Olympus RC TTL mixed-flash vagaries

I guess what you're demonstrating is that you're doing it wrong.

Would working through the Strobist tutorials be a way to learn how to do it right?

453C wrote:

Would working through the Strobist tutorials be a way to learn how to do it right?

Strobist is a good source of information on how to use a flash, but it can be a hard slog work going through all the videos etc.

Some very simple advice was given in the last thread, but it appears to have been ignored.

I just wonder whether this thread is an attempt to prove someone right, when they are wrong,.

OzRay wrote:

453C wrote:

Would working through the Strobist tutorials be a way to learn how to do it right?

Strobist is a good source of information on how to use a flash, but it can be a hard slog work going through all the videos etc.

Some very simple advice was given in the last thread, but it appears to have been ignored.

I just wonder whether this thread is an attempt to prove someone right, when they are wrong,.

Ah, I will jump to yonder thread and take it from there.

There are many things going on here I feel, I will try to cover them in a coherent manner.

*--------------------------------------------------*

1) In my own tests, combining a FL-50R and DMW-360L (Panasonic version of FL600R) together did not come out with the expected results. In my own tests, with a GX7, it seems that the FL50R and 360L/600R have different ideas on the fill flash balance ratio. It should not be that way...yet it seemed that way IRL.

It could be that the differential in power output confuses the RC system. Try your shots again, making sure that each flash is on a different RC channel. In this way both the camera and you can independently control the differing flash outputs.

2) I could be wrong but I believe the significant problem is that you are bouncing the flash behind the bear.

Therefore, either 1 of 2 things seems to be occurring:

• a) the flash power is insufficient to properly illuminate the subject area, leading to overall underexposure (your images have that overall visual appearance, at least).

Did the in-camera flash signal identify a properly exposed image? Did the flash OK light blink 'insufficient' or was it OK?

The lack of exposure difference between the evaluative and spot metering seems to be a telltale, there certainly should have been a difference. The lack of difference speaks "The flash did not have enough power, regardless of what exposure you were hoping for".

Or,

• b) the behind-the-subject flash is simply fooling the exposure system

Here are my thoughts:

Your comment

I didn't do it today but have on occasion tried higher than +0.7, all the way up to +1.7 and beyond, and sometimes when the flashes decide that they aren't going to do it, they just don't seem to want to do it. Bright backgrounds are harder.

(emphasis added)

makes a bit of sense: TTL metering is expecting the center area to respond to flash brightness (fill, 'dynamic'), with the background being the preexisting (ambient, 'static') part of the equation. You've essentially flipped that around: your flash aiming is making the background the dynamic, varying with flash output, and the center [foreground] simply coming along for the ride.

I am not sure TTL flash was meant to handle that situation. All camera metering is center-weighted but you are asking the TLL control system to try to alter the center exposure by altering the background fill, while simultaneously ignoring the actual background exposure.

That might be too much to ask of a machine. You know what you want to do, we know what we want to do, but the machine is pre-programmed with certain parameters.

As for Oly TTL flash, be it on camera, off camera via cable or off camera via RC control for E-PL1, E-P3, E-PL5, and E-P5 I have found that the happy default for flash compensation is to leave it at plus 1.3 - sometimes need more, sometimes less, and if macro or closeup then best left at or near 0.

So start with the notion that Oly TTL definitely always under-exposes and take it from there. Compare the flash in the hotshoe TTL result with a flash self auto result and see what I mean.

Use the blinkies to look for over-exposed spots in the review and adjust flash comp to taste.

I suspect that with "studio" type setups it would always work out better to use manual flash and then just alter the power settings until it looks right.

Regards...... Guy

ctlow wrote:

Okay, ladies and gentlemen, I've got my fire extinguisher handy, but let's generate light, not heat. I submit the following both to teach and to learn.

I didn't find it practical to do every possible permutation.

All were made under similar settings, ISO 200 with ambient light, ISO 400 with flash, f5.6, 60mm lens on an Oly E-M1 (Mark I). There is the little on-camera flash, the trigger, set at "0" but it still flashes enough to give the little fellow an eye-highlight.

Camera-left is an FL-50R, about even with the bear, and pointed variably into a back wall/ceiling corner of the room (fore-lighting) or a front corner of the room (backlighting). Camera right is an FL-600R, positioned as the FL-50R.

The ambient light exposures varied, and were all longer than 1/6 second, so ambient light is almost irrelevant in the flash exposures, all at 1/125. I'm pretty sure that the flashes never went anywhere close to full power; it didn't sound like it, and I've examined some of the exif files and they show far, far less than maximal output.

I have used the RAW files to make smaller but otherwise unmodified max-quality JPEGs. The files went through ACR and then PSE. WB adjusted in one photo (my error).

Every different lighting setup I exposed in evaluative, centre-weighted, and spot metering. The centre-weighted sometimes veered one way and sometimes the other, more often barely different from evaluative, so I'm generally omitting it below.

I've numbered the images from my own file-names.

First, we've got this humble-but-cute little teddy bear, and all he wants to do is spread a message of love.

464 -all-around incandescent ceiling lighting, evaluative-metering

466 -all-around incandescent ceiling lighting, spot-metered, so brighter foreground than with evaluative metering

This bear has a fairly shiny ribbon, and the tufts of "fur" are bright, so there's probably a wide range of tones in that creature.

463 -incandescent ceiling back-lighting, spot-metered, so the bear stays relatively bright, as it should, and the background brightens, as expected

468 -flash lighting, aimed behind the camera into the far walls/ceiling corners, so essentially diffused front-lighting -evaluative metering - both flashes on RC TTL at "0"

Would this brighten up by converting to spot-metering?

470 -same lighting as previous, but spot-metered -brightness essentially unchanged from evaluative metering

Let's try backlighting, aiming the two flashes into the corners in front of the lens, behind the bear.

474 -flash-lighting, aimed into the walls/ceiling corners behind the bear -spot metering -so: backlighting, essentially unchanged exposure from front-lighting -the bear has gone a bit darker than 470

All right, 'nuff foolin' around, let's make some adjustments. As many others seem to do, I generally start with the RC TTL exposures set a bit high - often as follows:

478 -flash, one RC flash at +0.7, the other at +0.3, diffused front-lighting, evaluative metering -bear still too dark and/or too little contrast

480 -same setup, except spot-metering - better than with evaluative metering, still dark and flat -it is brighter than the "TTL 0" image 470- but not as much as one would think from the settings

477 -flash, diffused back-lighting, +0.7 and +0.3 -only a trace brighter than the "TTL 0" 474

I didn't do it today but have on occasion tried higher than +0.7, all the way up to +1.7 and beyond, and sometimes when the flashes decide that they aren't going to do it, they just don't seem to want to do it. Bright backgrounds are harder.

Then, just when I thought I had at least some data to debate, I also altered the focussing algorithm. I did that because of a brain-slip. Nonetheless, to my amazement, as I went from "all-targets" focussing to "single-target" focussing, the image brightened up (what?!?) ... if I switched to a black background. This is from a different-but-similar series:

445 -flash, TTL +0.7 & +0.3, evaluative metering, all-targets focus -back-lit, so highly dependent on scattering of light hither and yon

(Yes, I'm sorry, I'm changing two variables.)

448 -flash, TTL +0.7 & +0.3, evaluative metering, single-target focus

And there, I got visually identical images with diffused front- or back-lighting.

So, two things from that:

• bright backgrounds do seem to be ("seem", not "are positively") a factor, and it's as if on RC TTL flash-mode, it's like plain, old-fashioned, full-frame, "dumb" auto-exposure;
• why changing from all-targets to single-target focussing should matter I have no idea, but the effect was dramatic.

Of course, even the brighter back-lit bear is probably no brighter -it just stands out better from a dark background.

It's complicated, but still I submit not as easy and transparent to use as ambient-light TTL. Again: RC TTL appears, to me, to behave like dumb full-frame metering - or worse.

TTL uses the camera`s metering, its through the lens metering, call it dumb if you want.

Think about the reflectance properties of your subject or its surroundings, at 18%, sometimes a bit more your TTL will nail it every time, if these properties are higher or lower generally your going to need to dial in some comp, either positive or negative, its just the way these things work.

No knocking anybody - just what I see.

Charles

-- hide signature --

ctLow Photog

What you need is a foreground fill flash so that the system can stay within its pre-programmed parameters, that is balance background and foreground exposures.  With 2 flashes, one background and one foreground, set to different RC channels, the camera should be able to meter a proper exposure.  It will test fire channel A first, then channel B, then compute an overall exposure which combines both, then fire all flashes during the picture process (all this happens automatically in the space of time between pressing the shutter butter and finally closing the shutter, so you aren't really aware of this).

The 2 flashes will give the metering what it expects, a major foreground, centered subject that it is flash metering directly for.  The background should be "along for the ride", that is the camera will compute the flash fill but it should be paying the greatest attention to the center.  After that, you should find that the flash exposure compensation actually has an apparent result.

This is based upon my own test of a FL-50R for a left/foreground subject (room "A") and a FL-360L for the right/background subject (room "B", an open floorplan room behind the main room) that I did after I acquired my second RC flash.  The setup actually demanded quite a bit of TTL exposure compensation to get the flashes balanced but balance they did, they fully responded to the exposure adjustments as they were supposed to.

Again, for my test, the [background] FL-360L was prone to overexposure, it wanted to blast the background out as compared to the foreground, and I believe this was due to the center-weighted balance of the system (even in evaluative metering) and it was programmed to bring that center up, regardless of other parts of the frame. This was a great surprise to me as the background flash was less powerful than the main, and a good learning experience.

First pic - I think I can see some smoke.

Second pic - now you are in the ball park. No need for TTL at all there but certainly experiment with TTL to see if you can duplicate/improve that, and also get the same from shot to shot.

Remember that flash compensation is additive, that is the SCP flash comp adds to the RC menu flash comp adds to the individual flash's flash comp.

I really think that in the teddy bear "studio" sort of setup that manual power solves all problems, it just takes a few guesses and trial shots to get there.

Regards..... Guy

ctlow wrote:

Thanks, dinoSnake,

Helpful.

I have studied Strobist, and now course number 3 is out - will have to check it. Overall: not helping with this.

Let's just take my front-lighting examples, by which I mean the ones with the flashes bouncing into the room corners behind me. I had the bright (groan) idea to repeat a similar test I posted long ago: just put it on full power and see if the flashes can even do this (but recall that the exif data showed that they were far from full power). This is what I did to the poor critter:

He couldn't even see for a few minutes.

So, there's lots of juice in the flashes.

These are unretouched but from the OOC JPEGs (a bit faster and I'm on the run), "Vivid".

I'll set it up again later and just keep pumping up the TTL correction and see if I can finally get it good.

-two flashes on RC-manual, 1/4 and 1/8 power

More later!

Charles

I don`t know what all you fussing is about.

You have already been told how TTL works yet you still think there is a problem, in fact its behaving exactly how it should, its rendering the white bits middle grey due to the reflectance levels.

The solution is simple dial in some positive comp, this is not a fault of the flash system, its just that your not telling the flash system what you want it to do, it can`t read your mind, its pre programmed to do what it is told.

Picture this, your shooting a bride and a groom at a wedding, the brides wearing a ivory dress covered in shiny sequins, the grooms wearing a black tux.

Your going to use a flash but how are you going to measure the light, what would you measure, the light falling on the subjects or the light reflected off and back to the camera, manual flash or TTL.

Both are good options but you still need to know how to control the flash units.

ctlow wrote:

Guy Parsons wrote:

First pic - I think I can see some smoke. :-)...

Thanks, Guy, and also to dinoSnake who is sticking with this very thoroughly. I'm enjoying this and learning and finding it interesting.

It seems to me that idea of using foreground fill flash is a work-around. That's not to say "unnecessary" but it runs counter to the whole idea of creating a large, diffused light source (neilvn.com/tangents/, http://strobist.blogspot.ca, "L, S & M " and others). Theoretically, the Oly RC TTL system should be able to do that.

I think the TTL flash can properly handle fully diffuse light, that is coming from all directions. If you place a TTL RC flash inside a softbox that creates a very large, very diffuse light pattern, the camera will properly expose based on the light conditions.

I don't think that the diffuse part is your problem, I think the direction of the light source is the problem. "Diffuse", from the camera's perspective, should still mean "towards the subject". In your case you're talking backlighting via flash, and in-camera exposure meters pretty much mess up ambient backlighting so how is flash going to turn out well without (major) user intervention?

I am using A & B sub-channels; that's how I control the two flashes separately. I have tried but am not currently using overall flash-compensation from the SCP nor on the flashes themselves - just from the RC screen.

I'm interested in dinoSnake's information about flash A and B being tested sequentially by the Oly RC TTL system. Is that documented somewhere? I have, a year ago when struggling with this topic, tried to reverse which flash was which, and could discern no effect.

Yes, someone has decoded the Olympus RC flash system

http://strobehacker.tumblr.com/post/78490593051/the-rc-code-has-been-cracked-well-mostly

OK, I remembered the TTL RC control pulses wrong.  If the decode is correct, all flashes fire on preflash for measurement and then are programmed for power output.

This just adds to the difficulty of [our] test here, as the camera is not testing and setting power for each flash unit, it is testing and setting power levels as a whole for the exposure.  That's even worse when it comes to complex setups like backlighting: how can it determine the different power levels for the different lighting zones?

It can't!!

So, if true, this is why you (and I, and everyone else using TTL RC multizone lighting) have a problem, why we need lots of flash EC (both plus and minus) to get what we want.  The camera can't, and isn't, programming each flash zone as it isn't testing each flash zone, only all of them for a total, complete exposure.  So we end up fiddling with the EC.

dinoSnake is using two dissimilar flashes too (one like mine, the other not), and having challenges, which he overcomes. I am interested in more detail and images about that room photograph, if that can occur.

So, having not needed to run my furnace for all of yesterday after all the heat from those full-power exposures, I did some testing to try to replicate that with TTL.

This is - stubbornly - with both flashes beside the bear and aiming up into the walls/ceiling corners behind the camera, i.e. diffuse front-lighting. That's the setup where, on manual flash, I could get it extremely over-exposed (previous post, image 481), i.e. the RC TTL underexposures are not for lack of available flash-power.

488 - RC TTL on-camera flash "off", FL-50R (A) at "0" and FL-600R (B) at "0" - simply dark

This goes all the way to this:

501 -A on +5.0 and B on +4.7 -badly over-exposed and in the same ballpark as 481 (previous post) with both at full power. The histograms look pretty ugly (you don't want to see 'em), and 481 is brighter.

About right on my uncalibrated monitor is this:

492 -A on +1.7 and B on +2.0 - roughly "correct" exposure

492 histogram - only a little room available to the right, and the next highest exposure (not shown) is still all contained, but just looks a bit washed-out to my eye.

And if I tighten that up a bit, like this:

tightened histogram for 492 -minimal sacrifice of the black/white extremes

I get this:

which isn't bad if you like teddy-bears. This is from a "Vivid" JPEG (not otherwise manipulated), not from RAW, so not 100% congruent with some examples in earlier messages.

Recall that with ambient light, none - well, perhaps "less" - of this "wild" exposure adjustment was necessary (image 466, first message of thread).

So: you're all correct. Just crank it. But it has to be cranked - sometimes - so much that it just feels - "felt" - to me and I believe to others, that something is just not working and so to give up and revert to manual.

And manual would be fine in a studio with a static setup. It's not fine in a dynamic shoot with everything or everyone moving around.

My RC TTL starting point has been +0.7 and +0.3. (It's a "MySet".) My resistance to starting higher was, quite frankly, probably just from over-thinking and not simply doing what was necessary (in testing only - on a shoot, I move quickly to do what I need to do to get it good). And on occasion, that will still probably be too much. But I might start higher. (And I set them a bit differently, hoping for some "modelling".) I set the more powerful flash to the higher compensation - seems to make sense.

By the way, this thread is an off-topic off-shoot from https://www.dpreview.com/forums/post/58988322.

FWIW this is why, in fixed setups, like studio, pros go with manual and use flash meters: since exposure is based upon light falling on a subject and quantity of light is based upon 2 variables, source brightness versus distance, when we can control those 2 things just go all manual.  Once set it will stay set, as long as the 2 variables remain the same.

ctlow wrote:

All right; getting there.

Now, we have heard users of other makes (higher-end models?) claim that their optically-controlled RC TTL does better than this, but in the real world, your observations about the limitations of the the Olympus RC TTL system sound valid.

The message I'm getting from pretty well everywhere is - pardon me going right back to basics - don't, for the love of all that's holy, ever point a bare, on-camera flash at anyone, if one has any respect for "good" lighting at all.

That is not true.

https://www.google.com/search?q=fill+flash&ie=utf-8&oe=utf-8#tbm=vid&q=daylight+fill+flash

for example

https://www.youtube.com/watch?v=jaBExauPQTo

Bare flash is indeed harsher but it is also far more powerful. Don't consider any "rule" in photography as a rigid law - I've learned, in only a very short time, to ignore those who make proclamations to others that "You must / must not do this!" The bare flash can be harsh but sometimes that is what you want, maybe for a special effect such as ghoulish ambiance or as you see full sunlight backlit daylight fill, and if you learned "Never do that!" you'll miss getting the most of the shot.

For example,

https://www.youtube.com/watch?v=3RKA8K0PO00

and

https://www.youtube.com/watch?v=cDwPnqYZFMY

The common wisdom of "Don't shoot bare flash" is because, especially here, most photographers are men and most of their subjects are women, and they want a soft, flattering light to create soft, flattering features. But that is not a rule. That is a preference. And all too many people think that their preference is, indeed, a general rule for others to honor.

Almost all portraits will end up using diffused lighting. Almost all. But not "all".

So, not in the studio, no umbrellas or light-boxes, "they" advise that, if you have no other recourse, bounce the flash off something handy, usually a wall or ceiling.

I was at a wedding where the photographer had one on-camera flash ... aimed pretty well constantly up over his left shoulder. When I first noticed it, before the ceremony, he was getting some preliminary shots in the church with a high, dark ceiling. Fancying myself somewhat of an expert (stifle those guffaws, people!), I asked him about that, mentioning that my flashes couldn't do that. The young man just smiled at me and quietly said, "This flash will."

This was in 2010, and I'm not sure what the latest Nikon systems (can I say that here?) were capable of back then, but man did he get beautiful photos.

That's the Neil van Niekerk technique, bounce the flash over a shoulder. You can read about it on his website that I previously posted a link to.

ctlow wrote:

Paulmorgan wrote:

I don`t know what all you fussing is about.

You have already been told how TTL works yet you still think there is a problem, in fact its behaving exactly how it should, its rendering the white bits middle grey due to the reflectance levels.

The solution is simple dial in some positive comp, this is not a fault of the flash system, its just that your not telling the flash system what you want it to do, it can`t read your mind, its pre programmed to do what it is told.

Picture this, your shooting a bride and a groom at a wedding, the brides wearing a ivory dress covered in shiny sequins, the grooms wearing a black tux.

Your going to use a flash but how are you going to measure the light, what would you measure, the light falling on the subjects or the light reflected off and back to the camera, manual flash or TTL.

Both are good options but you still need to know how to control the flash units.

Maybe, Paul, but it's still I think patently and obviously true that under ambient-light TTL conditions our cameras perform much better than this, the "evaluative" metering being truly amazing.

If have this sailboat shot and at a photo seminar I was presenting, I just threw in that by the way, this would have required major manual compensation back in the days before "smart" metering. As it was, I just lined it up and snapped.

This all depends on you metering mode, centred weighted would have dealt with this just fine but TTL flash is a little different, its still metered through the camera lens but all that light bounced back in gets averaged out.

reasonable "smart" exposure despite having a very bright area in it

I understand completely about having to adapt in the real world to the strengths/weaknesses of a system. When on a shoot, that's exactly what I do. But that's not my topic here. It's just to illustrate for those who might find it helpful that the Oly RC TTL system does not behave like that, and learning how it does behave takes some doing.

Part of your problem I think is that your bouncing both units and in doing so your creating a single light source (just maybe)

Charles

-- hide signature --

ctLow Photog