Canon 400D onboard Flash problems...

04/28/2007 11:50:52 PM · #1

I've been having problems with the onboard flash since getting my camera, and I think I've just worked out what it might be...

Flash photos seem to come out OK at around ISO400 or 800 only. At ISO 100, the flash is too weak. At ISO 1600, the flash is too bright. I thought TTL flash was supposed to be intelligent, but it's not taking ISO into account. I can push the flash EV depending on the ISO, but that's buried away in a menu, so taking multiple exposures for every flash photo, and adjusting the EV each time is really not worthwhile.

In the end, what I want to know, is this user error, or a camera problem? Do I need to get it looked at, or do I just need to learn how to use it, and maybe only ever take flash photos on full manual and adjust the EV to suit each shot. Or always use ISO 400 for flash shots?

Thanks for your thoughts.

04/29/2007 12:11:53 AM · #2

Two questions about what you're doing when you're using the flash:

1) What camera mode are you in, and 2) is the ambient light so dim that your on-board flash will be the "main light", or is the ambient light bright enough that the flash is merely supposed to be fill light?

04/29/2007 12:13:41 AM · #3

Also, how about some sample pictures to show what you were taking. And when you upload, post the stats as well (ISO, shutter, aperture).

04/29/2007 12:35:04 AM · #4

OK, I will do this. For the shots where I'm having the worst problems, flash is the main light. There is other light around, but not enough to use a reasonable shutter speed, so I pop the flash.

I'll try and upload some samples soon.

xantangummi

Canon EF 135mm f/2.0L USM

04/29/2007 12:39:46 AM · #5

which aparture do you use?

Message edited by author 2007-04-29 00:49:06.

04/29/2007 12:43:40 AM · #6

Originally posted by surfdabbler:

OK, I will do this. For the shots where I'm having the worst problems, flash is the main light. There is other light around, but not enough to use a reasonable shutter speed, so I pop the flash.

Okay, so when you do this, what mode is your camera in?

(M, Av, Tv, P, etc)

04/29/2007 08:17:33 AM · #7

OK, here's a couple of pics...

ISO 100

ISO 400

ISO 1600

All taken in P mode, with flash. All taken at 1/60s, F4.5, except the ISO 1600 shot, where the camera closed up to f/5.0. Just for the record, A and T modes were all picking the same exposure values.

Shots without flash were coming out at .8s for ISO 100, or 1/6s for ISO 400. I was just wanting to use the flash to fill-in and to speed up the shutter, but the actual shot was darker with flash.
(EV and flash ev both set to zero)

I tried to create some more controlled shots to show this behaviour, but couldn't get it to happen as convincingly as the earlier shots. (It's night now, so I can't get the same scene.)

I've tried a lot of different flash stuff with this camera, including diffusing and bouncing the on-board flash, and I got some nice results in a few shots, but I'm still really struggling to get consistent reliable flash results with this camera, so I'd appreciate your thoughts.

P.S. I realise that this discussion is in the wrong forum. If SC wants to move it, that's fine with me, but I don't know how to move it myself . :)

04/29/2007 09:56:08 AM · #8

Okay, here's my guess:

The one thing that fools eTTL is a highly reflective surface. eTTL works by sending out a pre-flash, measuring how much light comes back, and then opening the shutter and firing the real flash. But if you have a highly reflective object in the frame, the eTTL system will see too much light coming back and will greatly reduce the amount of flash.

Now, you may not notice it at first, but take a look at the chair the girl is sitting on. The reflection of your flash off the chair is nearly white. That means the flash pumped enough light to get a nearly white light coming back at it. So it probably reduced the light.

Try your test again, without any reflective surfaces in it, and see if you get better results. (again, this is just a theory as to what the problem is at this point based on what you've shown, so we're looking to test the theory and if it's wrong, figure out what the real answer might be)

By the way, I'm not sure how you're trying "bounce flash" with an on-board flash, but depending on the angle of the bounce, that could solve your problem as well by keeping the direct reflections down.

04/29/2007 07:39:29 PM · #9

Yes, interesting idea. I've heard that. I looked through the whole image. At ISO 100, the highest exposure I could find in the whole image is 252,250,251, and that's in the catchlights in her eyes. I would hope that catchlights in eyes aren't big enough to cause problems with TTL! :) I don't want to be limited to taking photos with closed eyes.

The highest reflection on the chair is 216,214,217, but we're talking no more than 2-3 pixels anywhere near this level. Even in the flash reflections, it's hard to find pixels over 200. They only look white compared to the rest of the image. Looking at the histogram, the blue and green peter out at about 50%, and the red struggles to reach 70%, and there's really nothing over that. Certainly no detectable spike to fool the TTL.

For comparison, in the ISO 1600 shot, there are a few of the flash highlights that hit 255, and there are significant areas (the window in the background, and a few other natural light reflections) that also hit 255.

(By the way, when I've been playing around with bouncing and diffusing, I've just been holding a mirror or a piece of paper in front of the flash. Very educational, and sometimes gets some surprisingly nice results.)

Message edited by author 2007-04-29 19:41:01.

04/29/2007 09:12:44 PM · #10

So go ahead and "test the theory". Try your your varying ISO tests again but without a reflective surface in the frame. And no, I do not expect the highlights in her eyes to be big enough to cause the problem, the reflections on the chair are far more likely to be the culprit. And they don't have to be 255/255/255 .. the camera wants to see GRAY remember?

The problem with on-board flash is that the light bounces right back at you. The further away from the lens you can get the light source, the better. That's why the external flashes are so tall. But even that is not enough. I use a flash bracket to put my flash even further away from the lens access.

04/29/2007 11:16:23 PM · #11

Ah, now you're just making me work. :) A photo without any reflections is going to have to be a totally different location. I will think about whether I can find a similarly lit location without any reflective objects. I will cover the chair with something and try in the same location.

I tried some tests last night, with and without any artificial lighting. Those relying totally on flash were all consistently slightly darker than I would want. I couldn't recreate the same variation in results by varying the ISO. With no ambient light, the ISO made almost no difference to the exposure. However, even on a tripod taking shot after shot, there was a quite noticable variation in exposure from shot to shot, regardless of ISO. Also interesting to note that I was using the same chair in most of the test shots. Taking the chair out of the shots made no difference.

Also, the same highlights are present in the ISO1600 photo at even higher levels, yet the ISO1600 photo is slightly overexposed.

Message edited by author 2007-04-29 23:16:58.

04/30/2007 01:19:03 AM · #12

OK, I think I've found the culprit...

I just tried the same shots at the same time of day, same settings (ISO100, F4, 1/60s) with a teddy bear sitting in the chair. No problems at all. My daughter got home from Kindy, so I got her to sit in the seat. Problem. Put teddy back. No problem. Put daughter back. Problem. It's HER!!! :)

OK, teddy bear was brown, so the next test was put teddy bear with sheet of white paper on stomach. Underexposed. White piece of paper on bear's head. No problem.

I also tried it with a piece of yellow paper. Sometimes it was way underexposed, other times, it was fine.

The camera is set to evaluative metering, but the flash metering seems to be very much based on the centre. I've read something about the TTL metering from the auto-selected focus points. Is this true? Can this be overridden?

What's really annoying is that the EV adjustment doesn't do anything if the flash is on. Wouldn't it be nice if when the flash was up, the EV would adjust the flash EV? Canon don't actually use their own cameras, do they? :)

04/30/2007 01:28:35 AM · #13

Interesting results.

Now try this... Set your camera to M mode. Set the aperture to what you want. Then move the shutter speed to a point where the view finder says that you are under exposing by -1 to -1.5 stops. Now, with the flash up, take the picture. What do the results look like?

04/30/2007 01:30:54 AM · #14

I will try this tonight. My wife has my camera for the afternoon. :)

04/30/2007 01:37:14 AM · #15

One more test, go ahead and pick a shutter speed that totally under exposes the image and try that too. For example, if your meter reading says 1/4th of a second at ISO 100 and f/4, then try it at 1/125th with the flash up. The subject should be exposed the *same* way as in my "-1 to -1.5 stops under exposed" setting. But the background (where the flash doesn't hit it) will go dark.

For what it's worth, I use flash in two ways:

1) As fill flash when there is plenty of ambient light. I just set my camera to Av mode and dial in the aperture I want. Then I let the camera figure out the amount of fill flash to use.

2) As main light when the ambient light is too dark. In this case, I am almost always in M mode. I dial in the shutter speed and aperture I want, and I let the camera/flash system figure the rest out. It's fast and easy.

04/30/2007 01:41:38 AM · #16

Thank you for your detailed help and ideas on this. I really appreciate this. I will do these tests tonight and see what results I get.

04/30/2007 10:06:42 PM · #17

OK, here's some of my test images...

Firstly some teddy bear shots. All taken in P mode, all using f3.5, 1/60s, ISO 100...

Any light area near the centre of the photo is really underexposing the show shot, but move it off to the side, and it's fine. This does not seem to be related to the active focus point, as I took many shots with different focus points, both automatic and manual. It was also a little inconsistent. There were times when with exactly the same settings, the shots with the paper in the middle were exposing fine.

I also tried the M mode approach of dialling up 0,-1,-2, and using the flash to make up the difference. This worked well for shots where there was ample ambient light, and using as a flash for fill-in, or to slightly increase the shutter. Where there's really insufficient light, this method obviously creates very low shutter speeds, but still seems to expose OK.

Interestingly, I did some more experiments this morning, with white paper, and a different bear, and no problems at all...

Ok, it's slightly underexposed, but this is understandable with the subject, and can very easily be corrected in PP. More importantly, it was fairly consistent, with and without the sheet of paper.

Compare this shot to the earlier underexposed shots of the other bear.

Just to add more confusion, check these out...

These are successive shots 5 seconds apart, same settings (P mode, 1/60s, f4.5, ISO100), same environment, move the paper, HUGE difference in exposure, and the one with the paper in the centre is fine!

What is going on??? Stuff like this, I can't explain.

Message edited by author 2007-04-30 22:12:49.

04/30/2007 10:23:50 PM · #18

Your first set of images would seem to suggest the exposure is heavily weighted towards the center. With the light paper in the center, the rest of the image goes dark because the camera decides there's enough light. But when the paper is away from the center, the camera sees the darker light-absorbing bear and pumps up the light.

Your second set, with the white paper centered, also indicates that the center exposure is more important than even the background (dark or light) as it manages to light the animal with the same amount of light regardless of background.

So I dunno... is it working correctly? Yeah, I think it is. Is it what you want it to do? Perhaps not, if you can't predict how much light you're going to get, it's going to be hard to consistently light your subjects.

How about anyone else reading this thread? Your thoughts?

04/30/2007 10:41:35 PM · #19

Yes, I would agree with the first four photos showing the consistent very strong localised centre-weighted behaviour. (Not really ideal behaviour, but consistency is better than nothing)

However, the rest contradicts this, and I've got other examples that contradict this as well. As part of the middle set of photos, I have shots of the monkey against the light wall without the piece of paper, and they are exposed exactly the same as the one with the paper, so even having a big white sheet of paper vs dark brown fur in the centre of the shot makes no difference in that scenario.

I don't know if anyone else is following this thread. :) David, your help has been great so far. Even if I haven't found the problem, I've appreciated your excercises and thoughts so far. I've learnt a lot.

Anyone else out there?

SamDoe1

Tamron SP AF 28-75mm f/2.8 XR Di for Canon

04/30/2007 11:33:18 PM · #20

Is your camera set to center weighted metering? If so, that might cause the camera to meter the whole shot off the center. Also, have you changed any of the custom functions?

05/01/2007 12:12:49 AM · #21

Hi Sam,
Camera is set to evaluative metering, so from what I understand, this should look at the whole scene.

Interestingly, the icons really don't seem to represent what the metering modes do. From the manual, 'Partial' metering means spot metering, and it's icon is the surrounding lines ( ) with no spot in the centre!

As for custom functions, I hadn't changed anything ... but I have now! :)

You can change the SET button to be flash EV! Thank you for making me look there. I bemoaned the lack of good FEV adjustment on the 400D in another thread (and possibly in this one too), and here it is in the custom menu! This is great - that will certainly help LOTS!

There is also a custom function for ETTL II modes, evaluative or average. It's currently set to the default of evaluative. It has another option for average. The manual doesn't give a lot of details, but perhaps this will give a more consistent result for the whole scene, and with my new FEV button, I can then adjust this better for the situations where special handling is required. I will try this out.

Regardless of how this goes, you have no idea how excited I am just to have an FEV button now! :)

Elmakias

Canon EF 17-40mm f/4.0L USM

05/01/2007 12:37:03 AM · #22

im interested in what you are explaining, i don't really understand if though, could you break it down?

SamDoe1

Tamron SP AF 28-75mm f/2.8 XR Di for Canon

05/01/2007 01:14:10 AM · #23

Try setting the camera in the evaluative metering mode, the one that has a symbol that looks sorta like this: [(*)]. That should average the whole frame as opposed to just the center of it. The setting you have for the ETTL2 is right. I got this from another site:

""Average" turns off the reflectance adjustment, the distance feedback adjustment, and all other flash metering adjustments. ETTL-II will perform a straight average of all flash metering zones."

You can try it out and see how it works for you.

Originally posted by Elmakias:

im interested in what you are explaining, i don't really understand if though, could you break it down?

Was this directed at me or surfdabbler?

Message edited by author 2007-05-01 11:20:40.

05/01/2007 06:59:03 PM · #24

Aha, DPC is back online! I couldn't get here all yesterday afternoon and night. Anyway...

Yes, my camera has always been set in Evaluative for both the normal metering, as well as the TTL2. I still think it's worth testing out the average setting to see how it goes.

Elmkias might have been asking about the "Custom functions". These are functions which for some unknown reason do not have their own normal menu function. Instead, you go through the menu and select "Custom functions". They have their own little simple menu, and can change a few cool things in there. Just make sure you read the manual and know what you are changing before you do anything.

SamDoe1