Chuck, thanks again for the insightful post. I feel like I "know" most of what you touched on above, but maybe not. I'll go over it again to see if something sinks in that I'm missing. I haven't visited your blog yet... will do that when I have a little more time.

As for point #4, I don't understand how ISO matters. It doesn't change the ratio between the background and subject; why would it matter what ISO is set at?

From your post I can't tell if you're a proponent or opponent of ETTL, maybe that's not the best way to put it. Myself, being someone who occasionally uses flash (mostly as fill), ETTL seems like a headache. For the most part I don't shoot with my camera in any AE mode so I can be in control the exposure, I feel the same about letting the flash make decisions, or as you more accurately stated, "best guess", for it's role in the exposure. I know some people rely on their ETTL setups and will spend lots of $$$ for wireless transmitters and receivers, but with the FEC and chimping involved it seems like I might as well use manual. I guess I'm lucky not to have encountered many situations where the light is rapidly changing.

I hope I'm not coming off like some sort of ETTL hater, don't want to sound that way at all

Dawei Ye wrote:
580EXII is the way to go. A 430EXII costs more because you'll eventually sell it (lose money on it) and buy a 580EXII anyway.
That is the most retarded statement I've heard in quite a while. He's new to photography and doesn't need/want/afford the top of the line equipment. I've used my old companies 420ex flash for 5 years and never had a complaint. It worked perfect for me. When I left that company I had to buy a flash for myself and I got a used 430ex for $195 and I'm perfectly happy with it. I have no need to upgrade to a more expensive model.

Reread cgarnder's post again above about bouncing. It makes a huge difference in the quality and professionalism of your photos by softening the shadows. I purchased the Flip-it with diffuser from here: http://www.dembflashproducts.com/

As you can see from my picture using the flip-it there are no hard shadows being cast onto the wall behind them:

back when the 550EX was current I tested it against the Sunpak 393, Vivitar 285HV (both in A mode) and the Sigma 500 Super. The Sunpak and Vivitar gave the most consistently well exposed exposures, *but* they tended to blow highlights, the 550EX and Sigma 500 gave identical exposures, except they tended to expose to protect the highlights no matter what, so even if the overall exposure was 2+ low, you had your highlights... with my 20D I could not recover more than 2 stops underexposure in raw witout having serious shadow noise issues. Bottom line, the Sunpak and Vivitar dumb flashes gave much better exposures with no adjustment. Yes, for perfect exposure you have to adjust either... but once you are making manual adjustments I don't think it matters how you are doing it, so ETTL offers little advantage. The biggest reason to choose a single ETTL unit is for focus assist and *maybe* high speed sync (I bought a flash that has it because I thought it was cool, but have used it maybe a handfull of times).

Jacob:

What I'm referring to is situation indoors in a dark room where the camera is already wide open and as slow as can be hand-held then higher ISOs are used to create more background ambience with the flash lit scenes foreground. At lower ISOs when the comparison of ambient and flash is done in the evaluative flash metering very less ambient light would be detected and more flash would be used. As the ISO rises so does the overall ambient exposure and less flash is needed. Why its an issue is because the reason to use flash indoors often isn't the lack of ambient light intensity but the fact it hits faces from a steep unflattering angle (i.e. dark eyes shaded by brow) or its a different color temp than the flash (nearly always the case indoors).

I use both ETTL and M modes as the situation dictates. ETTL when either I or the subjects are moving around, and M for things like portraits or still life where the flash/subject distance doesn't vary.

Functionally there is no difference between changing power manually or changing it via FEC adjustment in ETTL mode: IF THE SUBJECT IS STATIONARY. The difference manifests itself when shooting situations which aren't static. ETTL reacts automatically to changes in subject distance. Once FEC is set to get correctly exposed highlights if the subject moves around the exposure will stay consistent automatically. WIth manual mode you'd need to either change aperture or flash power as the subject distance changes, which just isn't as practical.

When using my Vivitar flash manually in dynamic situations I'd shoot very systematically from pre-determined distances of 16, 11, 8, 6 feet. Just like f/stop numbers those distance increments change flash power by one f/stop due to the inverse square law. Starting from a baseline of 11ft @ f/8, I'd stop down to f/11 if moving into 8ft, and f/16 if moving into 6ft for a shot. If moving back to 16ft for a wide shot, the lens would be opened to f/5.6. After doing that repeatedly your brain will start to automatically measure distance in f/stops, not feet. It works with dual flash also. As the shooting distance changes you move the off camera flash in or out to keep the ratio the same. Fill / Key distance combinations of 16/11, 11/8, 8/6, and 6/4 all produce a 3:1 highlight/shadow ratio on a face (2x brighter key + 1x fill : 1x fill = 3:1)

FWIW: The idea that flash is "fill" is incorrect in most situations outdoors. On a sunny day there are two sources of light: the direct rays of the sun, and light reflected from the sky in all directions as fill.

When a person is standing facing the direct sun with shaded eyes, or face half in sun and half in shadow adding flash lifts highlights and shadows equally. The flash "fills" the shadows in the sense they get lighter, but so do the highlights. the contrast range between shaded eyes and sun lit cheek doesn't change because both are skin of the same reflectance and both are getting hit with the same amount of light. In that scenario the sun on the face is the "key" light, the light from the sky is acting as fill for the areas not hit by the direct sun (like the eyes) and flash? Since its hitting both the highlights and shadows at the same time its neither key or fill, its acting as both which really doesn't help much in the technical sense of exposure or contrast reduction.

If we start with normally exposed skin in the sunlit cheeks then add flash the highlights become overexposed. If we lower the ambient with shutter (assuming we are below x-sync) the eyes we just lifted with the flash get darker again. On a perceptual level we will react to the "twinkle" of catchlights in the eyes the flash creates in that situation and will accept lighter looking highlights as "normal", but at some point the balance gets out of whack the the brain says "that looks fake and over-flashed". Because caucasian skin reflects more red, green and blue in that order, the red channel in skin clips first. The brightest skin highlights, when overexposed by just 1/3 stop will start to get a flat yellowish look.

Back in the days B&W fill flash wasn't needed in direct sun to put detail in shaded eyes because the "normal" dynamic range of the film was about 10 stops scene illumination and with reduced negative development or the use of less contrasty print papers scenes with ranges of 12 or 13 stops could be reproduced normally. Contrasty lighting became more of a problem when photography switched to color which only has a range of about 5 stops.

When flash was used photographers found a better strategy in sunlight: put a person's back to the sun and expose the background normally, then lift the shady side of the face to match the background INDEPENDENTLY with the flash because there is no overlap of flash and sunlit highlights.

In that scenario the shady side is filled by skylight which is very soft but -3 stops darker than the sunny side. A digital camera with a 6-8 stop range renders the shaded face quite dark before flash is added if the background is exposed normally. When flash is added to the front of the subject it creates a highlight pattern on top of the skylight fill. The areas on the face not highlighted by the flash remain "filled" by sky and unchanged in tone.

Creating highlights is what a "key" light does by definition. So in that scenario the sun changes its role to "accent / rim / hairlight", the flash becomes the frontal "key" light, and the fill for the shadows (which remain quite dark) still comes from the sky. The flash lit face will still look "fake" unless the it is raised above the head of the subject. That gets us back to the initial point I made about adding flash: the direction matters. If we want flash light to look "natural" the direct it comes from must mimic the direction of natural light sources.

In that backlight scenario outdoors if we want the shadows on the face any lighter, we need to add a secondary fill source to boost the light from the sky from the direction of the camera.

The "magic" of the photographic process is that when exposure is correct in the technical sense of having detail in both the darkest shadows and brightest solid objects at the same time, all the tones in between fall into place perfectly perceptually because the response is engineered to be linear. That's part of what makes the Master on bracket / slave on stand arrangement with hot shoe flash both a practical and technically sound solution to matching the contrast of any scene to the range the sensor can record with detail.

Chuck

Dang, some real flash buffs here! I'm all lit up!

>>Functionally there is no difference between changing power manually or changing it via FEC adjustment in ETTL mode: IF THE SUBJECT IS STATIONARY.

No, when you change your framing at all your camera is metering off a different part of your subject, and unless it is a gray card you have to re-adjust your FEC etc. since the camera is adjusting to the changes it sees... so you have to anticipate the complex ETTL camera function. Dumb flash is easier to anticipate most of the time.

Chuck, you're a wealth of flash knowledge!

Your high ISO situation makes more sense after further explanation. I honestly try to avoid using flash in that situation because I don't have the skill to make it look good. This is where I grab a fast prime and turn the ISO up to 1600 or 3200 and let 'er rip.

Outdoors I typically (try) to...
a) expose for the bright side of the face (assuming side lighting)
b) not shoot in harsh direct sunlight to avoid racoon eyes
c) shoot in the shade or backlight or side-light with fill from the front or dark side of the face
d) try not to mess up or get any light where I don't want it, and sometimes succeed!

I always advise to get the best equipment one can afford, because one can grow into it as one gains experience. One may not think he or she may want more, but it often happens. If one settles for lesser equipment, there's less room for growth.

Sometimes it's better to keep using what one has, and save until later for something significantly better, rather than buying now something only marginally better.

That said, a used 550EX (or maybe even a 580EX (original, not Mk II)) might be had for US$150.00 if one shops around. That would allow for growth into wireless off-camera flash at a later date, if desired at that time.

On the other hand, there's much to be said about manual flash as a learning tool, and a Vivitar 285HV is a darned good flash even if it's never used in Auto mode. For around US$90.00 brand new, it's a tough act to beat. (And you can be saving for a brand new 580EX Mk 7 while you're using the 285 for the next several years.)

+1 for the Demb Flip-It.

BrianO wrote:

On the other hand, there's much to be said about manual flash as a learning tool, and a Vivitar 285HV is a darned good flash even if it's never used in Auto mode. For around US$90.00 brand new, it's a tough act to beat.

If you're going manual the Chinese flashes are pretty good, Yongnuo YN460 or YN462 - for $35 it is even more of a "tough act to beat" !

mh2000 wrote:
>>Functionally there is no difference between changing power manually or changing it via FEC adjustment in ETTL mode: IF THE SUBJECT IS STATIONARY.

No, when you change your framing at all your camera is metering off a different part of your subject, and unless it is a gray card you have to re-adjust your FEC etc. since the camera is adjusting to the changes it sees... so you have to anticipate the complex ETTL camera function. Dumb flash is easier to anticipate most of the time.



That's not the case with evaluative metering which is "smarter" than that.

For clarification, by "stationary" I meant nothing changes: subject/ framing. Perhaps I should have said "if camera, flash, and subject are kept in the same relationship". The underlying point was that FEC and manual power settings both do the same thing: change the flash duration which affects exposure, but ETTL mode will compensate exposure if the scene changes, M will not. ETTL doesn't always compensate perfectly as scenes change where something is in the frame affects the exposure decisions with evaluative, and because the metering zones are still rather large and the metering can only deduce by comparing reflections what tone and how far away the nearest objects are. It makes a logical guess, but will only guess perfectly about half the time. The rest of the time the result is close, but requires tweeking to get the results you had in mind. Bottom line? The camera metering can't read your mind it can only react to the reflected light you put in the viewfinder in a programmed, logical way.

At the end of the day Canon flash / metering is what it is, and you need to learn to deal with it or move on to something else. I'd used something else, manual flash and various TTL metering systems for years, but I quickly came to the conclusion I just could not predict in advance how the camera would meter the scene. So rather than try to out guess with spot metering or dialing in adjustments beforehand, I started pointing, shooting, evaluating, then adjusting from the camera's "I think this is correct" baseline of EC=0 and FEC=0. I found I got to the correct exposure faster than way.

The best aid for determining ideal exposure turned out to be the "idiot light", the warning that blacks out the playback when the highlights clip. But to use it to record a full range of tones I needed to see when the highlights are clipping. I've always used control targets like gray scales in my day job in reproduction and printing, but a solid white looks the same, 255.255.255 if overexposed by 1/3 stop or 33 stops, so I looked around for something white but with texture and grabbed the nearest thing: a white towel. With some experimentation and comparison of camera warning to results in the RAW file I noticed the texture in the towel and skin started to clip at the same time. So the towel texture is an ideal indicator for correct skin exposure when setting portrait exposure.

Chuck - Have you tried the 580exII's automatic exposure mode yet? If so, I would be interested in your thoughts.

BrianO wrote:
That said, a used 550EX (or maybe even a 580EX (original, not Mk II)) might be had for US$150.00 if one shops around. That would allow for growth into wireless off-camera flash at a later date, if desired at that time.

I'm wireless right now with my 430ex. You don't need a 550 or 580 to be wireless.

You can see my wireless setup HERE.

Good morning. To add one more point/question to the ETTL discussion...

I had read that using FEL would producce a more accurate result when shooting with ETTL. If I recall the reason given was that the flash would "spot meter". I tried this with my 5D and 580ex II and didn't see any difference between results. I'm not talking about a focus and recompose situation, I was simply attempting to see if the spot metering theory held up. Perhaps the 1 series bodies (or other newer bodies) behave differently. Maybe later I can dig up the source of that info. There was also mention of changing the camera's ETTL mode (under custom functions) to "average" rather than evaluative would improve metering accuracy under most situations. I did this and I *think* it made a small improvement. I never pulled out my manual to see what the difference is, and I admit that I haven't used the flash in ETTL enough to compare the evaluative metering vs. average.

>>That's not the case with evaluative metering which is "smarter" than that.

that's what I meant, it can be hard to out-smart a "smarter" metering system.

>>At the end of the day Canon flash / metering is what it is, and you need to learn to deal with it or move on to something else.

That's the bottomline... I don't have any problem with ETTL as a system.

Jacob:

FEL allows the flash metering to lock on something a specific distance from the flash. But when using it realize a few things:

1) Any spot meter reading tries to render the spot you measure middle gray. That's the nature of meter calibration. The meter can't know if the spot is dark, light, or in-between, so it provides the exposure settings needed to put what is measured in the middle of the tonal scale in the photo (Zone V in ZS parlance). If the spot measured is lighter or darker than middle gray, then its up to the photographer to INTERPOLATE from the reading to obtain the exposure needed to render the tone correctly. That's a function of the camera's total dynamic range. If the sensor can record 6 stops with detail (about average for most DSRLs) if you spot meter a white dress, you'll need to add 2.5 - 3 stops more exposure to the indicated reading to render it white in the photo. To render a spot metered face correctly in the highlights you'd need to add about 1 - 1.5 stops, depending on the tone of the complexion.

2) Flash is only correct in the highlights at one distance for any given power level. So lets say you set FEL on a face 16 feet away, and adjust the spot reading +1.5 stops so the highlights are rendered correctly. That will expose the face at 16ft perfectly, but if there are also faces 11ft away they will be over-exposed by 1-stop. Any faces at 8ft will be overexposed by 2 stops. Those at 6ft will be over-exposed by 3 stop. So FEC will provide correct exposure on the spot that is measured (after interpolation) but it will not suspend the laws of physics regarding light propagation.

All things considered I think its easier and faster to just use the clipping warning and scene highlights to gauge when flash exposure is correct. Raise FEC to the point where any more flash would cause clipping in what you want correctly exposed. In the scenario above with a person at 16th the warning would tell you closer objects are clipping. That allows you to make a more informed decision to either: 1) shoot as cropped and allow the foreground to be blown out, or; 2) change the camera position or flash strategy to avoid putting anything closer to the flash source than what you want correctly exposed.

What I conclude with Canon metering is that trying to outsmart it is a futile exercise, so when shooting new scene I just let it take its "best shot" at FEC=0 in evaluative and see where it lands (point of correct exposure front to back in the scene) with the clipping warning then adjust the aim with an FEC override.

Try it that way. If you don't like it and find it more convenient you can always go back to how you did it previously

Chuck

Mark Peters wrote:
Chuck - Have you tried the 580exII's automatic exposure mode yet? If so, I would be interested in your thoughts.

I don't own a 580exII. I already have two 580ex and a third avaliable when needed and thought most of the changes on the mkII were more for marketing (catering to the "manual is better" Strobist ethic) than functional. So I am sitting that one out hoping for something better in the next model, like one that doesn't switch randomly to TTL mode and fire at full power. I correspond with Chuck Westfall occasionally and have provided suggestions on flash features, but I don't think he has much clout when it comes to design input.

I think both the PC socket, waterproofing and auto mode on the 580exII are classic examples of the marketing department dictating design. How is having flash controlled via a sensor on the flash any different in function than just setting the flash metering in the camera to "Average" mode? Connecting a radio trigger to a 580ex is simply a matter of using an accessory shoe, which is actually better because it also solves the problem of mounting the flash to the stand. Water proofing? How many people use flash in the rain, and why did the convenient "Master/Slave/Off" switch need to be a casualty? The old battery door design on the 580ex had a detent to prevent the batteries from falling out expectedly; canon just failed to provide good instructions for using it: Open it to the detent, then pull downward slightly to open it fully.

I used "auto" Vivitar flashes since the 1970s. They worked OK for single flash use with a 50mm on a 35mm camera with negative film because the sensor on the flash and the 50mm lens captured the same field of view, and negative film has a 2 stop latitude for overexposure. Today with digital and zoom lenses the scene the camera captures may be much different than what a sensor on the camera is metering and exposure tolerances are much tighter because files are exposed for the highlights where even 1/3 stop variance is noticed.

Auto doesn't work with multiple flashes. When flash A sees the light from flash B is will curtain its output. Even in the best case scenario where one flash doesn't see the light from the other the metering in the flash is based on the light reflecting back in the direction of the flash, not what is reaching the camera.

So auto mode on a 580exII is a bit like the tail fins they put on cars in the 1950s - something that makes little functional sense: average TTL mode is the functional equivalent and will yeid better results with zoom lenses. Most of the mkII features seemed based on marketing to counter the trend towards "Stobist" style use of manual flashes with radio triggers and auto flash doesn't even fit that scenario. Its as if the head of Canon marketing handed the head of Canon flash design a Vivitar 285HV and said, "Add all the same features to the new 580exII so people will stop buying these &&%$$ things". I can't recall reading an account anywhere of a 580exII actually using the auto mode and finding it different / better than the TTL metering in average.

Chuck

cgardner wrote:
Jacob:

FEL allows the flash metering to lock on something a specific distance from the flash. But when using it realize a few things:

1) Any spot meter reading tries to render the spot you measure middle gray. That's the nature of meter calibration. The meter can't know if the spot is dark, light, or in-between, so it provides the exposure settings needed to put what is measured in the middle of the tonal scale in the photo (Zone V in ZS parlance). If the spot measured is lighter or darker than middle gray, then its up to the photographer to INTERPOLATE from the reading to obtain the exposure needed to render the tone correctly. That's a function of the camera's total dynamic range. If the sensor can record 6 stops with detail (about average for most DSRLs) if you spot meter a white dress, you'll need to add 2.5 - 3 stops more exposure to the indicated reading to render it white in the photo. To render a spot metered face correctly in the highlights you'd need to add about 1 - 1.5 stops, depending on the tone of the complexion.

Makes sense, there's really no difference in use of the camera's spot meter.


2) Flash is only correct in the highlights at one distance for any given power level. So lets say you set FEL on a face 16 feet away, and adjust the spot reading +1.5 stops so the highlights are rendered correctly. That will expose the face at 16ft perfectly, but if there are also faces 11ft away they will be over-exposed by 1-stop. Any faces at 8ft will be overexposed by 2 stops. Those at 6ft will be over-exposed by 3 stop. So FEC will provide correct exposure on the spot that is measured (after interpolation) but it will not suspend the laws of physics regarding light propagation.

Should be true whether or not using FEL.


All things considered I think its easier and faster to just use the clipping warning and scene highlights to gauge when flash exposure is correct. Raise FEC to the point where any more flash would cause clipping in what you want correctly exposed. In the scenario above with a person at 16th the warning would tell you closer objects are clipping. That allows you to make a more informed decision to either: 1) shoot as cropped and allow the foreground to be blown out, or; 2) change the camera position or flash strategy to avoid putting anything closer to the flash source than what you want correctly exposed.

What I conclude with Canon metering is that trying to outsmart it is a futile exercise, so when shooting new scene I just let it take its "best shot" at FEC=0 in evaluative and see where it lands (point of correct exposure front to back in the scene) with the clipping warning then adjust the aim with an FEC override.

Try it that way. If you don't like it and find it more convenient you can always go back to how you did it previously

Chuck

I think a lot of what we're talking about here is sort of elementary exposure stuff, which is fine, I went off topic a bit with ETTL vs. Manual. I understand inverse square light fall off. My comments were based on my own experience (or inexperience) with ETTL and some unexpected results (not all the time, just sometimes).

With regard to FEL, I had read that using it would lead to more predictable results. This would make sense given it's a spot meter and if one knows how to use a spot meter properly the results shouldn't be too surprising. When I experimented with it I didn't see any major difference in the results....but it's probably not something that can be compared in any way other than a decreased rate of "random" incorrect flash exposures.

To summarize some points that have been raised by Chuck or me, so I can get them organized in my head, but also for anyone else following along...

1. Canon ETTL with no FEC is generally going to underexpose a scene, it's apparently designed this way to protect highlights (or to provide fill light depending on how you look at it). I think we both agree here.

2. ETTL is smart but not a magic bullet. Scenes with widely varying dynamic range is almost certainly going to require some FEC to obtain the desired exposure. I think we both agree here.

3. ETTL with FEC may occasionally cause incorrect flash exposure. For purpose of this thread, this is just my experience. Most likely it was caused by me making small adjustment in the way I framed the scene causing the metering to change.

4. Using flash exposure lock (FEL) may be useful in getting the exposure correct in any situation where the ETTL system would select some other exposure based on the scene or changes in the scene between shots. FEC will need to be used accordingly unless the FEL target is neutral gray and large enough to cover the majority of the spot meter. I think we both agree here.

5. It is probably not practical to FEL every shot unless you happen to shoot with a 1 series body that can lock the exposure for longer than a single shot (I believe). So this is probably more of an academic exercise than an every day solution. I think we would both agree here.

6. All of the rules of flash exposure apply whether or not the flash is set to ETTL or M. I think we both agree here.

7. A lot of this doesn't matter if you get your flash off camera, unless you're using off camera ETTL... in which case you probably already know what you're doing

mattr762 wrote: I'm wireless right now with my 430ex. You don't need a 550 or 580 to be wireless.

I was refering to wireless ETTL flash using the Canon system. A 430EX won't work for that, because it cannot function as a Master unit.

Obviously there are Pocket Wizards, Radio Poppers, Cyber Commanders, and myriad other ways to go wireless, some of which would work with almost any flash unit up to and including studio strobes, but with varying degrees of compatibility with Canon's ETTL functionality.

Jacob D wrote:

4. Using flash exposure lock (FEL) may be useful in getting the exposure correct in any situation where the ETTL system would select some other exposure based on the scene or changes in the scene between shots. FEC will need to be used accordingly unless the FEL target is neutral gray and large enough to cover the majority of the spot meter. I think we both agree here.


Your last sentence is why I think FEL is waste of time. If its necessary to spot read, take a test shot, then adjust FEC based on the test, its just as simple to take a test shot in evaluative, evaluate it and adjust FEC, skipping the spot reading step.

You understand how FEL and flash in general works, but many don't and seem to think FEL is a magic bullet that will make the entire scene correctly exposed based on where they aim the spot. Like you also said sagely; it will if the spot is gray (and everything else is at the same distance to the flash)