1D X White Balance Bug NOW WITH SAMPLES...

[clicstudio]

Hmm. Perhaps if you posted 50 or 60 more of those samples it would help to diagnose the problem.

There are about 5000 "samples" on my website...
Http://www.clicstudio.com

 

[Skulker]

The "flash" WB setting is intended to work with speedlights that feed back info to the camera. It should be set at approx 6000K with other flashes. It could be that there is some sort of lack of feed back or faulty feedback between your kit and the camera.

You could set a custom WB. You can take a single shot of a white card and use that to set Custom WB. Then the WB should be set and be consistent. I do this using a color passport and get good consistent results, but any old 18% grey or light grey should do it fine.

Thanks for your comment. But unfortunately it seems nobody understands the issue here. It's not a white balance problem or color inaccuracy. It's random images with the wrong color...
Definitely not flash related problem. Because it happens with any WB setting even without a flash firing.

Well I certainly thought you were saying you were having problems with a "white balance bug". I'm sorry if I have misunderstood your problem that I was trying to help with.

I didn't realise you had tried "any WB setting" I apologies if my attempt to help has waste your time and for not understanding your "issue" better.

 

[wockawocka]

To rule out the flash setting variable I would do a test shoot of say, a beach ball but set the white balance with a grey card.

Then it'll be fixed and you can then see if it's truly variable.

 

[clicstudio]

The "flash" WB setting is intended to work with speedlights that feed back info to the camera. It should be set at approx 6000K with other flashes. It could be that there is some sort of lack of feed back or faulty feedback between your kit and the camera.

You could set a custom WB. You can take a single shot of a white card and use that to set Custom WB. Then the WB should be set and be consistent. I do this using a color passport and get good consistent results, but any old 18% grey or light grey should do it fine.

Thanks for your comment. But unfortunately it seems nobody understands the issue here. It's not a white balance problem or color inaccuracy. It's random images with the wrong color...
Definitely not flash related problem. Because it happens with any WB setting even without a flash firing.

Well I certainly thought you were saying you were having problems with a "white balance bug". I'm sorry if I have misunderstood your problem that I was trying to help with.

I didn't realise you had tried "any WB setting" I apologies if my attempt to help has waste your time and for not understanding your "issue" better.

You didn't waste my time. We are all helping each other here and trying to learn at the same time... But the original post clearly states the problem. It's white balance shifting in some images, not a white balance color issue.
The samples are contiguous images side by side showing the problem.

 

[Harry Muff]

On my screen, the left ones look more realistic. It could just be the flash head output varying slightly for some.

Is not about color accuracy. I'm trying to illustrate a random problem with the WB, not the accuracy.
20 images out of 500 are Different, that is the bottom line.
This happens everyday. Randomly.

Lightbulb outputs fluctuate and this can affect colour temperature. Did you take any breaks? Can you correlate those breaks or set-up changes with the times that the images varied?

 

[Skulker]

The "flash" WB setting is intended to work with speedlights that feed back info to the camera. It should be set at approx 6000K with other flashes. It could be that there is some sort of lack of feed back or faulty feedback between your kit and the camera.

You could set a custom WB. You can take a single shot of a white card and use that to set Custom WB. Then the WB should be set and be consistent. I do this using a color passport and get good consistent results, but any old 18% grey or light grey should do it fine.

Thanks for your comment. But unfortunately it seems nobody understands the issue here. It's not a white balance problem or color inaccuracy. It's random images with the wrong color...
Definitely not flash related problem. Because it happens with any WB setting even without a flash firing.

Well I certainly thought you were saying you were having problems with a "white balance bug". I'm sorry if I have misunderstood your problem that I was trying to help with.

I didn't realise you had tried "any WB setting" I apologies if my attempt to help has waste your time and for not understanding your "issue" better.

You didn't waste my time. We are all helping each other here and trying to learn at the same time... But the original post clearly states the problem. It's white balance shifting in some images, not a white balance color issue.
The samples are contiguous images side by side showing the problem.

Your first post clearly stated you were using flash white balance. I still suggest that a custom white balance might help, but it seems you do not consider that you can learn anything from my help.

and BTW like someone else said in some of the side by side comparisons you posted the image you consider has a problem looks better to me than the one you have fixed.

 

[clicstudio]

The "flash" WB setting is intended to work with speedlights that feed back info to the camera. It should be set at approx 6000K with other flashes. It could be that there is some sort of lack of feed back or faulty feedback between your kit and the camera.

You could set a custom WB. You can take a single shot of a white card and use that to set Custom WB. Then the WB should be set and be consistent. I do this using a color passport and get good consistent results, but any old 18% grey or light grey should do it fine.

Thanks for your comment. But unfortunately it seems nobody understands the issue here. It's not a white balance problem or color inaccuracy. It's random images with the wrong color...
Definitely not flash related problem. Because it happens with any WB setting even without a flash firing.

Well I certainly thought you were saying you were having problems with a "white balance bug". I'm sorry if I have misunderstood your problem that I was trying to help with.

I didn't realise you had tried "any WB setting" I apologies if my attempt to help has waste your time and for not understanding your "issue" better.

You didn't waste my time. We are all helping each other here and trying to learn at the same time... But the original post clearly states the problem. It's white balance shifting in some images, not a white balance color issue.
The samples are contiguous images side by side showing the problem.

Your first post clearly stated you were using flash white balance. I still suggest that a custom white balance might help, but it seems you do not consider that you can learn anything from my help.

and BTW like someone else said in some of the side by side comparisons you posted the image you consider has a problem looks better to me than the one you have fixed.

I said I used Flash and also ambient light. I didn't fix or process the images. I am merely showing the 2 images taken a second, or less, apart being different.
20 bad images out of 500 doesn't seem like a problem but 5% of anything bad is not good.

Again, this is NOT a white balance accuracy problem...........

 

[Loswr]

To confirm: you see this issue with the WB set to something other than Auto or Flash?

 

[clicstudio]

To confirm: you see this issue with the WB set to something other than Auto or Flash?

\
Yes, but it is always a "Fixed" WB. Never Auto.

 

[risc32]

can't help, but damn dude, good work. very nice shots of very nice subjects. what am i doing shooting weddings?

 

[clicstudio]

can't help, but damn dude, good work. very nice shots of very nice subjects. what am i doing shooting weddings?

Thanx , it's a hard job but someone has to do it...

 

[docholliday]

So what happens if you say, shoot a macbeth against a gray background, from a tripod in a dark room? 20 shots, each 5-10 seconds apart, fixed WB, manual everything? Then, use an eyedropper tool and measure the patches / background to see how much drift you're getting. Since it is a slow shooting process 10s apart (just because the strobe recycles doesn't mean that it'll push out correct color, unless you are shooting Bron or Profoto), the strobe should recycle fully and with no change

With strobes, I've had problems with most lights shifting +/-200K-400K in the past and that's why the high end lights are around. They are more accurate shot to shot. I am very color critical since slight casts and differences make the difference in the shots I do. The tints you show would be enough to drive me nuts. But, I shoot a MacBeth for every frame and I definitely don't shoot any PCB lighting products. Was loaned some for extra lights and gave them back quickly as they were trashing every shot sequence with inconsistencies. I know that the Einstein's are supposed to be accurate, but I still prefer Bron for true accuracy.

Ambient is whole 'nother ball of wax with the changing environment (the movement of a tree in the breeze has caused a reflected shift of color on subject's I've shot that made a few hundred K difference).

I use a Minolta Color Meter to do color temperature checks when necessary and dial into Kelvins - and still shoot a MacBeth in every shot when necessary. I've never had skin tone problems with RAW - especially in studio. The ColorChecker Passport has the patches for warming or cooling tones. The trick is to shoot it once at the beginning test shot, then get the temperature like you want it. Since the lighting doesn't change, it should be consistent shot to shot. So, I would automatically apply those settings to all subsequent shots until a lighting/set change and do it all again. There are reasons that tools like Capture One and Lightroom exists - if you don't "like" LR, you should quantify why and learn to work around it. It is a much better tool than Aperture for studio shooting, especially in tethered configuration. I can see exactly what each shot looks like a second after firing on a large monitor next to me that is color accurate and calibrated.

 

[Shawn L]

Skin tones always look like crap on RAW. Its hard to make them look natural. I like what I see on the jpg version and can't get even close even tweaking for minutes on a RAW file.

Some "off the top of my head thoughts" that might help -- though I make no promises =)

• RAW files are 14 bit, JPG are 8 bit; that means some form of conversion is done when the JPG files are written out in the camera from the RAW data on the sensor
• since you like the skin tones in JPG, but not in RAW, that conversion is clearly baking a white balance setting into the images for you
• but JPGs are lossy and the color space transforms they do are a function of the image content
• it's possible (and here I'm speculating) that slight differences in image content (more red background, more exposed hardwood floor), etc are causing the JPG file you get out of the camera to have subtle shifts in white balance
• do you know if you see the same issue if you shoot RAW; if you do, that would seem more indicative of a sensor issue;
• here, you're looking converted data, and the conversion process in question is known to be lossy and to be a function of image content; so, it might actually make sense that you're not always getting exactly the same results
• also note, ambient light due to electric lights is going to fluctuate at 60 Hz; it's possible that two successive pictures would actually get a different amount (and potentially a different quality) of light hitting the sensor

Shawn L.

 

[clicstudio]

So what happens if you say, shoot a macbeth against a gray background, from a tripod in a dark room? 20 shots, each 5-10 seconds apart, fixed WB, manual everything? Then, use an eyedropper tool and measure the patches / background to see how much drift you're getting. Since it is a slow shooting process 10s apart (just because the strobe recycles doesn't mean that it'll push out correct color, unless you are shooting Bron or Profoto), the strobe should recycle fully and with no change

With strobes, I've had problems with most lights shifting +/-200K-400K in the past and that's why the high end lights are around. They are more accurate shot to shot. I am very color critical since slight casts and differences make the difference in the shots I do. The tints you show would be enough to drive me nuts. But, I shoot a MacBeth for every frame and I definitely don't shoot any PCB lighting products. Was loaned some for extra lights and gave them back quickly as they were trashing every shot sequence with inconsistencies. I know that the Einstein's are supposed to be accurate, but I still prefer Bron for true accuracy.

Ambient is whole 'nother ball of wax with the changing environment (the movement of a tree in the breeze has caused a reflected shift of color on subject's I've shot that made a few hundred K difference).

I use a Minolta Color Meter to do color temperature checks when necessary and dial into Kelvins - and still shoot a MacBeth in every shot when necessary. I've never had skin tone problems with RAW - especially in studio. The ColorChecker Passport has the patches for warming or cooling tones. The trick is to shoot it once at the beginning test shot, then get the temperature like you want it. Since the lighting doesn't change, it should be consistent shot to shot. So, I would automatically apply those settings to all subsequent shots until a lighting/set change and do it all again. There are reasons that tools like Capture One and Lightroom exists - if you don't "like" LR, you should quantify why and learn to work around it. It is a much better tool than Aperture for studio shooting, especially in tethered configuration. I can see exactly what each shot looks like a second after firing on a large monitor next to me that is color accurate and calibrated.

Very informative, thanx! Unfortunately, I work in a completely different way. I am not as technical or picky as you are.
I have been using PCB flashes for over 10 years and never had a problem.
They might not be Profoto but they work for me.
I have one camera and one lens. No backups or filters or polarizers.
I don't shoot tethered. I work really fast, hand held and moving around a lot. My girls don't pose, they move freely but listening to what I want them to do. When I see something I like, I shoot it. I don't have time for elaborate studio setups, and my clients don't care for delays either. I work alone. No assistants or any help and just the model and me. Nobody else in the studio.
I setup my 2 or 3 lights, do a couple of test shots, and if I like it, we start shooting. That simple. No color meters, no color charts...
My problem is not the flashes or the way I work or the flash brand I use. I never had this problem with my previous cameras. There is something wrong with my 1D X and I just want to find what it is.

Why blame it on the photographer when it could be a defective sensor or processor... Nothing man made is infallible.
Specially technology.

I will call Canon and send the camera to be checked on Monday. We'll see what happens.

As far as LR. I love Adobe products. i've been using PS since 1993 but LR just doesn't do it for me.
I want to like it, I tried it, I even downloaded the latest trial version but it just doesn't work the way I like.
Aperture is so much faster at displaying images, comparing, smart albums... I love it... and since I've been using it for years now, I am comfortable with it. BWT, I am using version 3.2.4, not the latest 3.4.5 which I hate. All those grey icons are terrible and there are video glitches on dual screens... Apple should step up their game but it still works for me.
We will see Monday at the WWDC... Hopefully they haven't forgotten about us Pros and they will introduce a new Mac Pro and Aperture 4.0...

Again, thanx for sharing the way you work. I wish I had more time to be as precise and meticulous as you...

All the best,

P

 

[docholliday]

Very informative, thanx! Unfortunately, I work in a completely different way. I am not as technical or picky as you are.
I have been using PCB flashes for over 10 years and never had a problem.
They might not be Profoto but they work for me.
I have one camera and one lens. No backups or filters or polarizers.
I don't shoot tethered. I work really fast, hand held and moving around a lot. My girls don't pose, they move freely but listening to what I want them to do. When I see something I like, I shoot it. I don't have time for elaborate studio setups, and my clients don't care for delays either. I work alone. No assistants or any help and just the model and me. Nobody else in the studio.
I setup my 2 or 3 lights, do a couple of test shots, and if I like it, we start shooting. That simple. No color meters, no color charts...
My problem is not the flashes or the way I work or the flash brand I use. I never had this problem with my previous cameras. There is something wrong with my 1D X and I just want to find what it is.

It sounds a lot like you want precision and straight-out-of-the-can accuracy, but aren't working in a method to do so (and that you can't/won't try). It sounds more like random chance that shots will be right. The bigtime pros have shown that chance doesn't get predictable results, just watch some of the top fashion and model shooters and their workflow. Hand meters, color meters, MacBeths, tethering - it all serves a purpose and not just in product/commercial photography, in fashion/modeling too. I've shot my share of fast paced fashion and dance where the model was just doing their thing and my shooting when I see a good frame. I shot by myself in my studio - and I still did the pre-work of metering, color balance, etc. And, I shot tethered, wirelessly and wired. Frames dumping by the hundreds onto the tether box. Still got to see my previews and lost no workflow speed, nor hinderances with the model's flow.

I'm not saying to change your workflow - I'm suggesting that you take 10 minutes of your time and do a simple experiment to remove variables in it that may or may not be causing the issue.

Why blame it on the photographer when it could be a defective sensor or processor... Nothing man made is infallible.
Specially technology.

I'm not blaming it on you - I *am* blaming it on possible equipment issues. But, you insist that your lights are fine and I'm saying for you to verify that just to be absolutely sure. I know that your current lights aren't 10 years old (they haven't been out that long), so there is a possiblity that they have some issues (bad caps, drifting IBGTs, tubes fade, etc - I am also an electronics engineer).

Well, then keep doing things your way...there's quite a few of us here who have offered you a way of trying to narrow down the possibilities of what is going on and every time, all we hear is "that's not me, it's gotta be the camera" and "I don't like LR, but I've never really tried to use it" and "my lights are perfect, can't be them".
But it sounds like you'll just do what you want to do, so send it off to Canon and hopefully they'll find something wrong with it - that would be a nice speedy fix. Good luck!

 

[clicstudio]

UPDATE... RAW test with Temperature shift...

Since all of you think the issue is related to jpg, I did a RAW+jpg shoot today. (203 images)

Somehow, just 2 images were "off". But, the problem still showed in the RAW version.

The metadata in the jpg version shows 5000K (Tint 0) for ALL the images. Even the bad ones.
Now, in the RAW metadata, the Color Temp shown is 6341K (Tint 4)
The bad image shows 5885K (Tint -6)

Changing the Temp to 6341 makes it look more like the "good" image but it still not quite the same. Tweaking the red hues makes it look more like it...

So, at least in the RAW file I can see there is definitely a ~700K shift. Somehow too, it's always the same shift.

I will do another test tomorrow in RAW and see...

What do u all think? :-\

 

[clicstudio]

Very informative, thanx! Unfortunately, I work in a completely different way. I am not as technical or picky as you are.
I have been using PCB flashes for over 10 years and never had a problem.
They might not be Profoto but they work for me.
I have one camera and one lens. No backups or filters or polarizers.
I don't shoot tethered. I work really fast, hand held and moving around a lot. My girls don't pose, they move freely but listening to what I want them to do. When I see something I like, I shoot it. I don't have time for elaborate studio setups, and my clients don't care for delays either. I work alone. No assistants or any help and just the model and me. Nobody else in the studio.
I setup my 2 or 3 lights, do a couple of test shots, and if I like it, we start shooting. That simple. No color meters, no color charts...
My problem is not the flashes or the way I work or the flash brand I use. I never had this problem with my previous cameras. There is something wrong with my 1D X and I just want to find what it is.

It sounds a lot like you want precision and straight-out-of-the-can accuracy, but aren't working in a method to do so (and that you can't/won't try). It sounds more like random chance that shots will be right. The bigtime pros have shown that chance doesn't get predictable results, just watch some of the top fashion and model shooters and their workflow. Hand meters, color meters, MacBeths, tethering - it all serves a purpose and not just in product/commercial photography, in fashion/modeling too. I've shot my share of fast paced fashion and dance where the model was just doing their thing and my shooting when I see a good frame. I shot by myself in my studio - and I still did the pre-work of metering, color balance, etc. And, I shot tethered, wirelessly and wired. Frames dumping by the hundreds onto the tether box. Still got to see my previews and lost no workflow speed, nor hinderances with the model's flow.

I'm not saying to change your workflow - I'm suggesting that you take 10 minutes of your time and do a simple experiment to remove variables in it that may or may not be causing the issue.

Why blame it on the photographer when it could be a defective sensor or processor... Nothing man made is infallible.
Specially technology.

I'm not blaming it on you - I *am* blaming it on possible equipment issues. But, you insist that your lights are fine and I'm saying for you to verify that just to be absolutely sure. I know that your current lights aren't 10 years old (they haven't been out that long), so there is a possiblity that they have some issues (bad caps, drifting IBGTs, tubes fade, etc - I am also an electronics engineer).

Well, then keep doing things your way...there's quite a few of us here who have offered you a way of trying to narrow down the possibilities of what is going on and every time, all we hear is "that's not me, it's gotta be the camera" and "I don't like LR, but I've never really tried to use it" and "my lights are perfect, can't be them".
But it sounds like you'll just do what you want to do, so send it off to Canon and hopefully they'll find something wrong with it - that would be a nice speedy fix. Good luck!

I am not saying it's me or the camera or the lights... I just don't know what to blame it on.
All I know is that I work the same way I've been working for years and I never had this issue.
It started happening with the 1D X and that is ALL I know...

I did try LR. For a few days. Just not for me. The Einstein Flashes are less than a year old. I have 5 of them and aside from a physical speed ring problem in one of them, they have been the best investment I've ever made (I was using X-800's before them). They are 100% digital. Super light, small and I can change the output power directly from the camera for up to 3 flashes using Pocket Wizard Mini TT1 with AC3 and plug-in PowerMC2's on each flash... I can also see everything I need to about flash output and modeling lights on the huge LCD screen... And the firmware can be downloaded and installed via MicroSD card... Works for me

And yes, I do what I want to do. That's why I work alone and never had an assistant or a partner. My decisions are all mine. Everything that comes out of my studio is 100% mine. My good decisions and my mistakes, but that's how you are forced to learn and the way I like it and why my business has grown 1000% in just 10 years.

I have a lot of fun "working"... Although I never worked a day in my life I don't want to lose that feeling.

 

[docholliday]

Just for the record, I never said it was related to JPG. The compression in JPG would cause localized shifts, but not global shifting, unless the compression engine was wacked.

Don't worry, I've never "worked" either - I have a philosophy that when it starts to seem like work, it's time to quit. That's why I do everything myself too - if you want it done right, do it yourself!

 

[clicstudio]

Just for the record, I never said it was related to JPG. The compression in JPG would cause localized shifts, but not global shifting, unless the compression engine was wacked.

Don't worry, I've never "worked" either - I have a philosophy that when it starts to seem like work, it's time to quit. That's why I do everything myself too - if you want it done right, do it yourself!

U know it Doc!

So, now that I can read the temperature shift in the RAW file... What do u think it could be?

 

[clicstudio]

Skin tones always look like crap on RAW. Its hard to make them look natural. I like what I see on the jpg version and can't get even close even tweaking for minutes on a RAW file.

Some "off the top of my head thoughts" that might help -- though I make no promises =)

• RAW files are 14 bit, JPG are 8 bit; that means some form of conversion is done when the JPG files are written out in the camera from the RAW data on the sensor
• since you like the skin tones in JPG, but not in RAW, that conversion is clearly baking a white balance setting into the images for you
• but JPGs are lossy and the color space transforms they do are a function of the image content
• it's possible (and here I'm speculating) that slight differences in image content (more red background, more exposed hardwood floor), etc are causing the JPG file you get out of the camera to have subtle shifts in white balance
• do you know if you see the same issue if you shoot RAW; if you do, that would seem more indicative of a sensor issue;
• here, you're looking converted data, and the conversion process in question is known to be lossy and to be a function of image content; so, it might actually make sense that you're not always getting exactly the same results
• also note, ambient light due to electric lights is going to fluctuate at 60 Hz; it's possible that two successive pictures would actually get a different amount (and potentially a different quality) of light hitting the sensor

Shawn L.

Hi Shawn. I did a RAW test and found something interesting. Read my last post.
Thanx
P