600 EX-RT Evaluative vs Average E-TTL

[pwp]

I rarely get what I'd call good exposures with my 600 EX-RT speedlights. It's generally underexposure. Maybe it's a settings thing. Evaluative vs Average E-TTL? Who uses which, and why?

At indoor events my default EC +2/3 stop, high iso 800-1600 and largish aperture f/3.5-f/5, 5D MkIII & 7D MkII, bounce if there are low white ceilings or on-flash mods like Joe Demb, or less often Gary Fong or Stofen

It's annoying.

Thanks!

-pw

 

[luminaeus]

I gave up on trying to get consistent results from ETTL a long time ago. Maybe I'm just a flash noob that was never able to figure it out, but generally, if I need to use flash, I'll expose to get the background at however dark I want it, and take my best guess at how many stop brighter the foreground needs to be and try to pick a decent manual power option. Two, three, or four chimps later I'll settle on the right settings.

 

[Jack Douglas]

I've recently invested in a second 600EX-RT primarily for fill flash shooting birds with the 6D from a blind. I also don't feel confident with the ETTL results and have been playing around with manual settings. Just when I've run through all the various menus and think I understand it all, a week or two goes buy and I'm not so sure again. In other words I have little to offer but am hoping this thread gets some discussion going.

I have not found a way to eliminate the pre-flash that frightens the little birds just prior to the actual flash. Also when I do a manual verification of dual flash firing there is a noticeable delay in the off-camera flash, which doesn't seem to happen in actual shooting but it's made me wonder what's happening.

Jack

 

[StudentOfLight]

I did a series of tests a while back bounce flash and direct flash using evaluative and average TTL metering. I used a specular dark subject as well as a matte mid-tone subject. Note there are harsh highlights in the "background".

These were the results:

 

[Jack Douglas]

Interesting and useful, how do you interpret or generalize on your results?

Jack

 

[pwp]

Thanks for posting your tests. Your Average ETTL files look more like the one's I'd prefer to be taking into Lr.
I'll run my own comparisons for good measure.

-pw

 

[neuroanatomist]

...have been playing around with manual settings.
I have not found a way to eliminate the pre-flash that frightens the little birds just prior to the actual flash.

If you're setting flash power manually (on all flashes), there will be no preflash.

Also when I do a manual verification of dual flash firing there is a noticeable delay in the off-camera flash, which doesn't seem to happen in actual shooting but it's made me wonder what's happening.

The test flash fires each flash sequentially with a short lag between them, so you can easily confirm they're all firing. I find it helpful with four flashes, especially when they're in different places around the room.

 

[Jack Douglas]

Neuro, you have preserved my sanity, thanks. Great news about the pre-flash - I hadn't perceived that since not all birdies are fast enough to react so the auto versus manual distinction would have escaped me for quite some time!

Jack

 

[ashmadux]

At indoor events my default EC +2/3 stop, high iso 800-1600 and largish aperture f/3.5-f/5, 5D MkIII & 7D MkII, bounce if there are low white ceilings or on-flash mods like Joe Demb, or less often Gary Fong or Stofen

Help -

Im in this situation now with a huge show coming. And ill be passing off jpeg at the end of the day to editors, so good lighting i critical, because they are now goign to be doing tons of photoshop. Behind the scenes with designers/ makeup artists.

For a less experienced person for shows, do you think i should just go with a fong (and a LOT of batteries). I have a flashbender that will be pretty crap for tall full body images.

Also, when there is ok light for iso 800, have you ever tried 2.8 / f4 and then just change the shutter speed as needed? hat about using the flash as an AF assist only when possible?


Thx in advance

 

[Jack Douglas]

Hmm, this sounds like it'd be good info. Hope you get some responses.

Jack

 

[Zeidora]

It all comes down to what the light meter expects to see: neutral grey all over.
You have three factors:
- tonality of objects: light objects will be underexposed, dark objects overexposed. Exactly the same as with ambient light.
- distance of objects: close objects will be overexposed, distant objects underexposed. Light fall-off is proportional to square of change in distance.
- Contribution of available light. With fill flash, the first two factors are reduced, with pure flash, exacerbated.

Now you have to estimate what percentage of the image area is how far off from neutral grey and make adjustments accordingly. Full black is -2, snow white is +2. With some practice, you can get the pre-visualization right within about 1/2 f-stop. That works quite well for (center weighted) average metering. Evaluative metering is a crap shot. It may improve your shots if you don't do any adjustments, but is utterly unpredictable.

E.g. Egg occupies about 10% of image area and is very bright (say +1.8), remainder (90%) is far away so will be rendered quite dark, as well as the horse (say -1.6). The egg is in the center, so in center weighted average gets a bit more emphasis. I would shoot at -1.3 or -1.5, depending on whether you have camera set to 1/3 or 1/2 f-stop adjustments.

If you want to be accurate on first try, do incident light metering with flash meter (I have a Sekonik 558L) and set flash/f-stop manually.

 

[Jack Douglas]

Zeidora, that's excellent relative to my somewhat basic flash use, and much appreciated. Mind you I can't quite picture myself in the scenario that was painted in this thread.

If I understand this correctly because the flash itself uses a type of averaging based on the three points you've mentioned, the choice of compensation necessarily must be based strictly on ones mental perception of the objects that are around, how distant, large, and what shade they are. Now if more than one flash is used things would change considerably, I suspect? How about bounce?

Jack

 

[Zeidora]

Hi Jack,

The flash is dumb, the camera controls the flash output. So you need to understand how the camera controls the light output of the flash, and that is by the camera's light meter. Re mental picture, I try to mentally figure out how the light meter in the camera perceives the scene, and how it need to adjust the exposure so that I get what I want.

Re multiple flashes, then it depends on how they are positioned. If you put them side-by-side to double the power/get one more f-stop, then all of the above is the same. If you stagger them, then the light-fall-off will be reduced. Then you can also add the complication of changing ratio of flash powers, etc. This is where a flash meter comes in handy.

Re bounce, it has two effects: softening light, and reducing distance fall-off because the apparent light source is now between camera and subject on the ceiling. [also reduces flash-power by quite a bit, and may induce color cast if the bouncing surface is colored]. Scattering angle of light (or zoom setting) can further affect it.

IMHO, once you enter those more complex situations, calculating everything and get a hole-in-one is pretty much impossible. There are just too many variables.

P.S. The sunshade smiley should have read -1.8 No idea about all those keyboard shortcuts.

 

[Jack Douglas]

Zeidora , thanks for that. There is so much to factor in but at least I'm understanding the principles. I've started using fill flash with blind shooting of birds with two 600s, one 1 1/2' off camera and one forward and off to the side some 8', and so far the results are promising but I'm having trouble balancing the natural light, especially as the sun slowly goes down. Guess it's trial and error. Trouble is my models don't always cooperate. What's your reaction to a bird when there are two dominant eye reflections - negative?

Jack

 

[Zeidora]

For your situation, I would try best guess, add +/- 0.5 or 2/3 exposure bracket series with sequential shooting. Shooting RAW, you can pull out a lot in post.
With fill flash, which I don't do too often, I set f-stop at around half to one stop smaller (=higher number) than what ambient would give me for flash-sync time of 1/200s. That means, your background (assuming it is overall neutral grey) will be under exposed by half to full f-stop; the birds will be correctly exposed. Assuming the birds are less than one quarter of the image area, then I do exposure adjustment of -1/3 for background -0.5.

Last but not least, you could go all manual. Set exposure time manually at >1/200, set flashes manually by trial and error, then wait.

Re reflection, add a diffusor over your heads (stofen box or similar), that will at least soften the catch light. I guess you figured that a good bird shot is 1 out of hundreds. I don't do birds, too crowded a field. I prefer slime molds and slugs.

 

[Mt Spokane Photography]

Zeidora , thanks for that. There is so much to factor in but at least I'm understanding the principles. I've started using fill flash with blind shooting of birds with two 600s, one 1 1/2' off camera and one forward and off to the side some 8', and so far the results are promising but I'm having trouble balancing the natural light, especially as the sun slowly goes down. Guess it's trial and error. Trouble is my models don't always cooperate. What's your reaction to a bird when there are two dominant eye reflections - negative?

Jack

For a situation where the sun is affecting colors, put the flash at full output and adjust the camera accordingly. A better beamer will concentrate the flash out put and get more light on the subject as well.

 

[Zeidora]

Alternatively, use color gel of same color temperature as sun over flash, and adjust color temp in camera or in post. The same also works for mixed lighting in-doors (incandescent 3200 K with flash + orange gel). However, all light sources must be of same adjusted color temp. So incandescent + sun + fluorescent cannot be corrected. (if they are in discrete areas, then you can use masking in photo editing software) In general, you can ignore off color temp that contribute less than ~10% to overall exposure = 3–4 f-stops less exposure by offending source compared to rest.

Re full power vs. low power, there it depends on how you want far background rendered. Full power makes it black, low power renders it a bit darker, but not black at all.

There are many ways to use flash and each have pros and cons. Have fun playing around with it!

 

[StudentOfLight]

Interesting and useful, how do you interpret or generalize on your results?

Jack

Sorry, I completely forgot about this thread...

Disclaimer: I do not claim 100% accuracy as this info is not clearly laid out in the Canon literature I've found so far. This is how I believe the ETTL flash metering works in the Canon system.

It would have been nice if Canon used a different terms for camera evaluative metering (full scene ambient measurements) and ETTL evaluative flash metering. (flash contribution measurements) This cause me quite a bit of confusion for many months when I started shooting with flash.

When you are using ETTL flash, the camera automatically also meters the ambient across the entire frame regardless of what camera metering mode you are using. When you subsequently push the shutter release, the camera sends a pre-flash and measures a flash contribution exposure. It then compares the ambient exposure with the pre-flash exposure and calculates the required power to "properly" exposure and add/subtracts whatever flash-exposure-compensation(FEC) you've dialed in before firing the shot.

There are two ETTL metering modes Evaluative and Average. ETTL Evaluative metering takes focus distance info as well as subject reflectivity into consideration when determining a "proper" exposure. Average only considers how much light is being bounced back from the pre-flash and ignores harsh reflections. I would avoid ETTL average metering for direct flash, especially if there is a close foreground subject. IMO average is best used for bounce flash.

If you have highly reflective objects in the frame then evaluative will underexpose to try to protect against hotspots. If you do not care to preserve the specular highlights then average will work fine. If you do want to preserve the highlights then evaluative gives you some highlight protection. If using direct flash (e.g. fill-flash) then evaluative is useful as it takes depth info from the autofocus system to dial in the correct power.

I prefer to use bounce flash. I use ETTL evaluative most of the time and dial in FEC as required. I don't have the presence of mind to shoot manual flash outside of a studio setting.

 

[privatebydesign]

Put more simply, Evaluative tries to determine the subject within the scene and attempts to expose that to 12% grey regardless of the surrounding scene. It does this by comparing the reflectivity within the various metering zones and takes subject distance into account when using a lens that reports it.

Average doesn't do the zone thing and doesn't try to work out what your subject is, it just trys to put out enough light to get the entire frame to 12% grey.

Both have their strengths and weaknesses, the nice thing about Evaluative is the subject prioritization, any FEC only needs to account for the subject reflectance, but it is proprietary and we don't know how much power the flash will contribute for any single exposure as it changes as the EV changes. The nice thing about Average is it is more predictable but more difficult to work out required FEC on the run, is the subject dark, how far is the background, does the flash have the power to light the entire scene etc all need to be answered to work Average with any consistency.

So a scenario, you are at the Superbowl on the sideline, you need some fill flash for the end of game rush onto the field and congratulations, you have the perfect angle kneeling and looking up with the stands in the background and the winning coach and quarterback shaking hands just a few feet in front of you, in AV and Evaluative the flash will expose the two subjects well and the shutter speed will drag to get the stands not black, in Average it will over expose the subjects and the stands will be dark because it is trying to light the stands and the subjects so will default to full power.

 

[Zeidora]

Interesting re taking focus distance into account. Hadn't thought of that, as I pretty much always use MF lenses.

The imperfect performance, then, comes from the problem that the metered areas containing in-focus points may also include distant background (or blown out foreground). And with side light, you also get the problem of lighting contrast of 3D objects. Interesting ...