Some questions to Canon on how ETTL works

[digital]

OK Canon. I have been in many threads and read quite read a lot of your documents and others work like the great Syl Arena.  All types of tests and theories.  It really does not make a lot of difference to me as I view ETTL as a good tool in a busy mobile situation but I know I will have to use FEC. Sometimes I can view a scene an predict FEC required and sometimes my predictions don’t work.

I’d like to hear from Canon but others are more than welcome to contribute.

The 3 main questions. 

1. Using a lens that does not provide distance data. We know that if you take a shot of a bride alone in a white dress, a groom in a black tux alone and the bride and groom together all 3 scenes will reflect light back differently during the pre-flash.

Does the flash system just rely on light reflected back from the pre flash to determine subject distance much like a flashes guide number only or the does the flash exposure system try expose the subjects based on the scene? Kind of like the cameras metering system. Or both?  NOTE: I realize the flash does not use the cameras light meter but does use the system to determine exposures. 

2. Using a lens that does not provide distance data. Does the system compensate for a diffuser?

Part a

First off the system has now way of knowing a diffuser was installed on the flash. If the answer to question #1 is it measures distance only by reflected light like a flash guide then it would not compensate. If it takes the actual scene (subject/s) into account then I would think it would compensate. I know Canon says it does compensate but we need more detail on what actually happens. Also we know that when the flash is evaluative mode it tries to isolate the subject/s by comparing ambient to the pre-flash. Again – distance or scene?        

Part b

If so then it should compensate when you bounce the flash. If it just measures distance like the guide number then it would not be able to compensate? More detail on how the system handles bounce flash would be helpful.   

3. Using a lens that provides distance information. Which takes priority? Pre-flash, distance or both? I just came from another forum (POTN) where a person has done this test with all his lenses. When he placed a diffuser on his flash and shot direct the images were underexposed.  When he tilted the flash up or to the side (whihc disables distance info) the exposures were good. Seems to me that distance info takes priority when available.

Well Canon explain this to us so we are better prepared when using our flash systems. Anything else yo can add would be greatly appreciated. 

Thanks in advance. 

[p4pictures]

Since the flash emits the pre-flash at a known power level (1/32 is often quoted) then the camera sees what areas of the frame are illuminated by the flash and uses them to evaluate the correct exposure. White subjects cause it to under expose and black subjects cause it to overexpose just like they do with normal ambient meter readings. E-TTL II makes use of available distance info as part of it's programming.

If it's really bright & sunny then the pre-flash is not seen by the camera and the camera thinks that it needs to max out the power for the exposure resulting in way overexposed flash subjects. You can improve on this by ganging up multiple Speedlites in the same group, in effect making the pre-flash more visiible.

The flash / camera cannot compensate for a diffuser like a stofen fitted over the flash head - it doesn't know it's there. So this as you say can lead to under exposure. Bouncing, twisting or taking the flash off-camera switches off distance info part of the E-TTL II calculation so this then takes in the effect of the diffuser.

So E-TTL II flash calculation uses several factors to determine the exposure.

1. The brightness of the subject lit by the pre-flash compared to it's brightness lit by ambient light

2. Lens distance information (if the flash is on camera and pointing straight at the subject and not a master)

3. Some kind of detection of 100% flash reflection, eg off mirrors to discount these from the metering

Depending on the camera the weighting of the distance info is different. There was something in the EOS-1D Mark IV where the flash algortithm was tweaked to place more value on the distance info than the previous cameras. Some working pros found that for red-carpet work the Canon flash became much more consistent than ever before with that camera. Unfortunately there is little public information from Canon about how E-TTL II really works and the differences between each camera models implementation of E-TTL II.

Brian / p4pictures

[digital]

You replied

"The flash / camera cannot compensate for a diffuser like a stofen fitted over the flash head - it doesn't know it's there. So this as you say can lead to under exposure. Bouncing, twisting or taking the flash off-camera switches off distance info part of the E-TTL II calculation so this then takes in the effect of the diffuser".

I believe the flash has no idea the modfier is there and does not compensate but 50% of people on vaious forums think it does and I have read many threads on the subject. You may as well ask if a filter is neccesary to protect your lens .

That is why I'm curious to see how Canon responds.     

Thanks for the great feedback.       

[Mykolas]

Hi Digital,

I have tried my best to answer your three questions in order below:

1. The pre-flash is fired, then what is reflected back is compared with an ambient meter reading to determine the best exposure.  Distance information is not determined by this process.  It is only factored into the overall E-TTL II algorithm for fine tuning the exposure if the lens you are using provides this information.  Essentially, the overall scene is what will be used entirely when you are not using a lens that provides distance information.

2.  When you are using a diffuser on the flash, the camera has no way of knowing that.  Since the pre-flash is being compared to the ambient meter reading to determine how to best expose the scene and the subject, I would say that using a diffuser will effect this process anyway.  It will change what is being reflected back to the camera from the pre-flash, thus changing what is being compared to the ambient meter reading.

3.  As I noted earlier, the distance information that is provided by a lens that supports this is added into the overall calculation, so this explains why the exposures were a bit different in the example you provided.

Not knowing if you have seen what we have on our own Canon Flash Work website about E-TTL II technology, as well as other flash technology, I am providing a helpful link below:

http://web.canon.jp/imaging/flashwork/ettl2/technology/index.html

This site provides a wealth of information about how our flashes work together with our cameras.

To conclude, I would like to note that the exact E-TTL II algorithm is not published information, so I tried to answer your questions using the best of my knowledge and experience with E-TTL II.  I hope that you find this helpful.  Please do not hesitate to ask any additional questions if you have any.

If this is a time sensitive-matter, additional support options are available at Contact Us.

[digital]

Thanks Mike. I appreciate it. I'm sure other members do as well and persnally I found no surprises. 

Thanks to the link I will check it out. Here are a few more links I found helpful if anyone is interested.

http://cpn.canon-europe.com/content/education/masterclass/canon_flash.do

http://photonotes.org/articles/eos-flash/