Bounce flash

If I bounce the flash of a colored surface, will I be able to easily adjust the colors afterwards or should I avoid to bounce if the surface is not white. I shoot RAW.

If you take a sample shot with a good reference somewhere in the shot (in light representing the majority of the light in the scene) it's easy to tweak the white balance in the RAW conversion process.

I strongly suggest a standard 18% gray card rather than using a white reference for this purpose, as the gray card is much less likely to be overexposed when shooting it as a part of the scene. A white card or paper could easily be blown out and would then be useless as a white balance reference.

Thank you Skip. That sounds like a excellent advice. I'll do that.

I have had considerable doubt in the past about the result of using White Balance correction in postprocessing to fix the tint created from bouncing off tinted ceilings. Finding myself with holiday time available to conduct experiments, I decided to do a controlled study. I put my 20D on flash color balance, then put the camera on M at 1/60 f/8, mounted it on a tripod and aimed it at a Macbeth ColorChecker. I aimed a Metz 54MZ upward at 45 degrees, put it in Auto (not ETTL) mode. (I have many years of trust for Metz Auto for exposure accuracy in shooting weddings, I don't trust ETTL necessarily!) In a room with very subdued window light, I then held a white foamcore board overhead and shot. Then I taped a 7x12" piece of green paper on the foamcore (about the same green as in the second column of the Macbeth) and fired again. I then used Rawshooter Premium to look at and adjust the RAW files which resulted. (I also had the camera grab JPG files at the same time as the RAW, as reference photos.) As expected, the exposure and color balance was perfect with the Metz in Auto mode, so that only tweak to the images was the use of Shadow Contrast to bring the black framing of the Macbeth closer to native tonality. Here is the flash off foamcore:

Hopefully you'll find time to clean your sensor, too.

Thanks for this, Wilt. The whole area of color management is mostly a thick fog with me, and I have wondered about this sort of scenario.

The admittance committee has deemed this experiment sufficient for your initiation into the Measurebator's Anonymous club. Soon you will receive your official certificate, suitable for framing.

I'm thinking this is what made your walls neutral in the WB-corrected shot. The WB correction tool made the grey squares grey, when they were really beige from the counter reflection. This made your beige walls grey, too.

I think you need to try this again, with a piece of white paper or something covering the counter.

OK, Curtis, I reran the test, this time with a foamcore board over the counter so that the reflected light is neutral, and only the green-paper bounce contaminates the lighting.
Here is the flash-white-bounce...

Wilt, if you make both images as "equal" to each other as possible, i.e., the white, gray and black are as close in value to those of the respective colors in the other image, and apply the same white balancing on both images, you will be able to see that the WB of both are very close to each other. There still is a very, very slight hint of green on the "green" image but it's only noticeable if you do a side by side comparison of the images.

If you look at the two samples in your previous post, you can see that the WB of each mage are not the same. And neither are the intensities of the white and black (as evidenced by looking at the histogram). You obviously adjusted the gray of the "green" image so that the mid-gray is mid-gray but you didn't do the same with the other image which shows a warmer mid-gray for the mid-gray. That's like comparing apples and oranges. You really should use the gray eye dropper to WB both images similarly and adjust the black and white points of both images so that they are as similar as possible for an valid comparison.

Funny, in that second set the Macbeth cards are extremetly close to each other but you can still see a difference in the color of the wall behind it.

Still, it's pretty amazing what this technology can do. I think this shows (though it doesn't prove conclusively) that you can bounce your flash off whatever you like, as long as you shoot RAW (do we really need another reason to shoot RAW?).

Now if parts of your subject, or other things in the image, are receiving either direct flash or some other light source, then it is a much bigger can of worms.

PacAce wrote in post #2452714
Wilt, if you make both images as "equal" to each other as possible, i.e., the white, gray and black are as close in value to those of the respective colors in the other image, and apply the same white balancing on both images, you will be able to see that the WB of both are very close to each other. There still is a very, very slight hint of green on the "green" image but it's only noticeable if you do a side by side comparison of the images.
If you look at the two samples in your previous post, you can see that the WB of each mage are not the same. And neither are the intensities of the white and black (as evidenced by looking at the histogram). You obviously adjusted the gray of the "green" image so that the mid-gray is mid-gray but you didn't do the same with the other image which shows a warmer mid-gray for the mid-gray. That's like comparing apples and oranges. You really should use the gray eye dropper to WB both images similarly and adjust the black and white points of both images so that they are as similar as possible for an valid comparison.

The first image is set to the flash color temp. So it is the example of the native setting with a known light source.

The second image is the BEST I could obtain in adjusting White Balance with the dropper tool of Rawshooter Premium, plus it was necessary to manually use of Tint control and Hue control to render the wall as free from off tint as possible, as seen both in the wall color (the warm off-white paint) and the white foamcore below the card. In getting the foamcore as neutral as seen in the first shot, the wall was rendered neutral as well, and there was no hint of warmness even if I tried to tweak the White Balance setting in addition to the automated White Balance setting that had been chosen. Other settings such as the color saturation where otherwise matched between the shots, except for Exposure Compensation, which the greenish second shot required a greater EC for similar tonal range.

Because the greenish tinted illumination causes an imbalance in the rendition of colors especially those which have some of the complement of green, it is not a surprise that those colors are not rendered as well.

Wilt wrote in post #2453078
The first image is set to the flash color temp. So it is the example of the native setting with a known light source.

The second image is the BEST I could obtain in adjusting White Balance with the dropper tool of Rawshooter Premium, plus it was necessary to manually use of Tint control and Hue control to render the wall as free from off tint as possible, as seen both in the wall color (the warm off-white paint) and the white foamcore below the card. In getting the foamcore as neutral as seen in the first shot, the wall was rendered neutral as well, and there was no hint of warmness even if I tried to tweak the White Balance setting in addition to the automated White Balance setting that had been chosen. Other settings such as the color saturation where otherwise matched between the shots, except for Exposure Compensation, which the greenish second shot required a greater EC for similar tonal range.

Because the greenish tinted illumination causes an imbalance in the rendition of colors especially those which have some of the complement of green, it is not a surprise that those colors are not rendered as well.

Then you don't have an apples to apples comparison. IMHO, both images should have been adjusted for WB using the same gray square in the chart with the gray eyedropper. And nothing else (tint and hue) should have been changed (other than exposure so that both have similar white and black points). I think that would be the best way to demonstrate whether adjusting for color casts of images shot using bounce off colored walls will yield (or not yield, which I think is what you were trying to prove) results close to those shot using bounce off neutral walls.

PacAce wrote in post #2454523
Then you don't have an apples to apples comparison. IMHO, both images should have been adjusted for WB using the same gray square in the chart with the gray eyedropper. And nothing else (tint and hue) should have been changed (other than exposure so that both have similar white and black points). I think that would be the best way to demonstrate whether adjusting for color casts of images shot using bounce off colored walls will yield (or not yield, which I think is what you were trying to prove) results close to those shot using bounce off neutral walls.

But if the flash color temp is the published specification of the unit, and the color temp set in the RAW convertor is that same color temp, why is that not a valid 'control' in the experiment?

I will go back to the RAW file for photo 1 and use the dropper on that as well, just to satisfy your suggestion, but I still think it is not comparing as-shot vs. the contaminated and corrected shot.

edit: I just posted both the flash shot and the contaminated bounce shot examples back into message #6. I used only the auto dropper tool for adjustment of White Balance in Rawshooter Premium. Note that both the green-contaminated shots corrected to about the same, with auto-only example vs. auto-with-tweaking (of color and hue) which I had previously posted. Note that the photo shot under neutral light still reproduces the wall color with greater accuracy than the corrected one which had been shot under tinted light.

Wilt wrote in post #2454558
But if the flash color temp is the published specification of the unit, and the color temp set in the RAW convertor is that same color temp, why is that not a valid 'control' in the experiment?

I will go back to the RAW file for photo 1 and use the dropper on that as well, just to satisfy your suggestion, but I still think it is not comparing as-shot vs. the contaminated and corrected shot.

It's not a valid comparison because you are using your flash shot as the "standard" while trying to make the colored shot look like the "standard" by using the gray eyedropper which adjusts color cast for true neutral color. You flash shot is not a true color standard because it has a warm cast to it. If you want the green shot to look like your flash shot, then the green shot should be adjusted to be neutral and then the WB should be adjust beyond that to be as warm as your flash shot is. But it'll be a lot easier to bring the flash shot to be neutral compared trying to bring the green shot to be as warm as your flash shot.

BTW, there is a thread that Curtis started which shows that flashes change color temps based on the intensity of the flash output. So, unless the Metz flashes are immune from color temp variations due to output intensity, then I don't think that the published color temp of that flash necessarily applies to your particular shot. It could possibly be give or take a couple of degrees from the published.

PacAce wrote in post #2454593
It's not a valid comparison because you are using your flash shot as the "standard" while trying to make the colored shot look like the "standard" by using the gray eyedropper which adjusts color cast for true neutral color. You flash shot is not a true color standard because it has a warm cast to it. If you want the green shot to look like your flash shot, then the green shot should be adjusted to be neutral and then the WB should be adjust beyond that to be as warm as your flash shot is. But it'll be a lot easier to bring the flash shot to be neutral compared trying to bring the green shot to be as warm as your flash shot.

BTW, there is a thread that Curtis started which shows that flashes change color temps based on the intensity of the flash output. So, unless the Metz flashes are immune from color temp variations due to output intensity, then I don't think that the published color temp of that flash necessarily applies to your particular shot. It could possibly be give or take a couple of degrees from the published.

I posted additional shots in message #10

Here are the same two shots you added to which I applied a gray eye-dropper to the middle of the mid-gray square of each shot to white balance the shots. They are so close it's really hard to tell them apart (unless you are doing a side-by-side comparison).

HOSTED PHOTO
please log in to view hosted photos in full size.

HOSTED PHOTO
please log in to view hosted photos in full size.

celter wrote in post #2449559
If I bounce the flash of a colored surface, will I be able to easily adjust the colors afterwards or should I avoid to bounce if the surface is not white. I shoot RAW.

Getting back to the OP's original question, after hassling Wilt about his comparison test , I'm going to say that Wilt does make a valid point. It is not always possible to color correct pictures that have a color cast to them from bounced flash off colored walls. It all depends on how strong or weak the color cast is and whether there are other light sources in the mix. If the only light source is from the bounced flash and the color cast is not extreme, it could easily be corrected. However, if the color cast is so strong as to cause clipping of some of the colors or uneven shifts in the colors of the object in the image, it'll be next to impossible to fix the image 100%. And if there are other light sources whose color temp is not blended in with the color cast off the wall, that image will also be impossible to fix across the entire image. If the parts of the image that have the color cast can be isolated from that lit the othe other light source, then of course, that can be corrected.

Another thing to consider is that some objects absorb certain colors and others reflect them. If the color cast is strong enough to visibly alter the colors of those objects, color correction maybe not necessarily correct all the colors in the image either.