Why or how does a camera change the color of something
 

I've been working on duplicating an extremely rare factory anodized color option for 928 wheels. Long road. Today we got something extremely close. A lot of celebration.

Took pics for the customer. The pics show two very distinct colors.

In person, both outside in daylight and under fluorescent lights, the colors are nearly identical.

How does that happen?

http://forums.pelicanparts.com/uploa...1614909940.jpg


http://forums.pelicanparts.com/uploa...1614909940.jpg


http://forums.pelicanparts.com/uploa...1614909940.jpg

You're supposed to photoshop your sample images before you send them off to the customer.

Seriously, it all comes down to color gamma. The CCD in the camera can do a pretty excellent job with color gamma these days, but for some reason the camera chip is picking up something on your refinished sample and during the conversion from RAW to your chosen colorspace it's not blending the colors the way your eye does.

And, by the way, what is the color temperature of your fluorescent lights?

https://www.gtilite.com/2015/08/the-five-keys-to-accurate-color-viewing/

If I had to guess, Shaun, I'd say it has to do with the refractive properties of the respective coatings.

_

Yes, try with a polarizing lense if you have one.

A CMOS is a mechanical sensor that due to the nature of it's design can be sensitive to the angle and amount of incoming light energy.

There are 3 basic types of materials, diffuse reflective, retro reflective, specular.

They will affect how light energy is dispersed back to the sensor.

Diffuse reflective will scatter / disperse some of the light energy hitting it so less energy hits the sensor than hits the object, think concrete.

Retro reflective will concentrate and amplify energy directly back, think reflective road paint or road signs.

Specular, tends to redirect concentrated light energy, think laser hitting mirror.

So it is entirely possible to have two identically colored but different materials return different results.

It's not the color, it's the material properties of the two different coatings that are effecting what the sensor can interpolate. One may be rougher / smoother than the other.

I may know.

Eye ball visuals show virtually the same.

I worked for a company that did huge volume of anodizing. It was NOT my expertise.

But, some of the variables (as you know) are dye, crystalline structure of the anodic built up, any clear coating.

I think that you have visually duplicated with a slightly different process. Hence, the discrepancy between eyes and photos.

Cool, BTW.

We used to fight the same issue in the olden days of film. Some dyes and pigments just reflect the color spectrum differently to a piece of film or in your case the camera sensor. Modern digital sensors are amazing, but they still "see" in a different spectrum than your eyes. Sensors see more into the UV spectrum than the human eyes.

We had one very famous western artist come in with a fantastic portrait of a horse. The owners wanted to have posters made to use in advertising and promotional ways. I would set up the lights, photograph the original oil painting with careful color balance lighting on 8x10 transparency film, and go process it. One hour later we would all agree everything was perfect, except the nose was too green. If I added magenta to the filter stack the entire piece was magenta, except the nose. Finally the artist came in with his paints, repainted the nose of the horse to a more magenta color he hate. The new transparency was perfect. He painter the nose back like he wanted, and everyone was happy.

If you want the wheels to match, try a couple of different cameras just to see the difference. Then use Photoshop to adjust the colors to match.

Thank you Gentlemen! I will take pics with a Nikon D7200 and see how they turn out as well as a Canon G9X and compare. Then photoshop as needed. Good news is the shop doing the car, an incredible 6.2L supercar build, is local so they can see the tests in person. Still would be nice to post in 928 forums, etc. so PS may be in order.

Pics were taken with an iphone which has produced good accurate shots in the past.

Welcome to the world of color reproduction and proofing.

Think carefully about photoshopping your proof images - how do you argue your case when a client isn't happy with the ultimate results and the only fallback you have is an image that you've digitally color corrected?

The camera sees what it sees, and your best bet is to find lighting that your eyes and the camera both agree on.
I wouldn't be surprised that if you were to place the wheel and the sample in 5000K ISO 3664:2009 compliant lighting that you would be seeing the same color variant that the camera is picking up. Using the industry standard light spectrum means your color adjustments become direct comparisons.

Matching color is difficult, and unfortunately subjective. Your best bet is to always lock down as many variables as you can so you compare apples to apples.

I blame auto white balance. If you're taking a sequence of images, or want to compare things, pick a setting (sun/cloud/whatever). When I dumbly leave things on auto it's super annoying how the tone shifts a little from image to image.

Or throw a grey card in the corner of the image to guide post-processing.

I'll post three pictures here.
They show how a digital camera (all of them) tries to make everything grey. At first I didn't believe this, but after taking the pictures I can see it's pretty much true.

http://forums.pelicanparts.com/uploa...1614962203.JPG

http://forums.pelicanparts.com/uploa...1614962203.JPG

http://forums.pelicanparts.com/uploa...1614962203.JPG

Try it yourself Shaun. Fill the view finder completely with a black, then a white piece and see what your camera does with the image. Don't change any settings between shots.

Oh, if I were to Photoshop these pics, I would let the customer know with both raw and then PS'd. "How they look in pics and how they look actually."

I've never had a client unhappy with my results. :) That is actually the raison d'être of Tru6. And I actually spend my own time making something perfect vs. billing the customer. It's only right.

And on these, I couldn't have a disappointed customer because the color matches in real life. The kids say IRL i've heard.

But yes, matching color is difficult!

I will try that Scott, thank you!

I was at the anodizer this morning dropping off some window frames and we looked at the original wheel and the test piece again inside and outside. They are so close. I am going to call the dye company and ask them how to get a touch more gold into the color. You can't mix colors in anodizing without adding buffers and adjusting pH, etc., and we've already spent a ton of time getting this far, but I'm hoping they'll say to add "5% of this to add a tiny bit of gold." Really I should just leave it as is but those last few feet in the last mile are where the fun is.

Never forget the opposite of pretty good is perfect

does the D7200 have a choice of modes to alter the color? lower level (amateur) Nikons do but they considered the 7200 a pro body so it might not have that

another easy way to do this is to change the light source or use a filter on the lens

otherwise you need to do it in post with Pshop or Lightroom or whatever

and... the monitor used will affect how the colors come out so yours may differ from the customers

That's really impressive for a WAG that you pulled out of thin air.

Sadly and much to my chagrin, I have not learned how to use the camera. Have another tech coming in for a test on Monday, with 2 I might have time to actually learn how to use it properly.

Just another SWAG (Schitty WAG) .... does my green font show up on yer screen correctly :D?

I work in optics, and have background in getting digital cameras to properly recreate color and hue correctly...

Definitely a built in white balance problem with the camera. You need to find a way to output RAW if you're going to compare at that level of detail.

Try this...3 pictures, one after the other, with the camera on a tripod. Recreate your wheel image. First, take an image with both wheels, then take away the small piece, then take away both and only image the ground. The three SHOULD look identical as you flip back and forth, but they won't. The camera is looking at the histogram of each color channel and finding the median (or some other internally decided value), and setting that to be "white". The full wheel is getting more sunlight and a brighter spot, so it's probably deciding the color balance. The camera might even have zone balancing, so it's setting each differently.

With RAW, you should be able to go between the 3 images and see the exact same colors/levels/hues between each, which is when you're looking at the actual raw pixel level data. THAT is what you need. The files will be huge (on the order of a Gb each).

Looks like the Nikon D7200 has a "manual preset" white balance, use that, the manual says settings from 1-6, see what "1" and "6" do on a test image.
Also, set output to NEF (RAW) and 14 bit depth, instead of jpeg. JPEG files have their own set of internal calibrations that change the picture even more...
ALSO, see if you can get an image using manual exposure and aperture settings, the auto settings might change the values between shots as ambient light changes.

In other words, take that expensive hi-tech camera and make it stupid, all of the bells and whistles are modifying your image behind your back, and you're critical enough that you can see it.

When you work on Photoshop, keep it in some high-bit format, like TIFF (or RAW, if it can keep the image that way). RAW file format is 3 numbers per pixel giving the RGB amplitude and nothing else. It's the actual data that the CMOS detector picked up.

:(:(:(:(:(:(:(:(:(:(

Centering on grey allows the highest detail threshold for each picture.

While that is true, a lot of people don't understand that the camera is going to attempt to do that with every picture. They expect a picture to be black if the object is black and white if the object is white.

Read this under the dying heading:

https://en.m.wikipedia.org/wiki/Anodizing

I think Porsche may have used a tin based inorganic dye vs an organic dye.

Inorganic is far more colorfast.

Again, I am not an expert!

Another test today to get more gold. On wood in the sun the colors are very accurate using the Nikon 7200. You can see the original piece is very close and is extremely close in person. The test piece is a little more gold. Even though the first parameters are closer to the original, the customer has chosen the more gold version. I can't blame them, it's a beautiful, subdued bronzy gold.

One more test to replicate the gold on a perfect surface, then one more on an entire manhole cover assuming all goes well, the set of 4 9x16 Club Sports.

http://forums.pelicanparts.com/uploa...1615413442.jpg

On the full-frame nikons there's usually a "WB" button and you use a scroll-wheel to cycle through the options. My 700 and 610 have little "sunshine", "clouds", "lightning bolt (flash)" and other options, as well as a fully manual setting.

And if you don't have a grey card, the cap from a 35mm roll of Kodak film was 18% grey I think. ;)

There are so many moving parts to what you are trying to do.
You think you see a color, but the brain is pretty tricky and lets you see what you think you ought to be seeing. The camera will see it differently. It's sensors are different than yours and it doesn't have a brain that says, "I know that should be black, so I'll let you see black."
So even if you clear the first hurdle and the camera sees and records exactly what you see (not likely), you then you have to display what the camera sees on something. Show the same image on two different monitors and they probably will not look the same. If you have prints made, who knows what they will look like.
The best you can do is a decent approximation.

Shaun - Taking photo's of the wheel to compare different anodize techniques should be treated like an experiment. Control the variables as best you can. This would include the back ground used, the lighting used, the angle the light hits the object, the camera angle relative to the object and lighting, all camera settings (ISO, aperture, shutter speed), even the distance from the object too.
Once you can get images with all those parameters the same, as much as possible, then you'll be able to really see differences in the anodize formula used.
Just my opinion from out here in the cheap seat in CA.

Also keep in mind the digital camera may be recording light that the human eye is unable to see, and in the photo print, it is printing it , or showing it on the monitor different than you see it with eyeball 2.0

A good example of this is many digital cameras can see the ir transmitters that are used in remote controllers ie tv remote.

Got a old film camera? It records the photons that pass through and hit the film.

A good explanation why some producers miss the old film cameras, even though so expensive to film using them.

most of this I gained from a conversation with a production friend at WDW, so I may have not understood perfectly.