ETTLII Query

[burchy]

My understanding of the difference between ETTL/ETTLII is that ETTLII takes into account the subject distance from the camera, i.e. the focus distance currently achieved by the camera. Does anybody know if this is still taken into account if the flash is bounced, i.e. if the flash head is pointed vertical, does the flash still take into account distance information passed by the camera?
The reason I ask is that I’m shortly going to upgrade to an ETTLII body and have a 550EX (ETTL only), however virtually all of my flash shots are bounced of the ceiling indoors. So I’m trying to work out whether I’d actually gain any benefit at all upgrading to an ETTLII flashgun.

[oni0n56]

I think so. It works by doing a pre-flash riight before the real one, and then analyzes it and all for the correct exposure. My experiences with bouncing have been great.

[dgcorner]

From what I understand -- all EX models are capable of ETTL-II.

[Curtis N]

Burchy,
Somewhere there is a post from Scott Berger who spells out the differences in detail.

But there is no such thing as an E-TTL II flash gun. E-TTL II is a function of the camera, not the flash. The flash just does what the camera tells it to do. Here is the text from Canon's USA website on the features of the 580EX:Canon's top-of-the-line Speedlite has been thoroughly revised. It's smaller and lighter than ever, yet provides a bit more power and significantly faster recycling. A new Select Dial makes changes, like exposure compensation, a breeze. It now covers lenses as wide as 14mm, and has new features for digital SLRs - including the ability to transmit precise flash color temperature information, and vary the zoom setting to match the imaging sensor.
Approx. 25% faster recycling time compared to 550EX.
Superior evenness of exposure, center to corner of frame.
Slightly higher max. guide number at 105mm setting (GN 190, feet).
Auto conversion of flash coverage with compatible digital SLRs.
White Balance info communicated instantly to compatible digital SLRs.
Full swivel, 180° in either direction (first TTL flash with this ability).
AF-assist beam now compatible with all AF points on every EOS SLR.
New Select Dial for easy inputs and user control.Notice they don't mention E-TTL II.

In terms of metering, the 580EX won't behave any differently than a 550EX.

[PacAce]

My understanding of the difference between ETTL/ETTLII is that ETTLII takes into account the subject distance from the camera, i.e. the focus distance currently achieved by the camera. Does anybody know if this is still taken into account if the flash is bounced, i.e. if the flash head is pointed vertical, does the flash still take into account distance information passed by the camera?
The reason I ask is that I’m shortly going to upgrade to an ETTLII body and have a 550EX (ETTL only), however virtually all of my flash shots are bounced of the ceiling indoors. So I’m trying to work out whether I’d actually gain any benefit at all upgrading to an ETTLII flashgun.

No, if the flash head is pointed away from it's normal straight-ahead position, distance information from the lens, if available, will NOT be utilized by ETTL II. Although the use of the distance information with ETTL II is a good feature, it is not a requirement to make ETTL II work. The main difference between ETTL and ETTL II is the fact that ETTL is AF point biased in evaluative mode (vs average mode) whereas ETTL II is subject biased. So you're not going to be getting less out of your ETTL II body just because you are bouncing your flash.

As for the 550EX flash, as has already been pointed out, it is ETTL II compatible so you won't have any problems using with with an ETTL II camera.

[burchy]

Excellent, thanks guys. I fully understand this now it has been explained.

[jcw122]

No, if the flash head is pointed away from it's normal straight-ahead position, distance information from the lens, if available, will NOT be utilized by ETTL II. Although the use of the distance information with ETTL II is a good feature, it is not a requirement to make ETTL II work. The main difference between ETTL and ETTL II is the fact that ETTL is AF point biased in evaluative mode (vs average mode) whereas ETTL II is subject biased. So you're not going to be getting less out of your ETTL II body just because you are bouncing your flash.

As for the 550EX flash, as has already been pointed out, it is ETTL II compatible so you won't have any problems using with with an ETTL II camera.

I just want to highlight what PacAce said. Not all lenses display or output distance information. Many Canon lenses do (there is a list somewhere on the forums), but IIRC most 3rd party lenses won't.

[Dellboy]

FWIW I have found ETTL II more of a problem than a help. If using the flash with a diffuser then pointing the flash staight forward FEC will need to used as the flash will underexpose ( ie: it will not automatically adjust its output to compensate for the diffuser ). If bouced with a diffuser no FEC will be required. Hence every time you take a pix of someone with a hat ( quite common at weddings ) FEC needs to be changed and then put back again. For this reason I always run in average mode.

Also the red focus grid will not work on the top and bottom auto focus points with a 550EX , except with a 1 series camera.

The 580EX has one one button for changing the flash angle where as the 550EX has two, If a fast adjustment is required ( from landscape to portrait ) this can make quite a difference.

[tim]

Excellent, thanks guys. I fully understand this now it has been explained.

I think it's impossible to fully understand ETTL-II, the best you can hope for is to be able to predict what it will do in most cases.

[jcw122]

I think it's impossible to fully understand ETTL-II, the best you can hope for is to be able to predict what it will do in most cases.

Why do you think it's impossible to fully understand?

I'm asking for two reasons:
-Simply because you didn't state why
-Because IIRC no one literally knows the algorithms to how it actually works, but I'm not too sure.

[J Rabin]

Tim's response is just too honest! ;)

There were two major advancements between E-TTL and E-TTL II.
1st, was forever breaking Canon's link between the active focus point and flash metering bias to this focus point, and the 2nd was improving subject distance in the flash metering calculation.

Of the 2 E-TTL II advances, DISTANCE is LEAST important innovation for obtaining improved reliable flash exposures. Of course, it's no possible to consider distance if you are bouncing, but once you get the hang, both E-TTL and II do fine in this regard.

But, doing away with ALL linkage between focus point bias ambient light metering and flash metering was a BIG advance for users. E-TTL II is a lot more reliable on-the-go, with or without distance.

[Wilt]

jcw, I will side with Tim...there are too many 'behaviors' exhibited by ETTL to be able to predict them all! And they are NOT documented by Canon, and you have to spend many hours lurking the web to discover some of them. You think that you have it mastered somewhat, and then you discover yet another behavior when someone inquires about "Why does...?" The answer often is "because the Canon designers chose to" and no understanding of the logic why!

The truly predictable thing about ETTL is that you will always be stumped by yet another lurking odd behavior.

[tim]

What J Rabin and Wilt said. If you tell me you fully understand ETTL-II i'll find a situation that you can't predict. The algorithm is complex.

[Curtis N]

And to further make it a head scratcher, there's no EXIF data that says how much flash power it used for a shot. So it's harder to measure the behavior of E-TTL in a given situation.

[cgratti]

I think only certain lenses are compatible with ETTL-II, the ones that return the distance info to the body...

[J Rabin]

I think only certain lenses are compatible with ETTL-II, the ones that return the distance info to the body...
This is just plain mistaken. Communicating distance information is completely unneccessary for implementation of E-TTL II evaluative flash metering.
When a lens is used that communicates distance to the flash metering system, and when the flash is directly connected looking forward, the distance info is used in the E-TTL II calculation for a more accurate exposure.

But, if you are using the flash with an Off Shoe Cord 2, or wireless, or with an ST-E2, or in bounce position, E-TTL II metering still works, and works fine, it just does not use this distance information. It still may use some information from the metering pre-flash timing to guess at distance, but Canon's flash engineers are not telling anyone.

I think Tim is sooooo right, it's funny.
Whenever new Canon camera products are introduced, Canon's evangelist Chuck Westfall answers technical questions. When E-TTL II came out in 2004, and professional users were asking these questions, Chuck would post a "well let me go and do some test shot experiments because Canon does not disclose this."
If he don't know how it works, we don't know!
Like Time says, all you can do is understand enough to get predictable reliable results under the situations you encounter. That takes systematic practice. It's like mental reverse engineering - saying to yourself, "now what were these flash engineers thinking when they...."
Jack

Back to the original poster's Q:
The 550EX is a workhouse. It will do E-TTL II fine on any new body.
You did not say which body you are moving to. Be aware that the 1-DMkII implements E-TTL II evaluative different than the consumer bodies. On a 1-DMkII, no matter the metering pattern the camera is in, when an EX flash is used under E-TTL II, the body reverts to evaluative metering pattern mode for the flash metering.
I get much more reliable flash exposures at work with the 1-DMkII than I do with my personal 30D. In addition, the Auto White Balance under flash conditions is more accurate on the 1-DMkII. Don't let anyone tell you E-TTL II execution is the same on consumer bodies and 1-D II bodies.

[tim]

I think only certain lenses are compatible with ETTL-II, the ones that return the distance info to the body...

ETTL doesn't require distance info, but uses it when it's available. Probably best that if you don't know what you're talking about, don't answer questions.

[jcw122]

jcw, I will side with Tim...there are too many 'behaviors' exhibited by ETTL to be able to predict them all! And they are NOT documented by Canon, and you have to spend many hours lurking the web to discover some of them. You think that you have it mastered somewhat, and then you discover yet another behavior when someone inquires about "Why does...?" The answer often is "because the Canon designers chose to" and no understanding of the logic why!

The truly predictable thing about ETTL is that you will always be stumped by yet another lurking odd behavior.

What J Rabin and Wilt said. If you tell me you fully understand ETTL-II i'll find a situation that you can't predict. The algorithm is complex.

These are the exact answers I was looking for. Thanks guys.

[René Damkot]

If using the flash with a diffuser then pointing the flash staight forward FEC will need to used as the flash will underexpose ( ie: it will not automatically adjust its output to compensate for the diffuser ). If bouced with a diffuser no FEC will be required.
Yes, it will. Light is metered through the lens. The camera will shut down the flash after enough light has been emitted. Bouncing as well as straight on.
The only difference is that the character of the light is completely different, since it is not bounced. Therefore it may require a different amount of FEC to look 'good' to you.

[PacAce]

...
Back to the original poster's Q:
The 550EX is a workhouse. It will do E-TTL II fine on any new body.
You did not say which body you are moving to. Be aware that the 1-DMkII implements E-TTL II evaluative different than the consumer bodies. On a 1-DMkII, no matter the metering pattern the camera is in, when an EX flash is used under E-TTL II, the body reverts to evaluative metering pattern mode for the flash metering.
I get much more reliable flash exposures at work with the 1-DMkII than I do with my personal 30D. In addition, the Auto White Balance under flash conditions is more accurate on the 1-DMkII. Don't let anyone tell you E-TTL II execution is the same on consumer bodies and 1-D II bodies.
Jack, I'm a little confused by your statement which I have highlighted above. Can you elaborate on that for me, please. As far as I know, the ambient metering that is selected is independent of and never has any influence on the metering mode that is used for flash, on the 1D-series camera or amy other EOS dSLR. The metering algorithm used for ETTL-II flash metering is a whole different animal from those used for ambient metering. Unless you meant to say that, no matter what ambient metering mode is selected for a 1D-series camera, it will revert to evaluative metering for ambient metering when a flash is in use (although I take exception to that, too, as I think it actually reverts to CWA metering mode instead).

[tim]

Yes, it will. Light is metered through the lens. The camera will shut down the flash after enough light has been emitted. Bouncing as well as straight on.
The only difference is that the character of the light is completely different, since it is not bounced. Therefore it may require a different amount of FEC to look 'good' to you.


If using the flash with a diffuser then pointing the flash staight forward FEC will need to used as the flash will underexpose ( ie: it will not automatically adjust its output to compensate for the diffuser ). If bouced with a diffuser no FEC will be required.

René is right, but I doubt major FEC changes will be required. FEC is usually altered depending on the subject being metered, not anything else. However I do agree that this isn't an exact science.

Honestly, people who don't know what they're on about should really keep quiet. René i'm obviously not talking about you.

[PacAce]

Yes, it will. Light is metered through the lens. The camera will shut down the flash after enough light has been emitted. Bouncing as well as straight on.
The only difference is that the character of the light is completely different, since it is not bounced. Therefore it may require a different amount of FEC to look 'good' to you.
Rene, what you seem to be describing is realtime TTL flash metering. With ETTL and ETTL-II, a preflash is fired before hand (before the shutter opens) to determine the flash exposure required for the shot. When the shutter opens, the predetermined amount of flash power is then fired. At that point the amount of light entering the light box is no longer monitored so there is nothing to shut down.

[René Damkot]

Yeah, I know, should have inserted "as determined by the metering of the preflash" after "emitted". The point I tried to make however, is still valid (unless Dellboy manages to attach a diffuser in the time between the preflash and the exposure :lol:)

[PacAce]

Yeah, I know, should have inserted "as determined by the metering of the preflash" after "emitted". The point I tried to make however, is still valid (unless Dellboy manages to attach a diffuser in the time between the preflash and the exposure :lol:)
LOL :lol: Well, that's very possible, too, Rene. Press the FEC button, attach diffuser, take picture. :lol: ;)

[Wilt]

Be aware that the 1-DMkII implements E-TTL II evaluative different than the consumer bodies. On a 1-DMkII, no matter the metering pattern the camera is in, when an EX flash is used under E-TTL II, the body reverts to evaluative metering pattern mode for the flash metering.
I get much more reliable flash exposures at work with the 1-DMkII than I do with my personal 30D.

Jack, I'm a little confused by your statement which I have highlighted above. Can you elaborate on that for me, please.

There is a Custom Function which you can set to Evaluative or Average (CF14 on the 20D), which is probably what he is referring to.

The Evaluative choice seems to be the 'more automagic' of the two, with the Canon flash logic built in with its harder to predict actions; Average might be said to behave more like the old TTL flash computation which could be scene fooled and requires photographer intervention to compensate. And from J Rabin's comments it seems the 1D Evaluative is better implementation than the 20D/30D Evaluative.

[PacAce]

There is a Custom Function which you can set to Evaluative or Average (CF14 on the 20D), which is probably what he is referring to.

The Evaluative choice seems to be the 'more automagic' of the two, with the Canon flash logic built in with its harder to predict actions; Average might be said to behave more like the old TTL flash computation which could be scene fooled and requires photographer intervention to compensate. And from J Rabin's comments it seems the 1D Evaluative is better implementation than the 20D/30D Evaluative.
The 1D MarkII series cameras (the non-Mark IIs are ETTL) have that custom function, too, and it works the same way it does on the 20D/30D.

[J Rabin]

Jack, Can you elaborate on that for me, please.

PacAce. Yes, I am talking about the fact that when using a 1-D series body, WITH an EX flash connected, it does not matter whether you have spot, partial, CWA, or Evaluative selected for the AMBIENT metering.

I was NOT talking about the custom function for Evaluative vs. Averaging E-TTL II, I was talking about the Evaluative Ambient metering.

On a 1-D with EX, the ambient metering reverts to, or behaves, according to Chuck Westfall, ...as if you were in Evaluative with the Center Point selected, which is similar to CWA...

This affects flash because flash metering works in the following way:
- With EX flash connected you press shutter half way.
- Camera takes an ambient reading independent of flash metering.
- E-TTL II preflash then fires and camera meters flash, metering some ambient light with the preflash.
- Ambient is then SUBTRACTED from the E-TTL preflash to get metered FLASH ALONE.
- As shutter is fully depressed another ambient reading is made to make sure there has been no change.
- Shutter opens, etc.

So, as you follow the above, the ambient METERING PATTERN chosen affects flash, because ambient is subtracted from the E-TTL reading AFTER pre-flash metering.
On a consumer body 30D, 5D, etc., the camera uses the ambient metering pattern user has selected.
On a 1-D no matter the pattern selected, the body uses an algorithm that behaves "as if" Evaluative with center point selected was chosen, even when some other pattern was selected by user on the body. It is close to CWA, but not exactly.

Heck, this is why Tim is right. Nobody but the engineers know. All we try and do is predict.

My empirical experience has has been - for whatever reason E-TTL II is implemented differently this way on 1-D bodies - I get far more reliable flash performance on the 1-MkII than the consumer bodies. On the consumer bodies, CWA and Evaluative give solid results, but it's not as reliable as the 1-D's.

PacAce, this probably way over the head of the original poster, and I did not want to hijack his/her thread.
I apologize.
Jack

[PacAce]

PacAce. Yes, I am talking about the fact that when using a 1-D series body, WITH an EX flash connected, it does not matter whether you have spot, partial, CWA, or Evaluative selected for the AMBIENT metering.

I was NOT talking about the custom function for Evaluative vs. Averaging E-TTL II, I was talking about the Evaluative Ambient metering.

On a 1-D with EX, the ambient metering reverts to, or behaves, according to Chuck Westfall, ...as if you were in Evaluative with the Center Point selected, which is similar to CWA...

This affects flash because flash metering works in the following way:
- With EX flash connected you press shutter half way.
- Camera takes an ambient reading independent of flash metering.
- E-TTL II preflash then fires and camera meters flash, metering some ambient light with the preflash.
- Ambient is then SUBTRACTED from the E-TTL preflash to get metered FLASH ALONE.
- As shutter is fully depressed another ambient reading is made to make sure there has been no change.
- Shutter opens, etc.

So, as you follow the above, the ambient METERING PATTERN chosen affects flash, because ambient is subtracted from the E-TTL reading AFTER pre-flash metering.
On a consumer body 30D, 5D, etc., the camera uses the ambient metering pattern user has selected.
On a 1-D no matter the pattern selected, the body uses an algorithm that behaves "as if" Evaluative with center point selected was chosen, even when some other pattern was selected by user on the body. It is close to CWA, but not exactly.

Heck, this is why Tim is right. Nobody but the engineers know. All we try and do is predict.

My empirical experience has has been - for whatever reason E-TTL II is implemented differently this way on 1-D bodies - I get far more reliable flash performance on the 1-MkII than the consumer bodies. On the consumer bodies, CWA and Evaluative give solid results, but it's not as reliable as the 1-D's.

PacAce, this probably way over the head of the original poster, and I did not want to hijack his/her thread.
I apologize.
Jack
Thanks for the clarification, Jack. That's what I thought you had in mind although it wasn't stated that clearly the first time. :)

BTW, my statement about taking exception to the fact that it reverts to Evaluative metering mode? Forget I said that. I just double checked my data I collected when I first looked into this phenomenum at couple of months ago and I got my wires crossed. I thought it reverted to CWA but it, in fact, does revert to Evaluative.

Thanks again for the clarification.

[J Rabin]

PacAce.
You're welcome. Maybe you can answer an 1-D vs. consumer body E-TTL II behavior Q for me.
Here goes. Does the 1-DMk II body have NO, or at least less, NEVEC than consumer bodies?

Every time I do on-the-go reportage, I get much better flash exposures on the 1-DMk II than the consumer bodies. The 1-DMk II in Av yields more reliable flash exposures than running the consumer bodies in Av.

Someone told me that the 1-DMk II DOES NOT do the evil auto reduction of ambient background exposure during fill flash, what the engineers kindly named NEVEC.
I never liked the look of photos where Canon NEVEC is present, so 80%+ I run the camera in M doing flash reportage.

Like most people, I lust after a 5D for its compact size, weight, and image quality. BUT, if it is TRUE that the 1-D bodies don't do NEVEC, I'll stay right where I am with the 1-DMk II.

What's the truth about this?
Am I making any sense at all?
Maybe Rene D. knows?

Thanks, Jack

[PacAce]

PacAce.
You're welcome. Maybe you can answer an 1-D vs. consumer body E-TTL II behavior Q for me.
Here goes. Does the 1-DMk II body have NO, or at least less, NEVEC than consumer bodies?

Every time I do on-the-go reportage, I get much better flash exposures on the 1-DMk II than the consumer bodies. The 1-DMk II in Av yields more reliable flash exposures than running the consumer bodies in Av.

Someone told me that the 1-DMk II DOES NOT do the evil auto reduction of ambient background exposure during fill flash, what the engineers kindly named NEVEC.
I never liked the look of photos where Canon NEVEC is present, so 80%+ I run the camera in M doing flash reportage.

Like most people, I lust after a 5D for its compact size, weight, and image quality. BUT, if it is TRUE that the 1-D bodies don't do NEVEC, I'll stay right where I am with the 1-DMk II.

What's the truth about this?
Am I making any sense at all?
Maybe Rene D. knows?

Thanks, Jack
I just checked this tonight in my office where the lighting is EV=4 and it looks like the exposure setting of the 1DmkII when using the flash is indeed affected by NEVEC. This is the same result I got the last time I did this test a year ago.

The test was done with the metering set to Evaluative in both Av and Tv modes (to avoid the change in exposure I would expect when the camera is set to any other metering mode with the flash off and the mode being forced into evaluative with the flash on).

When I did a similar test this afternoon (there was more ambient lighting from the outside in the room), the exposure setting when using Evaluative metering did not change whether the flash was on or not.

[J Rabin]

Thanks PacAce.
So, at low EV (4 is pretty darn dark), the 1-D Mk II flash still implements NEVEC on background ambient. No wonder why we all have to use M in lower light.

I am placing here one of Chuck Westfall's exhanges on Galbraith Forum couple years ago before I came to POTN. I saved it as a web archive. Please give a right-click download, read, and tell me how his finding that 1-DMk II does not do NEVEC squares with your observations?
http://postit.rutgers.edu/uploads/E%2DTTLII%5Fno%5FNEVEC%5F1DMkII.webarchive

Or is he refering to auto reduction of fill?
Damn this is confusing. I had three EX's set up with an ST-E2 today shooting some TaeKwonDo poses, and no matter how much I study, I still end up in trial and error mode. Especially since the ST-E2 does not do a "C" channel for background, so I had to Kludge "B" into double duty.

Hells Bells. JACK

[PacAce]

Thanks PacAce.
So, at low EV (4 is pretty darn dark), the 1-D Mk II flash still implements NEVEC on background ambient. No wonder why we all have to use M in lower light.

I am placing here one of Chuck Westfall's exhanges on Galbraith Forum couple years ago before I came to POTN. I saved it as a web archive. Please give a right-click download, read, and tell me how his finding that 1-DMk II does not do NEVEC squares with your observations?
http://postit.rutgers.edu/uploads/E%2DTTLII%5Fno%5FNEVEC%5F1DMkII.webarchive

Or is he refering to auto reduction of fill?
Damn this is confusing. I had three EX's set up with an ST-E2 today shooting some TaeKwonDo poses, and no matter how much I study, I still end up in trial and error mode. Especially since the ST-E2 does not do a "C" channel for background, so I had to Kludge "B" into double duty.

Hells Bells. JACK
Unfortunately that link is not working but if I had to make an educated guess, I'd bet that he was talking about the auto reduction of fill.

Re you ST-E2 and Group C dilemma, have you considered putting your "group C' flash in manual mode? This way, you can slave it with the ST-E2 but it won't fire a preflash that will mess up your ETTL exposure for the normal group A and group B flashes. The problem with setting your "group C" flash as part of group B is that the lighting from your background will also be taken into account when the flash expoosure is computed. Theoretically, group C lighting should not used in calculating the subject exposure since none of it directly falls on the subject.

[Dellboy]

René is right, but I doubt major FEC changes will be required. FEC is usually altered depending on the subject being metered, not anything else. However I do agree that this isn't an exact science.

Honestly, people who don't know what they're on about should really keep quiet. René i'm obviously not talking about you.

As I understand with ETTL II distance of camera to suject info is relayed from the lens to the camera flash/ camera combination and this info is used (along with other info ) to determine the flashs' final output. When the flash is bounced I beleive that this info is ignored as it is obviously not relevent to the flashs' final output, however when the flash is pointed straight forward this info is used. So there is a difference in how the flash calculates its output depending on wether its pointing straight ahead or bounced.
In my experience I have found that the flash will expose quite acurately when bounced (with or without diffuser ) or when pointed staight ahead with no diffuser but it will underexpose when when pointed straight ahead with a diffuser.
As the ambient light reduces this underexposure seems to me to become greater so for those that haven't tried this combination of straight ahead flash with diffuser, try it in a dark room with a low iso and a with lens that is capable of relaying the camera to subject distance info.This will ensure you see it's maximum effect. Also check CF 14
When I first noticed this flash behavior I though that something was wrong with my flash but my brothers 350D and 580EX combination behaves the same.
I think this is at least partailly responsible for this flashs' sometimes erratic behaviour, many people have had problems with this flash underexposing - try a google search. I also think that the flash is in a perminate fill in mode when using ETTL II. However as people have aready stated in this thread Canon haven't told anyone exactly how this flash works and as such all these theories we all seem to have will remain just theories.

[tim]

Dellboy, my point was that FEC shouldn't need to be changed if you use direct or bounced flash, since metering is through the lens.

[Dellboy]

Dellboy, my point was that FEC shouldn't need to be changed if you use direct or bounced flash, since metering is through the lens.

Have you tried this with a diffuser? Its certainly very different on a 350D and a 20D.

[tim]

Have you tried this with a diffuser? Its certainly very different on a 350D and a 20D.

I've used a bunch of diffusers. I might've not been clear, my last comment was ambiguous. If you take a well exposed photo with direct flash, then put a diffuser on or change to bounced flash, you shouldn't need to change your FEC.

[Curtis N]

If you take a well exposed photo with direct flash, then put a diffuser on or change to bounced flash, you shouldn't need to change your FEC.My experience differs with that statement. In general, I find that more +FEC is required for bounced flash if the goal is an equally illuminated subject with both techniques. My theory on this is that since bouncing tends to brighten up the background (at least if the room isn't huge), this affects the metering.

Dellboy's experience is plausible to me. If the camera is getting distance info from the lens and uses this info in its calculations since it thinks it's using direct flash, a diffuser is likely to lead to underexposed shots. The way to test this would be to angle the flash head just slightly, enough for the camera to know it's not straight forward, but not enough to affect things otherwise, and see what happens.

[PacAce]

I, on the other hand, would basically agree with Tim, but mostly it's from gut feeling than real experience with the diffuser (although I do use the diffuer and I've never had to add any FEC when using it). So, to find out for myself a little more scientifically, I shot a coulple of test shots. The results are shown below. What I found, at least for the room that I was shooting in, is that ETTL does, in fact, try to expose for the subject and adjusts the flash output level accordingly.

I shot all the pictures in Manual mode @ 1/60 and f/6.3. I set the ISO such that even with bounce, the flash would be able to generate enough output to get what it thinks is the proper exposure. At ISO 400, the flash had a working distance of 30 feet which is more than adequate for the room so that's what I used. FEC was set to 0 in all cases.

Also, it may not be obvious but the subjects and the arm chair was placed 5 feet in front of the back wall.


Since WB differs with the use of the diffuser and bounce, I used the eye dropper on the white baseboard so that the WB of all images would be the same.

The first image shows the results of using direct flash, with and without the difuser.

The second image shows the results of using the flash bounce at 45 degrees.

As you can see, overall, they all seem to be exposed about the same.

107266

107267

On the next post, I show what just the subject area of each image looks like.

[PacAce]

Oops. I forgot to mention that the lens I used is the EF 24-70 f/2.8 L which does give distance information.

[PacAce]

BTW, in case anybody thinks it's relevant, I was using the 30D and the 580EX flash.

[Curtis N]

Anyway, I think the camera and flash model info might be relevant only to answer the question of whether or not the flash zooms to account for the crop factor. In this case, yes. I'm wondering what focal length the flash was zoomed to, because in my limited testing with my Sigma flash, with the flash pointing straight-ahead, I found no difference in the measured range between the Omni-Bounce and bare flash with the wide angle panel extended (data here (http://photography-on-the.net/forum/showthread.php?t=197738)).

The interesting thing about your test to me is that the 45 degree shots both seem a tad brighter, and in my opinion more properly exposed. And assumedly the camera accomplished this without using any distance info.

[tim]

Thanks for doing the tests Leo. My post was made based on my understanding of ETTL, and gut feeling, I believe you backed up my thoughts with experimentation.

[PacAce]

Anyway, I think the camera and flash model info might be relevant only to answer the question of whether or not the flash zooms to account for the crop factor. In this case, yes. I'm wondering what focal length the flash was zoomed to, because in my limited testing with my Sigma flash, with the flash pointing straight-ahead, I found no difference in the measured range between the Omni-Bounce and bare flash with the wide angle panel extended (data here (http://photography-on-the.net/forum/showthread.php?t=197738)).

The interesting thing about your test to me is that the 45 degree shots both seem a tad brighter, and in my opinion more properly exposed. And assumedly the camera accomplished this without using any distance info.
Curtis, the lens was set to 35mm. I can try the set with other focal lengths if you think it would make a difference in the outcome of the test.

As for the 45° shots looking a tad brighter and more properly exposed, that's because of what you said in your earlier post. With the light bouncing off the ceiling, the background wall the the subjects are getting relatively the same amount of illumination. With direct flash, the subject will have more illumination than the background which is farther away from the flash.

[PacAce]

Thanks for doing the tests Leo. My post was made based on my understanding of ETTL, and gut feeling, I believe you backed up my thoughts with experimentation.
Sure thing, TIm! ;)

[Dellboy]

For my own sanity, I took a couple of pix myself. exactly the same as Leo (same lens too ) just used a 20d instead.

1st pic flash straight ahead no diffuser.

http://photography-on-the.net/forum/attachment.php?attachmentid=107293&stc=1&d=1157338675
_MG_2958.jpg

2nd pic flash straight ahead with a Stofen diffuser.

http://photography-on-the.net/forum/attachment.php?attachmentid=107294&stc=1&d=1157338675
_MG_2959.jpg

Maybe there's some difference between the 2 cameras - i certainly wish Canon would sort out this problem on my 20D.

[PacAce]

Dellboy, can you share the full image with us, please. I'm assuming what you posted was just a partial crop of the original. The difference in results you got is certainly interesting and I'd to figure out, if I can, why that is. If you can provide EXIF into, that'll be good, too. Thanks. :)

[PacAce]

Did you have the hood on. If not, can you try another test with the hood on?

[tim]

Try with something made of cloth, that thing has reflective metals in it which confuses ETTL.

[Dellboy]

Hi Leo,
I've just cropped the two sides of the image to fit in with the 800 X 800 pixels rule - otherwise it is full size.
I did not use a hood though but would not expect it too make a difference.


The EXIF data is;

ISO 100
shutter speed 1/100 sec
Appertue 3.2
distance to subject 3-4 feet
CF 14 - ETTL II
EC - 0 ( Zero )

These images are tipical of all the pix I have got cloth, faces or otherwise.
I find the effect is more pronounced when ambeient light levels are low, so I took these in virtual darkness.
If the flash is bounced or even pointed down slightly then there is no problem - this happens with my brothers 350D too so I though it too be just a product of the ETTL II system. As a consequence I now ues CF 14 set to average and the exposures are always predictable.
I certainly appreciate a helping hand to sort this issue out cause it was driving me nuts untill I changed CF 14

Thanks, Dellboy.

[Dellboy]

shot the pix in JPeg and at 70mm on a Canon 24-70mm F2.8
Flash set too ETTL and camera in Manual

[Curtis N]

I've just done a bit of testing myself, and I'm getting some variations that aren't making sense. I can get consistent exposure from one shot to the next if I don't change anything, but when I change something that shouldn't make a difference, such as ISO, I get different results (with exposure confirmation on the flash unit in all cases).

I'll post these two, not as conclusive evidence of anything but to indicate that apparently there's more to this issue than meets the eye.

20D & Sigma Super, EF-S 18-55 lens (no distance info) focal length 55mm
Subject is 18% gray cards on black music stand, backdrop is white.
1/250, f/5.6, ISO 400
C.Fn. 14 set to 0 for Evaluative (not average) flash metering
0 FEC except the bottom image
Flash on E-TTL mode, zoomed to 85mm

[PacAce]

You know what, Dellboy? I was able to duplicate what you are seeing with the straight-ahead diffuser shot. It occurs when the flash is zoomed in the 50 to 105 positions (either manually on the flash or automatically with the lens). I don't understand why that is but I'm certainly going to try to find out. With through the lens metering, this shouldn't be happening.

[Dellboy]

Curtis,
I too have experienced strange results and I carn't really get my head around it. This is why I don't like ETTL II except as a fill in flash.
Many thanks for making me feel a bit more sane over this issue, Dellboy.

[tim]

I'll be watching to see how things come out! In practice I just watch this histogram, and do test shots in each new setup.

[Dellboy]

You know what, Dellboy? I was able to duplicate what you are seeing with the straight-ahead diffuser shot. It occurs when the flash is zoomed in the 50 to 105 positions (either manually on the flash or automatically with the lens). I don't understand why that is but I'm certainly going to try to find out. With through the lens metering, this shouldn't be happening.

Many thanks, Leo
I have done quite alot of internet research on this and the only thing I came up with is that the ETTL - system is heavily weighted by its suject to camera distance ie the camera/flash "knows" how much flash is required for that distance and it just puts that amount of flash out. It of course doesn't know if there is a diffuser there or not.

[Curtis N]

This is why I don't like ETTL II except as a fill in flash.It would be interesting to see someone try the same experiments with an auto-thyristor flash to see if it faires any better.

[PacAce]

I think I've narrowed it down a little bit more. I'm almost certain that it has to do with the distance information transmitted by the lens to the camera. Here's why I think that. If I set the flash to manual zoom at 70mm, it underexposes. However, if I press on the button to move the head to a bounce position and keep it pressed while still keeping it pointed straight ahead, the flash exposure with the diffuser seems OK, or at least not as underexposed as before.

I'll have to try this with a lens that doesn't report distance to see if my hunch is correct. Unfortunately, I don't think I have such a lens anymore.

Curtis, I have an auto-thyristor flash but that test would not be valid because with non-TTL flashes (i.e. auto flashes), you have to position the head at 45° or more. Otherwise, the feathering of the light from the diffuser may reach the auto sensor and cause the shot to underexpose.

[PacAce]

I've just done a bit of testing myself, and I'm getting some variations that aren't making sense. I can get consistent exposure from one shot to the next if I don't change anything, but when I change something that shouldn't make a difference, such as ISO, I get different results (with exposure confirmation on the flash unit in all cases).

I'll post these two, not as conclusive evidence of anything but to indicate that apparently there's more to this issue than meets the eye.

20D & Sigma Super, EF-S 18-55 lens (no distance info) focal length 55mm
Subject is 18% gray cards on black music stand, backdrop is white.
1/250, f/5.6, ISO 400
C.Fn. 14 set to 0 for Evaluative (not average) flash metering
0 FEC except the bottom image
Flash on E-TTL mode, zoomed to 85mm
BTW, Curtis, with the 580EX, 45° bounce exposure with or without the diffuser is perfectly fine (just tried it again @ 70mm on my puppets in my computer room). Not sure why the Sigma is exhibiting such a huge difference in exposure.

[PacAce]

OK, I think I've just proven that this "weird" behavior of underexposing when using a diffuser with a 580EX flash pointed straight ahead is definitely caused by the distance information that's used in calculating the flash exposure in ETTL mode. When I use the Sigma 50 f/2.8 EX Macro which I doubt passes the distance info to the camera, straight-ahead exposures with and without the diffuser is almost identical. No disernable underexposure can be seen in the image or in the histogram.

Now, what bearing the distance info and the use of a diffuser has on the ETTL calculated flash exposures has me completely baffled, to say the least. :confused:

Anyway, it's late so it's time to hit the sack and get some shuteye for me now.

[tim]

So ETTL 2 using distance information could be thought of as a bug. Hmmm. Best to bounce if you diffuse, or just point the flash up or down a little so it ignores that information.

[Dellboy]

So ETTL 2 using distance information could be thought of as a bug. Hmmm. Best to bounce if you diffuse, or just point the flash up or down a little so it ignores that information.

That's the way I feel about it.

I just don't need to be thinking about what lens I'm using, How much ambient light there is and if my flash is pointing in a compatable direction, on top of everythink else, certainly if the action and conditions are changing fast.

I just don't think this is good enough from Canon, I would dearly love a function for the flash to ignore the distance info and still retain the other benifits of ETTL II.

[tim]

I'm pretty sure only my 70-200 IS returns distance info, so i've never run into this problem. The 50mm F1.4 and Tamron lenses don't, they're the ones I use most.

[René Damkot]

Interesting read... I have no lenses that return distance info I think, but if I get my hands on one, I'll give it a try...

[PacAce]

Well, I guess this just shows that a diffuser, at least the Stofen Omnibounce diffuser, is best used in bounce mode. ETTL then won't get confused with what to do with the distance info as it'll just disregard it in bounce mode. I, personally, have never used the diffuser in a straight-ahead position as I've always thought that doing so defeats the purpose of the diffuser and just wastes power, but now I have another good reason not to. :)

BTW, besides underexposing the shot when using the diffuser in the straight-ahead position, ETTL-II also changes the color temperature used in the shot. That's the reason they come out looking so warm or yellowish. Just underexposing a shot by a stop doesn't ordinarily change the color temperature so ETTL-II is doing something with that as well.

[Curtis N]

I, personally, have never used the diffuser in a straight-ahead position as I've always thought that doing so defeats the purpose of the diffuser and just wastes power, but now I have another good reason not to.I will concur with that. My experiments with the Sto-Fen showed that almost all of the light came through the front panel and very little through the sides, even when the flash head was zoomed to 28mm. It's not going to throw much light on the ceiling if you point it straight at your subject, and will give results similar to using the wide panel built into the flash.

[PacAce]

I will concur with that. My experiments with the Sto-Fen showed that almost all of the light came through the front panel and very little through the sides, even when the flash head was zoomed to 28mm. It's not going to throw much light on the ceiling if you point it straight at your subject, and will give results similar to using the wide panel built into the flash.
...but with an uglier yellowish color cast. :(

[Curtis N]

From what I've read, the color temp info from the flash unit is only used if the camera is set to AWB or Flash WB, so there are several ways around this. To be honest, white balance is usually the furthest thing from my mind when I'm shooting, since it's generally adjusted later anyway.

[burchy]

Guys, I’m hoping to might be able to add a bit of clarity for me. As hinted at the start of this thread I’ve since upgraded my body to a 30D (yippee) and as a result ETTLII.
For the purposes of this discussion assume that I’m talking about shooting indoors, an external 550ex bounced vertically off the ceiling with white card attached (or flip-it – same difference), where I’m relying on the flash being the main light source:

I discovered with my 300D that the only way to get reliable results in this scenario is to set the camera to 200-400 ISO in manual at around 100th sec, f5.6ish. When I tried this in Tv or Av, I wasn’t getting the same results, in fact the subject was only getting partially lit. I had convinced myself that this was because the 300D changed its metering mode to centre weighted average in manual mode, as opposed to evaluative in Tv/Av modes and as a result the metering was concentrated around the subject, hence giving better exposure with the flash. I lived a happy life with my 300D/550ex with these assumptions…

Now I’ve got my 30D, I have total control over the metering in manual mode, i.e. evaluative, centre weighted average, partial and spot. I put the 30D in manual, using usual settings and shot away with really good results. It wasn't until a while I noticed I was in evaluative metering, strange I thought, as this is what I assumed was cocking up my 300D in Tv/Av modes. Tried putting the 30D in centre weighted metering (as per 300D) and I honestly can’t tell any difference.

Is this because ETTLII is subject based, rather than AF point based? Or am I way, way, off the mark?

Any help will be appreciated, even if it’s “there’s another thread that explains all this”

[Curtis N]

As the cameras meter ambient and flash separately, they can also use different metering modes for each.

With the 300D as you mentioned you're stuck with either evaluative or CW average for the ambient portion, depending on the mode dial setting. However the flash metering will be evaluative unless you use the FEL button (*), in which case it uses partial metering with flash.

With the 30D you have more control over both. You choose a metering mode for the ambient which is pretty straight-forward (evaluative, CW average, partial or spot). For the flash metering, Custom Function 14 lets you choose between evaluative or average, and the FEL button again is the way to use partial metering for flash.

Now I'm not sure if these options make it easier, as there is plenty of evidence in this thread that there is still a lot of mystery with E-TTL. When using flash as the primary light source, I normally set custom function 14 to 1 (average). My limited experimentation with it suggests that it gives more predictable results, though adjustments to FEC will still be required depending on the scene.

[burchy]

Thanks Curtis, this has cleared it up nicely in my mind. I will do some experiments setting the flash metering to average to see if it sways my preference.

One thing I haven't quite got right in my mind was why with the 300D did Tv/Av only partially illuminate the subject, yet Manual gave a lot more illumination. i.e. switching between Evaluate/SW average for ambient on 30D didn't seem to make much difference, so why did it on 300D?

Just a side note for interest, with exactly the same settings and bouncing the flash vertically the 30D needed +2/3 FEC to achieve the same histogram as the 300D did with +1 FEC. Not relevant I know but it interested me, could be an ETTL/ETTLII thing.

[PacAce]

Now I’ve got my 30D, I have total control over the metering in manual mode, i.e. evaluative, centre weighted average, partial and spot. I put the 30D in manual, using usual settings and shot away with really good results. It wasn't until a while I noticed I was in evaluative metering, strange I thought, as this is what I assumed was cocking up my 300D in Tv/Av modes. Tried putting the 30D in centre weighted metering (as per 300D) and I honestly can’t tell any difference.
When you say you switched the 30D to CWA mode, do you mean with it still in Manual mode? Or did you try it in Av and Tv modes? If you try it in Av or Tv mode, I think you'll find that the results will be very similar to what you got when you were shooting the 300D in Av or Tv mode.

The reason the subject gets less light in Av or Tv mode is because the camera will be exposing for the ambient lighting and fill in with the flash as needed. Whereas, in Manaul mode, the flash is the primary light source and provides most of the illumination for the subject. And, in this case, you may notice that the background isn't as well lit as when shot in Av or Tv and that can make the subject seem even more brightly lit in Manual mode.

The difference between FEC +2/3 and FEC +1 may not really be all that significant. It could just as well be attributed to little tolerance differences in the setup (during assembly) of the flashes and/or the cameras. I read a post in another forum where a guy was wondering why his two 1DmkII camera were producing differently exposed images (not dramatically different but noticeably so) although they both had identical settings in manual mode.

[burchy]

The reason the subject gets less light in Av or Tv mode is because the camera will be exposing for the ambient lighting and fill in with the flash as needed. Whereas, in Manaul mode, the flash is the primary light source and provides most of the illumination for the subject. And, in this case, you may notice that the background isn't as well lit as when shot in Av or Tv and that can make the subject seem even more brightly lit in Manual mode.

I didn't try the 30D in Tv/Av mode, but I certainly will.

With reference to your above comment, is it actually documented anywhere (i.e. in a Canon manual) that in manual mode the flash acts as the primary light source and in Av/Tv it's configured for fill?

Thanks for the response

[PacAce]

I didn't try the 30D in Tv/Av mode, but I certainly will.

With reference to your above comment, is it actually documented anywhere (i.e. in a Canon manual) that in manual mode the flash acts as the primary light source and in Av/Tv it's configured for fill?

Thanks for the response
I personally have not seen any official documentation from Canon specifically stating anything about the flash being a primary or a fill light source, nor do I expect to. However, it can be inferred from documentation that Canon does put out, such as the Canon EOS System guide (www.canon.co.nz/ftp/eos%20system%20guide.pdf) in the section on "Speedlite Technology" for most applications.

But you can't generalize that comment (re M = primary and Av and Tv = fill) thought, without elaborating on it since it doesn't always apply when shooting in M or Av or Tv mode.

For example, if you put the camera in M mode and set the aperture to f/8 and shutter speed to 1/250 and you are outdoors at night or in a dimly lit room, you can be very certain that about the only light that's going to register in your shot is that coming from the flash. So, in this case, the flash is the primary light source.

However, if you are outside in the middle of the day and set your camera in M mode to f/8 and 1/400 ISO 100, so that the viewfinder meter needle is centered, and set the flash to FEC - 1 (to just lighten up the shadows on the subject), then the flash here is being used for fill while the ambient lighting (the sun) is the primary light source.

And similar examples can be given when shooting in Av and Tv. But, for the most parts, when you shoot in Av and Tv, the camera will always set the camera to properly expose for the ambient lighting and just use enough flash to provide fill if the subject is also lit by the ambient light.

However, for example, if the ambient light happens to be coming from the sun that is setting in the background and the subject's back is to the sunset, then the subject will be backlit and dark. In this case, the flash will provide enough light to properly expose the subject and would be considered the primary light source for the subject although, obviously, it not doing anything to light the background.

[Curtis N]

The only "official" documentation I have seen from Canon regarding primary vs. fill flash is on this page of their "Flashwork" online brochure (http://web.canon.jp/Imaging/flashwork/functions/reduction/index.html) which mentions automatic flash reduction.When shooting in bright settings, or using daytime synchronization flash to shoot extremely bright subjects, automatic flash reduction prevents overexposed "blowout" by automatically reducing the flash level. This function, available when using EX Series Speedlites on EOS cameras, essentially determines whether the flash should be used as the main or auxiliary light source.I suppose you could do some testing by taking a shot in Av mode and then switch to M using the same settings and see what you get.

[burchy]

Thanks for the insight guys, extremely useful as always. I'll be doing some experiments with the 30D this weekend, which will be a whole lot easier now I know what to expect.

Si.