[mpodesta]

Glad I came across this thread!

Last couple of months I really started to get into photographing indoor concerts and I've been running into this problem quite a bit

[UnknownVT]

Last couple of months I really started to get into photographing indoor concerts and I've been running into this problem quite a bit

Yes LED stage lighting is getting more and more prevalent - I guess because they have now become economically viable - that even small clubs can afford them.

They are also very versatile in that one can get just about any color lighting by varying the RGB output.

"Flesh" colored lighting which is a pinkish magenta has always been popular because it's flattering, and looks attractive. Since it is changeable and different people have different preferences - plus trying out different variations - the "flesh" magenta varies from almost blue to almost red.

I am sure magenta has always caused some problems in digital photography - but with LEDs - since magenta is made up of Red and Blue LEDs only that means the overall spectrum is only really 2 spikes with basically only red and blue components -
this causes some problems with capture on RGB Bayer matrix - as there are as many Green filters as red and blue combined -
so it's almost like using only half the sensor.....

In addition JPG compression is hardest on red (and to a certain extent blue) so there's a double whammy since any photo destined for the web or display on any computer monitor is most likely going to be JPG.

I hope some of the suggestions on how to mitigate some of the problems helps -
as this is a problem that is not going to go away -
in fact the opposite is true -
LED lighting is likely to become the dominant stage lighting -
photography is secondary to any venue
(and sometimes barely tolerated)
and photographers will just have to cope.

[UnknownVT]

Summary so far on how to cope with extreme lighting such as magenta made up of Red and Blue LEDs only.

My personal take is to try to minimize the adverse effect in the picture taken if at all possible - any post processing is kind of like correcting a mistake already made and the damage is already done - but that's just me and YMMV.

However I call it "extreme" lighting - because it is - and it is very difficult to mitigate - at least with my limited experience and knowledge.

No amount of white balance adjustment seems to be able to cope with something like magenta made up of red and blue LEDs only (since the light would lack any green component).

So far I have found using weak fill-in flash seems to help somewhat - as that throws some full spectrum white light into the scene - thus mitigating the extreme lighting.

I have now found this is different for different cameras -
slightly OT - I use a Canon G10 compact far more and therefore have a lot more experience with it - I have a custom setting of "P" with AutoISO, AWB, -1/3 stop overall exposure compensation and slow-sync fill flash at -1 2/3 stop compensation (when flash is fired on the G10 the AutoISO limits the max ISO to ISO250) -
this seems to generally do well - allowing most of the ambient scene to show well but giving just enough flash to mitigate most extreme lighting - the examples shown in Post #83 above are from the Canon G10 with -1 2/3 stop flash compensation.

So I thought using a -1 2/3 stop flash compensation would work with the K-x too - sometimes it did, but mostly it did not do as well - the results are no where near as consistently successful as the G10.

Pentax K-x settings - "P", AutoISO (200-5000), HighLight Correction On, AWB (w. subtle tungsten correction), -1/3 stop overall compensation.

Through trial and lots of error - a -1 stop slow-sync (first curtain) flash compensation seems to give better results - this could be because with flash the K-x does not drop its AutoISO to ISO250 like the G10 but stays with whatever the shot without flash would have used.

I would have thought any flash compensation ought to be relative to whatever AutoISO chosen - but maybe it isn't - whatever the reason I find I have to use about -1 stop slow-sync flash compensation to get more consistent results.

pudding eating -

No flash -


-1 stop slow-sync flash (all other settings exactly the same)


No flash -


-1 stop slow-sync flash compensation (all other settings exactly the same)

this last pair shows that perhaps -1 stop maybe a little too much - perhaps -1 1/3 or even -1 2/3 may have been better - but there is no way I can tell for sure - however as an overall picture I think it's just better with the fill-in flash.

Sometimes it just doesn't work as well -

No flash -


-1 stop flash

the flash shot doesn't look that much better than the no flash shot -
one of the main points to using the fill-flash is to try to retain details in the magenta light -

No flash detail -


-1 stop flash detail

in this case flash didn't do too well -
perhaps for whatever reason -1 stop wasn't enough - or the K-x over-saturated the red and blue channels, so adding white light didn't help?

Post Processing - I would much prefer to use this as a last resort - kind of like salvaging the shot if I didn't get it right at the taking stage.

To recover the use of select white or grey point in any photo editor seems to do quite well as shown in post #79 - for more extreme/difficult cases where RAW can't even manage - and i have shown this in Post #23 - the use of Pentax DCU/SilkyPix 4.11 on either the RAW/DNG or JPG and using the select grey-point seems to be able to recover the extreme lighting - see Post #86 for an example on a JPG.

[Marc Sabatella]

in fact I really struggled to get a neutral looking shot without a cast -
and I thought RAW was supposed to give more flexibility -
my straight JPG was so much better

In your opinion. To my eyes, they were "different", but neither was obviously "better" on my calibrated monitor. I say this not to be argumentative - you're certainly welcome to whatever subjective preferences you like - but I just want to emphasize for the benefit of others that that's all there is to this. There's nothing objectively wrong with ACDsee's color handling with the K-x; it's just that some might prefer one color rendering, others might prefer another. I also rather doubt any such preference even in one person would turn out to be all that consistent if looked at many images - some images you might prefer prefer one rendering, other images you might prefer the other. But we're talking about quite small differences. I've included a sample of my own below (well, not my own image, but one taken by a friend that I have access to).

Also if HighLight correction is turned on,
which I think my sample RAW/DNG paired JPG shot has -
then one would really be scr*wed with ACDsee Pro 3.0 -

It's true that ACDSee, like most RAW processing programs, is not designed to mimic the specific dynamic-range enhancing algorithms of different cameras, although I gather *some* might. But still, you're not really "scr*wed" - you'd just need to design your own custom pre-processing curve to mimic what the camera does in its own JPEG processing. Save that as a preset, and you could reproduce the effect on a whole batch of images in just a couple of keystrokes.

Part of the reason for shooting RAW, though, would be to get more control over the curves rather than having to settle for the one-size-fits-all curve applied by the camera or by a preset.

To be kind I think it's just that ACDsee Pro 3.0 hasn't yet fully supported the K-x

This is just factually incorrect.

To give people an idea of the nature and magnitude of the differences we are talking about, below is a side-by side comparison of the first image that caught my eye of the hundreds I have access to. I won't (yet) say which is the camera conversion and which is ACDSee's. There are indeed subtle differences here, but does anyone really see either of these images as being so bad it indicates a problem?

Attached Images

 

[UnknownVT]

In your opinion. To my eyes, they were "different", but neither was obviously "better" on my calibrated monitor. I say this not to be argumentative - you're certainly welcome to whatever subjective preferences you like - but I just want to emphasize for the benefit of others that that's all there is to this. There's nothing objectively wrong with ACDsee's color handling with the K-x; it's just that some might prefer one color rendering, others might prefer another. I also rather doubt any such preference even in one person would turn out to be all that consistent if looked at many images - some images you might prefer prefer one rendering, other images you might prefer the other. But we're talking about quite small differences. I've included a sample of my own below (well, not my own image, but one taken by a friend that I have access to).
It's true that ACDSee, like most RAW processing programs, is not designed to mimic the specific dynamic-range enhancing algorithms of different cameras, although I gather *some* might. But still, you're not really "scr*wed" - you'd just need to design your own custom pre-processing curve to mimic what the camera does in its own JPEG processing. Save that as a preset, and you could reproduce the effect on a whole batch of images in just a couple of keystrokes.

Just so I am not being argumentative -
I will concede to your take.

However I have a problem on ACDsee Pro 3.0 default rendition of my K-x RAW/DNG -
and it is not a small difference (I do realize this is just my opinion)

I took DNG+JPG paired photo:

JPG (IMGP3965.JPG) - (PhotoBucket seems to have dropped the EXIF metadata on these) -


(the paired) RAW/DNG (IMGP3965.DNG) ACDsee Pro 3.0 386 converted default settings -

The difficulty as you so kindly pointed out was due to K-x HighLight Correction being turned On and ACDsee did not understand it.

Yet ACR (Adobe Camera RAW) 5.6 seems to understand HighLight Correction -
the same DNG (IMGP3965.DNG) ACR 5.6 default conversion -


Perhaps using a custom profile for ACDsee could solve or mitigate the problem - but Pentax DCU/SilkyPix or even ACR already work with their default camera setting - thus I just prefer not to use ACDsee Pro 3.0.

Again, please, I am not being argumentative -
in fact I am grateful for your input -

Re: ACDsee Pro 3.0 not yet supporting the K-x
Quote:
To be kind I think it's just that ACDsee Pro 3.0 hasn't yet fully supported the K-x

This is just factually incorrect.

I actually got the unsupported bit from your very informative Post #185 in Do people really shoot in JPEG???

I'd guess is this a function of the K-x not being supported yet, and hence not having custom color profiles for it.
.....................
This is of course true in general, although I do suspect this particular case is a bit unusual because you are dealing with an unsupported camera and hence the colors are off in unusual ways.

[dragonfly]

Any chance of posting a raw image taken under this nasty lighting as a PP challenge?

I've got a few ideas for attacking the problem, but not with an already compressed and scaled down jpg.

I think it's a great challenge that we could all learn a lot from.

[UnknownVT]

Any chance of posting a raw image taken under this nasty lighting as a PP challenge?

I've got a few ideas for attacking the problem, but not with an already compressed and scaled down jpg.

I think it's a great challenge that we could all learn a lot from.

As requested -
I found a free file sharing site - MediaFire - I'm not sure how good/bad it is -
but it seems to rate well on a very cursory search -

I've created a Kx folder (link)
where I have uploaded the paired DNG and JPG as used in Post #23.
IMGP4490.DNG
and
IMGP4490.JPG

(I tested this and have successfully downloaded both files while NOT logged in)

be very interested in seeing the results of any processing/salvaging.

[Marc Sabatella]

However I have a problem on ACDsee Pro 3.0 default rendition of my K-x RAW/DNG -
and it is not a small difference (I do realize this is just my opinion)

Oh, I agree that if you use highlight correction and don't apply your own custom curve to mimic the one applied by the camera, the difference is not small.

Yet ACR (Adobe Camera RAW) 5.6 seems to understand HighLight Correction -
the same DNG (IMGP3965.DNG) ACR 5.6 default conversion -

Yeah, like in said, apparently *some* RAW converters do try to mimic this behavior. But it's really not something one should normally expect.

Again, please, I am not being argumentative -
in fact I am grateful for your input -

Likewise. I'm not out to convince you to change your mind about ACDSee here. I was just concerned that others might write it off just because it didn't happen to be the right program for you.

I actually got the unsupported bit from your very informative Post

Ah, yes :-). But that was over half a year ago. It's supported now.

P.S. Just looked at your RAW sample. ACDSee's tint slider maxed out at -100 doesn't go far enough, unfortunately. Playing with the advanced Color tool you can improve the picture some beyond that, but it would have helped if the WB sliders went further. Looks like they weren't considering LED lighting when they set the bounds for the WB sliders.

[UnknownVT]

P.S. Just looked at your RAW sample. ACDSee's tint slider maxed out at -100 doesn't go far enough, unfortunately. Playing with the advanced Color tool you can improve the picture some beyond that, but it would have helped if the WB sliders went further. Looks like they weren't considering LED lighting when they set the bounds for the WB sliders.

Many thanks for that input - very helpful -
that makes sense -
that's why I have the difficulty with the magenta lighting made up of red and blue LEDs.

On that sample - other than just being difficult lighting
do you think the red and blue channels are over-saturated or clipped?
therefore leaving very little, to no room room for manipulation/salvaging?

However I still think it's close to "miraculous" that Pentax DCU/Silkypix 4.11 can manage to salvage/recover that shot in either RAW/DNG or even JPG as shown in Post #86

[Marc Sabatella]

ACDSee says the blue channel is clipped in quite a few places on the white t-shirt; the red channel surprisingly does not seem to be. Looks like there's just enough useful info in that red channel to allow some detail after removing the color cast. Note there appears to be *no data whatsoever* in the green channel there, and that's probably why some programs would be reluctant to ever render that shirt white.

[UnknownVT]

ACDSee says the blue channel is clipped in quite a few places on the white t-shirt; the red channel surprisingly does not seem to be. Looks like there's just enough useful info in that red channel to allow some detail after removing the color cast. Note there appears to be *no data whatsoever* in the green channel there, and that's probably why some programs would be reluctant to ever render that shirt white.

Thanks so much for that information -
yes, that particular photo was really awkward to say the least.
ACR (Adobe Camera RAW) 5.6 could not do much on either the DNG or JPG
neither could my regular photo editor or PS Elements on the JPG version.

Now we know for certain why - there is a total lack of green -
like I surmised magenta light made up of only blue and red LEDs would only have two peaks/spikes in the spectrum at obviously red and blue and nothing in between...... thank you so much for confirming this.

However now I am really amazed that Pentax DCU/SilkyPix 4.11 could actually get a "natural" looking shot out of either the RAW/DNG or even the plain humble JPG - (I've reprocessed the DNG version since I thought I got a bit better at PP) to show here:

Original JPG shot -


Pentax DCU conversion of DNG using grey point on bassist's shirt -

this is an OMG! jaw dropping result....

and this one has my jaw on the floor...

almost UNbelievable -
since it is from the JPG! -
all I did was to open the JPG in Pentax DCU
and then do the same select grey point on the bassist's shirt ......

[kittykat46]

Very good colours at the end...
The software was doing the equivalent of on-camera Manual White Balance.
Its very fortunate that the guitarist was actually wearing an accurately white shirt, so the software took over from there.
Its very hard to get the MWB right when you can't find any suitably coloured object to act as a calibrator for the program.

[UnknownVT]

Very good colours at the end...
The software was doing the equivalent of on-camera Manual White Balance.
Its very fortunate that the guitarist was actually wearing an accurately white shirt, so the software took over from there.
Its very hard to get the MWB right when you can't find any suitably coloured object to act as a calibrator for the program.

Thanks for the comments -

BUT is was a lot more than that
I could not get anywhere near natural colors out of that shot when using either ACR (Adobe Camera RAW) 5.6 or LightRoom 3 Beta - since there was no Green at all - there is absolutely no "White Balance" as we understand it either in-camera or PP that could cope - since the light was two peaks of Blue and Red in the spectrum (nothing in between - no green component) - almost no possibility of approaching a natural looking shot - I also tried other RAW processors with basically no success.

It was only because I thought I'd try the supplied Pentax DCU (Digital Camera Utility) based on SilkyPix 4.11 (which most probably regard as a throwaway S/W - I mean how many people actually use DCU?)
and thought of selecting a grey point - that the close to "miraculous" recovery was made - again I have not been able to do this with any other PP software - RAW or otherwise.

The amazing thing was that Pentax DCU could work on the JPG,
and get basically same astounding result.

There were two requests for the original RAW/DNG file (post #97)
- I have posted them for anyone who cares to try some post-processing on these -

in Kx folder (link)
the paired DNG and JPG as used in Post #23.
IMGP4490.DNG
and
IMGP4490.JPG

Please try it......

[kittykat46]

I think the software was doing more than "just" the traditional Auto white balancing play with light.
It was adding in green light (you can see green in the curtain at the back) to re-make the White/Gray on the guitarist's shirt, even though the original LED lighting had no green component.

That's a neat trick. I bet the software can do the same with any other "missing" light - Red, or Blue as long as the user is able to point to a Real White or Gray object in the picture as a base to work with.

Now....purists may not be comfortable with software doing that, because that's almost in Special Effects territory rather than White Balancing...but if it yields a good, usable photograph at the end, why not ?

[UnknownVT]

I think the software was doing more than "just" the traditional Auto white balancing play with light.
It was adding in green light (you can see green in the curtain at the back) to re-make the White/Gray on the guitarist's shirt, even though the original LED lighting had no green component.

That's a neat trick. I bet the software can do the same with any other "missing" light - Red, or Blue as long as the user is able to point to a Real White or Gray object in the picture as a base to work with.

Now....purists may not be comfortable with software doing that, because that's almost in Special Effects territory rather than White Balancing...but if it yields a good, usable photograph at the end, why not ?

You got it - that's what another poster suggested over at CPF that the Pentax DCU added green - but so did other software such as my regular/trusted/used for years editor and the result was this mess:


or this one from PS Elements 7.0 using the same grey-point selected on the same area on the white T-shirt


It is not as if any software could do it - in fact NO software I tried came anywhere near getting a natural looking shot - that was not my intention - I just wanted to show that there are some shots/lighting that resolutely refute to be manipulated to a "natural" looking one - all the regular BIG guns editors could not do in either JPG or the much touted RAW/DNG.

and it was totally by accident that I found that the supplied Pentax DCU actually did the very thing that I didn't think was possible.

The way you put it originally makes it sound so obvious and common knowledge that all one has to do
- or should just do is select a white point on a conveniently white object
and I was so lucky to have that T-shirt -
- please try it with any other software than Pentax DCU
and see if anyone can manage to get to a natural looking shot,
and please tell us on how it was done?......

Once again here are the originals for download:
in Kx folder (link)
IMGP4490.DNG
and
IMGP4490.JPG

Thanks,