.
The fact Hobby used a camera with a higher sync speed than is normal today was the logical choice if one chooses also to only use manual flash and promote it as the only "ethical" way to do lighting. But most of the people following his advice weren't using camera's with 1/500th sync speeds and many new to flash might not even understand the difference unless it was explained to them.
If you take the trouble to read the my first Canon tutorial you will find it is an overview of the system that compares various approaches. Using the ST-E2 as a Master, using a 580ex as a Master, using manual radio triggers, and even forgetting about speedlights and purchasing studio lights instead.
Back before Radio Poppers going "Stobist" vs. using the Canon system (or Nikon's) meant taking the $300 430ex flash the hobbyist already at that point, spending $292.46 for a set of Pocket Wizards (what I paid for mine in 2005) and winding up with a single flash solution that moved the flash off axis, but without any fill.
http://super.nova.org/EDITS/BH.jpg
Going further down the Strobist path and adding an eBay SB-28 or Vivitar 285HV for fill would require another set of PW would cost another $300 for the radios plus whatever the flashes cost (e.g. $75). So all in for a two flash lighting solution they would have spent:
$300 430ex
$75 used manual flash
$600 two sets of Pocket Wizards
$100 two stands @ $50 each.
$40 two umbrella holders @ $20 ea.
$40 two umbrellas at @ $20 ea.
====
$1,115
At the time I bought mine 580ex flashes cost $380, so following the approach I used and suggested — putting the Master on a Stroboframe camera flip bracket (which still only costs $50) — they would spend:
Single flash solution:
$380 580ex
$50 Off Camera TTL cord
$50 Stroboframe bracket
$5 DIY diffuser
====
$485
Adding second flash off camera:
$380 580ex
$50 One stand
$20 One umbrella holder
$5 DIY diffuser
=====
$455
Grand total for two flash Canon 580ex based solution: $940
Apart from the cost I explain in my tutorial the difference in logistics of wrangling two stands and needing to meter lights vs. roaming freely with a flash bracket and changing ratio and adjusting exposure with their index finger on the back of the Master flash above the camera.
I explain that in any static situation manual control as Hobby suggested is better than TTL: have a tutorial explaining how to set Manual manual mode with the Canon M mode and ratios by distance without metering as I did for years with my Vivitars, a necessity with the Canon system because the pre-flashes affect a hand held meter reading. But I also explain the advantages of ETTL-II camera metering controlled flash which are the ability to adapt automatically to moving targets and control ratio and exposure from the Master flash (and now the camera).
Addressing William's comment about using TTL at weddings.....
The manual flash system Zucker taught thousands of wedding shooters over the years was to use a ratio that perfectly fit the range of black suit > white dress to the print range to produce a natural looking result in every photo taken vs. the flat "deer in the headlights" look a single flash on camera produces.
The lighting ratio that fit foreground to print was a 3:1 where key light which is 2x brighter than fill overlaps it. (2:1:1 = 3:1). Zucker made used two identical flashes and made the key light 2x bright by moving it closer to the subject than the fill on camera. Monte was all about KISS and I can't recall the words "Inverse-Square Law" passing his lips, but he understood how to apply it. If shooting at 11ft moving the identical slave 8ft from the tip of the nose at a 45° angle made the key light 2x brighter and created short lighting on the the faces with normal "seen by eye" contrast. That simple two-flash technique produced prints with a full range of tone and detail from the laces on the groom's shoes to the subtle difference between specular and solid whites on the lace and bead work of the brides dress. When moving closer to 8ft for a tighter crop you'd move the key light to 6ft to maintain the ratio and close the aperture one f/stop to keep the exposure the same.
I still use the same technique with my Canon flash because it is a no-brainer and requires no metering. This Bridge screen shot is from a session where I arrived and shot portraits in M mode with my speedlight at a meeting of our church "Lens Team":
http://super.nova.org/TP/MAGLens.jpg
The first two frames are a shot to set Custom WB off the card and a verification shot. The third frame should be familiar to regulars. Before taking it I had set both 580ex flashes to M at 1/2 power and the camera at f/8. Using my arm span (5-1/2 feet) between nose and center of my DIY diffuser I set the distance for the slave. Then I stepped back to 8ft, the distance I find flattering for H&S shots, focused and fired.
http://super.nova.org/TP/TowelGary.jpg
The reason I use that shot so much is because it illustrates what happens when a lighting ratio creates a scene range which EXACTLY matches the sensor range.
By coincidence the same 3:1 ratio fit scene to color print also does that on my digital camera. How did I discover that? By looking at the RAW images and the histogram in files which are rendered accurately in the highlights with solid white objects below clipping.
Being a technician and technical manager most of my working career I understand how the photographic process is engineered to work. If lighting ratio is controlled so there is detail the shadows and the highlights at the same time the DlogE response is linear. What linearity means in practical terms is that everything the camera records between solid black and specular white in the scene winds up looking "normal" per our seen-in-person perceptual baseline.
The histogram in the the camera playback represents the sensor range. It does not tell you directly what that range is, or what the range of the scene is, but if you know how to use and interpret it, the histogram will reveal whether or not the scene exceeds sensor range. That is determined by first exposing to the "right" placing the brightest white content just under clipping and then looking at the left side of the histogram. If the left side is piled up high the scene exceeds the sensor. If it it isn't the scene fits the sensor.
When a scene doesn't fit the sensor, which is most of the time outdoors the choices are:
1) Expose for highlights / lose shadows
2) Expose for faces/midtones clip highlights / lose shadow detail
3) Expose for shadow detail and clip face and other highlights
4) Find ways to change the range of the scene"
a) flat light vs cross light
b) use reflectors
c) use flash, preferably two
Indoors where the sensor doesn't need to cope with the contrast of sunlight fitting scene to sensor with two manual flashes can be done very easily by starting with a foundation of enough even shadowless fill to record detail in the shadows on the sensor then overlap key light until the solid white are just below clipping. When you get both ends accurately reproducing the scene the middle will look the same as it does by eye in average light.
Regardless of whether the lights are set by hand metering or using the playback/histogram/clipping warning the screen and print images will only have a full range of tone if the ratio if the scene matches the sensor range and the only way to judge that is look at the results both when shooting and on the computer. Once you test and find the ratio that gets both ends right based in the RAW in the computer you can look at the same file using the playback, histogram, and clipping warning on the camera, slap yourself on the forehead and say, "So that's what a optimally exposed file looks like!"
The advantage of Zucker's identical flash / distance based manual approach I use is that it requires no metering. That's what allowed use to shoot on the run at wedding receptions with two manual flashes and get perfect exposure. If I were shooting weddings today I would use that same method today because it is more predictable than ETTL. But I would still want the option to use ETTL for other situations where there isn't a huge white dress in most of the photos.
Always shooting from 11ft with the key flash at 8ft for a 3:1 ratio might seem to cramp one's style creatively, but with a constant aperture zoom lens like a 24-70mm 2.8 or 24-105 you could shoot just about every shot at a wedding that way and exposure of every shot would be exactly the same.
Because we shot with a fixed FL lens we shot from 16, 11, 8 and 6 feet placing the slave at 11, 8, 6 and 4 feet respectively to maintain the same 3:1 ratio. Those distances were selected because they are the same as full f/stops and easy to remember. The way the inverse-square law works moving camera and key light from 16ft / 11ft closer to 11ft/8th doubled the intensity so as we moved closer we closed the lens a stop to keep exposure the same. It sounds more complicate than it is. After a few times you find the spot, move the light and change the aperture as instinctively as focusing the camera, which was also done manually back then.
As mentioned previously color film had enough latitude that the lab had no problem adjusting for various caused by the ceiling bouncing more or less light or the distances of the lights not being perfect. The main thing to avoid was underexposed shadows you just erred on the side of overexposing.
That's where digital is different. There is no margin for error in exposure. The lighting ratio must match the scene to sensor's range and the highlights must be correctly exposed. So as with any other method of setting exposure with digital you are foolish of you didn't chimp, check for clipping as you shoot, and to correct as necessary.
The thing to take away here is that is you do know the lighting ratio that will match the scene to your camera's DR and use it, then all you need to do is use it and keep the highlights just below clipping to record everything with the same detail you would see by eye if your goal is to record a full range of tone at capture.
What makes it work is using fill and key to control the range. What made it practical at weddings logistically was using the bracket and only one stand and shooting systematically with pre-tested distance - aperture combinations. As for how real world that is? I shot weddings every weekend for two years (the longest two years of my life) and have used that technique ever since successfully.
Fitting scene to print isn't a new thing. That's what made a B&W print look real the difference was that flash wasn't needed to do it outdoors. #2 B&W print paper had a 10 stop density range. That meant to make a print with black shadows and paper white specular highlights the negative, from clear base to highlight needed a density range of 10 x .30 or 3.0. With B&W film and the zone system would measure scene with a spot meter, determine the range, then adjust development so the highlights of the scene had a 3.0 density. The same scene on clear and overcast days would have different f/stop ranges and require different development times create the 3.0 highlights needed for the full range.
Why do I mention this? As background for understanding why flash became necessary outdoors when color film arrived on the scene and started to be used. As late as the 70s some studios were still using B&W be cause color prints were not stable and faded in a few months if exposed to light. But the photographers started using color they discover the range of the prints was much shorter and negative development couldn't be used as with B&W to fit scene to sensor. The scene range needed to be manipulate with flash. Flash had been used outdoors with B&W but color film make using it necessary all the time.
The first type of photography to move to color was movies and the shorter range was one of the factors that moved movie production onto sound stages with elaborately constructed sets to shoot a horse opera rather than just shooting it outdoors because of the need for fill light to handle the contrast. Unlike stills movies need continuos light and a lot of it to balance the sun. It was easier to shoot in the studio and simulate the sun with rim light with a potted plant in front of it.
Color film forced still photographers to make Solomon-like decisions with every outdoor shot. Expose for the highlights and lose the shadows, expose for the shadow detail as they had with B&W or cut the baby in half, get the faces looking "normal" but blowing out the sky and losing the shadows.
The direction of the light became a more critical factor. For example here is a digital shot exposed for the white detail facing west at 11AM....
http://super.nova.org/TP/DR_FlatLight.jpg
Here's the same exposure on the same subject, also exposed to keep the highlights below clipping but turned 180° facing east into the sun in the SE part of the sky...
http://super.nova.org/TP/DR_Backlight.jpg
The DR of the sensor is exactly the same for both so is the exposure criteria (highlight detail) so why does the first look "normal" and the second underexposed?
The difference is perceptional. What seems normal depends on what the brain thinks is most important. You know what a white towel and gray card look like by eye and the first one is a closer match so by comparison it looks "normal".
Without flash what would you do in the second situation? Most would adjust exposure until the foreground in the shade looked normal the way the eyes do when they focus on a backlit face or object. The brain ignores the fact the eyes are blowing out the highlights and shadows because it remembers seeing the overall scene as normal, more like the first shot.
That was an ETTL-II HSS test and I was shooting a f/2.8 in Av mode to see how what camera metering at the default EC=0 and FEC=0 would do and what was needed to correct it to get a full range of tone on the target in the foreground. To get the second backlit shot below clipping in the towel I had to dial in - 2 EC. The EC=0 shot was lighter and more normal looking overall, but my goal in the test was not to blow the sunny highlights with the ambient.
What is interesting to grasp in that test is that the TTL metering, left to find the best ambient only exposure for that scene on it own, would expose as I would without flash manually based on what I saw in the playback making faces and other midtones look "normal" despite knowing it would blow the highlights.
This is what I got when I reached up and turned on the 580ex, used direct, in HSS mode at FEC= 0:
http://super.nova.org/TP/DR_FlashFill.jpg
The background is as underexposed as in the second ambient only shot, but the foreground just in range of the HSS mode flash 10ft away when measured separately with Levels has a fit range to sensor histogram...
http://super.nova.org/TP/DR_FillHisto.jpg
While the test was done in HSS mode to test range, the more important knowledge I gained from that systematic test from the baseline of ambient only shots was how the ETTl-II metering that controls the flash works, and how to make it work predictable. The key factor it turned out was what I had been doing all along because I don't like blown highlights in my photos: start by controlling ambient and keep the highlights under clipping. The second key factor is something I also always do when shooting with flash KEEP THE SUN OFF THE FRONT SIDE THE FLASH HITS.
Both are common sense if you think about it. If you start with ambient clipping highlights — what TTL metering will do by default — it will look better before flash is added but after you add the flash to lift the shadows a bit more you'll wind up with a normal face and blown highlights where the sun hits. To get a normal looking face on the shaded side and not blow the highlights where the sun hits you need to start with the highlights below clipping. Then you need to avoid clipping them with flash by overlapping flash and sun by keeping sun off the front.
If you use the sun as "key" light to model the face then try to add flash to get rid of the dark shadows on the nose and eye sockets you can't add flash to the shadows without also adding it to the highlights. With normal non HSS flash you quickly hit the sync limit (e.g. 1/250th). If you start at f/11 @ 1/250th and have the cheek exposed perfectly with just the ambient, then start adding flash over the highlights you must start closing down the aperture more to prevent clipping.
Here's what happens numerically. Without flash the incident ratio of sunny 16 / shady 5.6 on cheek and nose shadow is 3 stops or 8x greater in the highlights 8:1
H:S
8:1 cheek:shaded eye sockets
What happens to the ratio when you add flash 1/4 the intensity of the sun (i.e. 2 units)?
H:S
8:1 cheek:shaded eye sockets (ambient)
2:2 fill flash 1/2 the intensity of the sun
==
10:3 = 3.3:1
Adding flash is 1/2 the strength of the sun (i.e. 4 units) it hits both cheeks and shadows equally...
H:S
8:1 cheek:shaded eye sockets (ambient)
4:4 fill flash 1/2 the intensity of the sun
==
12:5 = 2.4 : 1
How lets stay you raise flash equal to the sun
H:S
8:1 cheek:shaded eye sockets (ambient)
8:8 fill flash 1/2 the intensity of the sun
==
16:9 1.7:1
or overpower it by 1 stop (i.e. 16 units)
H:S
8:1 cheek:shaded eye sockets (ambient)
16:16 fill flash 1/2 the intensity of the sun
==
24:17 = 1.4:1
The more flash power you add with a single fill flash the more it overpowers the ambient and lowers the net lighting ratio. By the time you add enough flash into the eye sockets and nose shadow to make them look "normal" the over look of the face is flat because most of the natural modeling has been cancelled. But because the eye sockets starting shaded and darker than the cheeks will still always be darker than the cheeks and forehead no matter how much flash is added if it hits cheeks and eyes equally. The eyes only look better and more normal because the flash adds the sparkle of catchlights.
That type of flat lighting became the "new normal" for flash assisted color photography because the cause and effect wasn't clearly understood. I didn't understand it myself until I started experimenting and thinking about the results I was seeing and what was causing them. That in turn lead me towards techniques with solve the problem
The solution? Find ways to fit the scene to the sensor on the flash lit foreground AND make the results look natural. The clues how to do that are found back in the studio.
Putting subject's back to the sun outdoors is exactly the same as using a hair light in the studio. Set it just below clipping where it hits, then on the front of the face add even fill until the shadow detail in dark clothing (groom's suit) is revealed, then overlap the key light until the white highlights (shirt collar / bride's dress) are just below clipping....
http://super.nova.org/TP/WhiteBGTowelCard1ZS.jpg
Outdoors in backlight if you do the same thing, back to the sun, keeping the sunny highlights below clipping you get this...
http://super.nova.org/TP/HSS/_MG_5034_Zones.jpg
Add one flash and it will adjust the exposure so it looks normal but kill the modeling of the natural light...
http://super.nova.org/TP/HSS/_MG_5026.jpg
What is missing? The natural modeling because the flash cancels the modeling of the skylight.
How to recreate the modeling? The same way as indoors with an off axis key light.
First pose the face into the skylight as if it were your "key" light. Skylight isn't "Flat" it is always downward and depending where the sun is in the sky in back one part of the sky can be brighter than the other. Clouds and haze will affect this so you just need to move the face around and observe to see the difference. Watch the both sides of the nose and eyes for the shadows and where they fall.
Then simply your second flash as key light at the same angle the skylight is hitting the face. For butterfly / full face aim nose directly into the natural light center the flash. For oblique / short light turn face 45° to the light and shoot into the shaded side, putting the flash at the same 45° angle relative to the nose...
http://super.nova.org/TP/HSS/_MG_5035_Cropped.jpg
The flash assisted lighting winds up looking "normal" because the key and fill flash allow the scene to match the sensor in on the flash lit foreground. The lighting winds up looking more natural than a single flash shot because natural light and flash key light hit the face at the same angle. The flash is complementing the natural lighting, not fighting it.
That's a different approach than "How can I overpower the sun with my flash" and one that starts with knowing how to pose a face to the natural light in a flattering way. That's one of the reasons I also suggest photographer experiment and learn with a window and reflector first before throwing money at a problem they don't understand.
As for ETTL vs. Manual?
Regardless of whether you choose to set ratios them with a meter and manual flash or ETTL ratios only one will match the black suit and white dress to the sensor. Just as with manual ratio with ETTL you'll need to test and find which one does that also.
When looking at this shot with a 3:1 ratio set by distance it looks unremarkable and uncreative but accurately captures what he normally looks like because it fits scene to sensor.
http://super.nova.org/TP/TowelGary.jpg
If shooting the same photo with ETTL A:B ratios I would have used A:B =1:2 Key light 2x stronger
H:S
1:1 A group is fill from bracket even on entire scene
2:0 B group is key light creating highlights over fill
==
3:1 Reflected ratio is the same as when set manually with key 2x brighter than fill.
The day after I got my flashes I did this test series to see what the ETTL ratios actually looked like.
http://super.nova.org/TP/CanonRatioTest.jpg
I find in most situations A:B = 1:2 fits scene to sensor. So when I shoot with ETTL I start with 1:2 and adjust FEC until highlights are correct. That fits scene to sensor, my definition of technically correct exposure because it uses the full DR of the camera.
How does these two shots look different in terms of "hard" and "soft" light and why?
http://super.nova.org/TP/TowelGaryLighter.jpg
http://super.nova.org/TP/TowelGary.jpg
It is the same file with a middle slider tweek in Levels. The DR captured by the camera didn't change and the extreme highlight and shadows didn't change. All that changed were the middle tones. Moving the middle slider makes them lighter and the perceptional impression is that the lighting is "softer" than the 3:1 baseline.
I learned that happens nearly 40 years ago. After working for Zucker learning to light faces in photos I spend the next four years working in the photo labs at National Geographic reproducing photos and adjusting them for optimal reproduction when printed in much the same way — manipulating the mid tones to change the brain's perception of what is there.
Perception is of contrast in a lighting pattern is also affected by context. The same photo will seem different perceptually depending the background surrounding it because brain calibrates perception to the overall field of view.
http://super.nova.org/TP/TowelGaryWhiteMat.jpg
http://super.nova.org/TP/TowelGaryBlackMat.jpg
But something else I came to realize by shooting mostly candids, not studio work in the same room all the time, is that a 3:1 ratio in big room will not look the same as a 3:1 in a small room where the key and fill are bouncing off ceiling and walls more.
What difference does the room make? With same incident readings there will be more fill in a small reflective room because the light not hitting the subject directly will bounce creating a "wrap around" fill effect not seen in the larger space or one covered with black drapes. When big modifiers are used in small spaces most of the "wrap" comes from: the 50% of the light that is hitting everywhere other than the subject and bouncing back in all directions. In a small space like my basement studio I can fill it with even light by using one bare bulb flash. Below did that with a rim light to show a furniture maker how to light it for a white background catalog with only the two Alien Bees he owned:
http://super.nova.org/TP/table/Setup.jpg
http://super.nova.org/TP/table/Table04.jpg
I didn't meter the lights because the incident ratio wouldn't predict results. I adjusted the bare bulb flash based on shadow detail and the back rim light based on highlight clipping.
The more I experimented with both my speed lights and studio lights the more came to realize that setting ratios with an incident meter while an effective way to precisely control the intensity of the lights didn't predict how the resulting ratio would look like perceptually on white and dark backgrounds, or with different amounts of bouncing, wrapping spill fill. For example a 3:1 ratio, set by meter looks "normal" for a face on a dark background, but that same ratio looks too dark and heavy on a white one.
My approach changed from "by the numbers" via metered ratios to starting from a baseline of first fitting range to sensor, looking at that, then adjusting perceptually based on the mood I wanted the lighting to project in the image. The camera playback doesn't accurately depict what is in the RAW file, but by systematically testing and comparing one with the other I trained my brain to understand what was happening in the RAW by interpreting the playback, histogram, and clipping warning.
Digital gave me new tools for measuring and controlling the final results I wanted. I found the best ways to use them for my workflow and in the process concluded the old tool — incident meter — wasn't really necessary in my workflow. When I did meter I would still wind up adjusting perceptually due to all the variables the meter can't measure anyway so the metering was redundant.
I use the same baseline method with ETTL ratios. I start with A:B = 1:2 most of the time because I know it matches scene to sensor. Then from the baseline of that "normal" looking full tonal range I decide whether to deviate from "normal" by changing the A:B ratio. But making face shadows darker globally with an A:B = 1:4 ratio with less fill will result in loss of detail in dark clothing. Making shadows lighter globally with more fill using an A:B = 1:1 will make the shadows gray affecting perception of the overall tonal range.
What Photoshop allows me to do is shoot "normal" for a full tonal then selectively adjust tones in the middle in the same way I dodged and burned prints in the darkroom. But instead of using the dodge and burn tools in PS I use adjustment layers as shown in this tutorial:
http://photo.nova.org/AdjustmentLayers/ and other method locally in specific areas of the photo, not just globally with the tools like Levels or Curves.
Using that approach I find ETTL ratios more convenient and use them most of the time. It's just a matter of keeping the highlights below clipping and everything else falls in to place more or less automatically at capture providing a file with detail everywhere I can adjust to match the result I pre-visualized at capture like I did 40 years ago with B&W and a darkroom.
There's more to my overall approach that is evident in the examples I post, but it's all explained on my web site for those interested enough to read the tutorials and grasp 100% of the way I do things 
Edited on Jan 19, 2012 at 05:05 PM · View previous versions
|
