70D bounce flash

one
Contributor
Hi,

Recently I upgraded my 650D to the 70D and I noticed that the pictures are a lot darker (underexposed) when bouncing the flash. I have tested this with several other cameras, the 600D, 650D and the 700D using the 430EXII and 580EX all with the 18-135 STM. All but the 70D resulted in acceptable exposure.

Is there a setting I am missing? I used both the [A+] and P setting.
47 REPLIES 47

No resolution to my knowledge...  😞

 

It's the only thing that I don't like about the 70D.  Really drives me nuts. 

I recently experience the same issue on Mar 22, 2014. I spoke with Canon Canada and did not achieve much. I decided the camera is defective and returned it.

 

I bought another 70D on Apr 25, 2014, which shows exactly the same behavior. I spoke with Canon again, this time I sent them pictures showing the difference in exposure between XSi and 70D. I did not get much of a reply, so I returned the second camera too.

 

I but I am still taking to Canon. I got them to acknowledge that there is something going on with the bounce flash, although in their experiments the results are not as severe as the one I observed. My call was recently escalated to the next level - people that I cannot talk directly with, but my Technical Support Representative will talk to on my behalf.

 

Here is a short summary of my findings so far and how to reproduce the issue:

 

1. Make sure you eliminate the ambient light from the experiment. How? Go manual mode, fix the ISO to 200 or 400, set the aperture to 5.6 and the speed to 1/250 (use 1/200 if you are comparing to XSi or to other camera that cannot normally sync above 200). Take a shot without the flash. The picture should be really dark, or you have too much light in the room. Why do you need to do this: because with the default settings in P mode and with bounce flash the camera will select ISO 1600, shutter speed 1/60 and the max aperture of the lens (F4 for example). With those setting you will get a dissent exposure without any flash. When I tried with 28 F/2.8 lens this is exactly what happened - the camera went to F4, it did not even need to be 2.8 and I got a perfectly exposed shot. To make things worse a lot of lenses are actually shower than what they tell you - check DXOMark and their Transmission graph to find out more. This can explain why some lenses work and some does not - sometimes the lens is more than 2/3 EV slower than it claims.

 

2. Now that you know how the picture looks like without the flash turn on the flash and take another shot. Whatever difference you see is the light from the flash. If you do the same with XSi you will get very reasonable picture. With 70D your results will vary and depend on the type of the flash (more specifically how low the flash power can go), the distance from the subject and the ceiling.

 

3. Now switch the flash to manual and try the picture with full power. That will prove to you that the flash has more than enough power to give you the proper exposure.

 

4. Go back to TTL, put a gray card in the scene (if you did not have it already, it is a very easy way to measure the underexposure) and try a FEL on it (focus, put the gray card in the middle of the frame, hit FEL, recompose, take the shot). This proves to you that the camera is fully capable of measuring the proper exposure and communicating with the flash.

 

The explanation that I have for the behavior is this:

 

It used to be that under these circumstances the flash was becoming main light. Now the flash is always used as fill light only. May be that makes sense to someone in Canon since at ISO 3200 or 6400 you have enough light (at home or in the office) to shoot with the available/ambient light, so you need flash for fill only. But to have enough light you need the high ISO and a slower shutter speed. I personally don't like that. It should be the photographer's choice if the flash is main light or fill light - may be you don't like the quality of the available light or you need to stop the motion through either high shutter speed or the short light pulse from the flash. Those use cases are still valid and I don't see them going away.

 

One way to prove the flash is only fill is to put a color gel on it. You will observe that the color tones change just slightly in the shades and the colored light is visible mostly in the very dark shadows. If you manage to get all of your picture in the gel color this means you are at the point when the flash just cannot go lower in power. You can verify that by switching to manual and setting the minimum power the flash allows (typically 1/64 or 1/128).

 

In all fairness it appears that the result depends on the lens more than it depends on the flash. In a experiment that I did just once at a camera show I found that with the stock lense (18-135 STM) I had a successful bounce flash with 3 difference flash models. I kept everything exactly the same but switched the lens to my 55-250 IS and the picture came out totally dark. I could not try it too many times, since the people I was bouncing the flash at were not very happy with me and I did not want to annoy them more than I needed to.

 

I still love the 70D body, but the bounce flash issue is a show stopper for me. I normally shoot with a remote flash triggered optically (non-TTL, digital slave mode) in manual mode. 70D was supposed to enable me to do TTL off camera flash, but instead I lost the ability to shoot with on camera bounce flash (I can still do the optical trigger, but that is not the point).

 

I will share any news I hear from Canon here. In the mean time can somebody re-try the experiment with 18-135 STM and 55-250 IS? Not that I will buy a new stock lens even if it works with it, but I still want to know.

 

On a side note I tried the bounce flash with 6D and 24-105 F4. That one worked too, although other threads are talking about exactly the same issue on 6D. I have seen the issue reported against 5D mark III too. The common thing between those cameras is the ability to use very high ISO (and the release dates), which makes the flash as main light way less important.

 

If someone has found a way of making the flash on these cameras act as main source of light, please enlighten us! I will buy my 3rd 70D immediately. Unless Canon decides to send me one for my contribution to this troubleshooting of course.

dg
Contributor

Had another conversation with Canon Canada - they escalated the issue to Canon USA and the answer is: the meter is confused by light coming from the viewfinder. Which is **. It is the one of the first answers that I received from Canon and I tried covering the viewfinder - does not change anything.

 

In the mean time I went to the store and tried a few pictures with different comibinations of both on camera and wireless TTL flash. Every time when the flash is pointing directly to the subject, either on camera or off camera I get proper exposure. Every time when the flash is bouncing from something, even if that is a wall 20cm from the flash, the picture is underexposed. One of the solid results is that with everything the same switching the flash to manual model and 1/2 power produces proper exposure, while +2EV FEC has practically no impact on the heavily underexposed image.

 

As a reference here is a thread about 580 EX II and 5D mark III showing the same issues:

 

http://www.dpreview.com/forums/thread/3222627

 

The last experiment that I will do before jumping the ship and buying Nikon D7100 is to try a real camera shop and asking the sales people to show me a working bounce flash with 70D.

 

Mod Note: Edited for language per forum guidelines.

dg
Contributor

I visited 2 real camera stores and asked about the underexposure. In the first store they guy spend the time with me, reproducing the problem with both my flash and the 430 EX II from the store, on a 70D body that they use as a demo. The clean experiment is like this:

1. Put the flash on manual, 1/2 power, pointed up (75 or 90 degree, with or without the bounce card)

2. Set the camera to M, shutter 1/200 and find a combination of ISO and apperture that provides proper exposure on the subject (about 1 m from the camera) - ISO 400, F/5.6 should be OK

3. Now switch the flash to TTL and you get a very underexposed picture


The gentleman told me he has never heard of this issue before, but obviuosly there is something. He suggests compensating for the overexposure, until hopefully Canon releases a firmware update.

 

In the second store I spend some value time with the guy in charge of accepting cameras for repairs to be sent to the manifcaturer. He denied any prior knowledge of the issue too. In his opinion the TTL is supposed to work only with the flash pointing directly at the subject, so the bounce should be expected to fail. He could not explain why this used to work with the old (XSi) body.

 

I tried to get some info from other photographers using the Canon system, but seems like the case of overpowering the existing light with the flash is not that common and when it is done it usually involves manual flash mode and softboxes. The TTL is used predominantly for fill flash.

 

This concludes my investigation - there will be no new Canon equipment in my camera bag. Unless I read somewhere that 80D magically fixed the issue, but I will not be holding my breath.

I have "read" of the issue with the 580EX II on a 5D III body, but I've loaned out my 580EX II to a friend ... a somewhat extended loan since I don't use it since getting the 600EX-RT units.

However the flash should (and normally will) always fire as a key light when flash exposure compensation is zero. If you want fill, you can cut FEC to -1 (or whatever your preference).

The person at the camera store who thought E-TTL shoudl only work when flash is pointed directly at subject is probably not aware of how the E-TTL system works.

With E-TTL (really E-TTL II these days) system, the flash and camera work together to identify the correct flash power. The camera meters the scene using evaluative metering -- but does nothing with the data other than saves it. The camera then triggers the flash at pre-flash power (by default it's 1/32nd power but you can change that default). As the flash fires, the camera meters the whole scene again. It compares the first evaluative metering to the second. It's trying to determine how much of a difference that 1/32nd power made and uses this to decide how much more (or perhaps less) it should use when the shutter is open.

Suppose there's a mirror in the background. On that particular zone (where the reflection shows up in a mirror) the difference will be enormous compared to all other zones. The camera is able to determine that this isn't true -- it's an anomaly -- and ignores it.

Likewise, suppose there's a lamp in the photo... in which case the lampshade emitting it's own light and the flash wont make much difference. The camera can detect that this is an anomaly as well and factor it out.

Old thyristor flash systems would have been fooled by those two cases -- but not E-TTL. This causes the E-TTL system to be a little smarter than previous systems.

There is one more factor... if (and only if) the flash is on camera and pointed straight ahead, then the camera knows it can use the lens focus distance to set flash power (since the flash has a known guide number). If the flash is not on camera then it has no idea how much closer or farther the flash is vs. the lens focus distance so it wont take lens focus distance under consideration for flash power. Likewise if the flash is on camera but is not pointed directly ahead then it has no idea how far away your bounce surface is located... nor how reflective it is. Consequently it is also forced to ignore lens focus distance.

But if flash is on camera and straight ahead then it CAN use lens focus distance. A 430EX II has a guide number of 43 meters (about 141'). If the lens focus distance is, say, 15', it's a very easy math equation to determine how much power to use on the flash (it also has to factor in focal ratio) to get a correct exposure.

If I understand you correctly, that's the only time it works correctly. This implies to me that it is correctly using math to work the flash power when it can rely on lens focus distance, but when it is forced to rely on metering alone it gets it wrong (does it matter *which* flash model is used? Because the meter seems to work correctly when there's no flash involved at all.)

The meter lives up in the viewfinder -- which is why if your eye is not looking through the finder (blocking nearly all light from entering via the viewfinder) then it is plausible that it would get a false reading from the extra light entering through the finder, resulting in the camera and flash underexposing the shot. It is, of course, trivial to just block the viewfinder and rule out that possibility.

Years ago, GM had a service complaint about a customer who had a car which "hated vanilla ice cream". This was baffling. The customer would stop on his way home from work to pick up ice cream at a local ice cream parlor that made their own ice cream. If the customer bought vanilla, the car wouldn't start when he returned to the car. If he bought a different flavor, the car started. A bizarre case. GM sent engineers to evaluate because this one really had them stumped.

Turns out the car had a vapor lock problem and the ice cream store had quarts and half gallons of vanilla ice cream flavor (and only vanilla) prepacked and available in the freezers at the customer counter. If the customer wanted any other flavor, an employee would have to hand-pack the ice-cream container and this took more time. The "more time" was just enough for the car to cool to the point that the vapor lock was no longer a problem.

So while wierd things do happen, there's often a rational explanation... the trick is trying to figure out why it happens. If Canon says light is entering through the viewfinder (a reasonable guess) it's easy to rule it out by blocking the viewfinder from any stray light.

Canon does claim their 70D is compatible with their current line of speedlites -- so it *should* work and there may indeed be a firmware problem.
Tim Campbell
5D III, 5D IV, 60Da


dg wrote (sometime in June):

I visited 2 real camera stores and asked about the underexposure. In the first store they guy spend the time with me, reproducing the problem with both my flash and the 430 EX II from the store, on a 70D body that they use as a demo. The clean experiment is like this:

1. Put the flash on manual, 1/2 power, pointed up (75 or 90 degree, with or without the bounce card)

2. Set the camera to M, shutter 1/200 and find a combination of ISO and apperture that provides proper exposure on the subject (about 1 m from the camera) - ISO 400, F/5.6 should be OK

3. Now switch the flash to TTL and you get a very underexposed picture


The gentleman told me he has never heard of this issue before, but obviuosly there is something. He suggests compensating for the overexposure, until hopefully Canon releases a firmware update.

 

In the second store I spend some value time with the guy in charge of accepting cameras for repairs to be sent to the manifcaturer. He denied any prior knowledge of the issue too. In his opinion the TTL is supposed to work only with the flash pointing directly at the subject, so the bounce should be expected to fail. He could not explain why this used to work with the old (XSi) body.

 

I tried to get some info from other photographers using the Canon system, but seems like the case of overpowering the existing light with the flash is not that common and when it is done it usually involves manual flash mode and softboxes. The TTL is used predominantly for fill flash.

 

This concludes my investigation - there will be no new Canon equipment in my camera bag. Unless I read somewhere that 80D magically fixed the issue, but I will not be holding my breath.


I have no dog in this hunt and little or nothing to contribute to the technical discussion that has taken place over the past several months. I do, however, have two points to make:

 

1.  The term "TTL" has been bandied about freely in many of the articles posted. Modern Canon cameras use E-TTL (Enhanced TTL), rather than the original TTL. They are not at all the same. If you set a Canon flash to TTL on a camera that expects E-TTL, you will get bizarre results. The flash always fires at full power, and most pictures are dramatically overexposed.

 

2. The guy in the second store is wrong. My experience with several E-TTL cameras (50D, 7D, 5D3) is that E-TTL's fill flash fetish works much better (or does much less harm, depending on your point of view) with bounce flash than with direct flash. Why that is, I'm not really qualified to explain, but it must have something to do with the fact that by the time the highlights are fully exposed, the rest of the scene has been illuminated as well. If this isn't the case on the 70D, I think it does support the contention that there's something wrong with the way the 70D calculates its desired flash power (or possibly in the way it communicates that calculation to the E-TTL flash unit).

Bob
Philadelphia, Pennsylvania USA

Bobn4burton, I add my kudos for your hard work trying to find a solution to this issue.  I'm new to the discussion, but I agree that this is a 70D problem (an otherwise wonderful camera).  As a new 70D owner having the same problem with underexposure of shots while bouncing the flash, I was stumped about the problem until I read the threads at DPREVIEW.  All the suggestions about operator error and to try this and try that are just noise.  The problem does not exist with the 430EXII with my XSi in the same conditions using the same lens.  The FEL button fixes the problem, but that indicates that something is not right with the normal (non-FEL) flash sequence.  The camera is just not calculating the correct primary flash power required to get a correct exposure.  It’s not even close (several stops underexposed).  I do hope Canon addresses the issue.  

Hello socal13,

There have been several answers, tests, suggestions and the like represented in this thread.  Even you yourself seem to have found a work around that maybe works for your needs.  That being said, this is not a known issue within Canon regarding the EOS 70D. 

If you feel as though your camera would benefit from the expertise of a service technician, it's urged that you set up a service evaluation with Canon.  That can be done through our web site, linked here.  Often times issues can only be truly understood under factory conditions, as there we could ascertain the true cause of this issue, if there is a malfunction with your particular camera.  

Did this answer your question? Please click the Accept as Solution button so that others may find the answer as well.


@Tim wrote:

Hello socal13,

There have been several answers, tests, suggestions and the like represented in this thread.  Even you yourself seem to have found a work around that maybe works for your needs.  That being said, this is not a known issue within Canon regarding the EOS 70D. 

If you feel as though your camera would benefit from the expertise of a service technician, it's urged that you set up a service evaluation with Canon.  That can be done through our web site, linked here.  Often times issues can only be truly understood under factory conditions, as there we could ascertain the true cause of this issue, if there is a malfunction with your particular camera.  


First, thanks for responding.

 

Although your response is horribly depressing 😞

 

This SHOULD be a known issue with Canon.  There are many many people that have confirmed the problem.  It's not hard to replicate.  So basically there is probably no hope for a firmware fix for this problem...

 

Using my 70D indoors with bounced flash is pretty much useless for candid shots...  I guess I'll be looking to sale this camera to replace with something else.  Too bad because other than this issue, I've loved the camera.


@bobn4burton wrote:

@Tim wrote:

Hello socal13,

There have been several answers, tests, suggestions and the like represented in this thread.  Even you yourself seem to have found a work around that maybe works for your needs.  That being said, this is not a known issue within Canon regarding the EOS 70D. 

If you feel as though your camera would benefit from the expertise of a service technician, it's urged that you set up a service evaluation with Canon.  That can be done through our web site, linked here.  Often times issues can only be truly understood under factory conditions, as there we could ascertain the true cause of this issue, if there is a malfunction with your particular camera.  


First, thanks for responding.

 

Although your response is horribly depressing 😞

 

This SHOULD be a known issue with Canon.  There are many many people that have confirmed the problem.  It's not hard to replicate.  So basically there is probably no hope for a firmware fix for this problem...

 

Using my 70D indoors with bounced flash is pretty much useless for candid shots...  I guess I'll be looking to sale this camera to replace with something else.  Too bad because other than this issue, I've loved the camera.


That thud sound you hear is canon burrying its head in the sand. "Nope no problems down here with focus or flash."

 

They dont need a customer copy to repeat this. Grab just about any 70D  laying around canon repair. stick a canon flash on the top and use it in bounce. It underexposes mine does the same thing. I generally run manual flash so its not such a big deal, but it should still work right for the people that use it in ettl  mode.

 

It's either

 

1. A Canon delvopment issue, not testing it when they designed the firmware.

2. It was designed to under expose and only act as a fill light for what ever bad reason.

3. Its just another defect in 70D's

 

It doesnt really matter which of the 3 it is. If they would just pick one and own it.

Avatar
Announcements