Depends on whether people’s faces were universally bright red or not. :-D
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
Colour accuracy of a picture
- Thread starter Marius Bocsa
- Start date
RL Photography
Forum Enthusiast
- Messages
- 407
- Solutions
- 1
- Reaction score
- 402
The same thing that has always happened when an image was turned over to someone who was not there and does not have a reference to make a print. Maybe it matches closely, maybe it mostly matches, maybe the colors are barely close. That is just based on what the operator might do. There is also the issue of whether the color gamut of the output media can match that color of turquoise.
Craig Gillette
Forum Pro
What I'm suggesting is that given the idea that "accuracy" might or might not be in the eyes of the beholder, a certain amount of consistency would still be a goal for at least some applications. Say Yearbook photos, our colors were orange and black for the gowns and collars. They shouldn't look black and peach. But even less would it seem ok to be 2 peach, 8 orange and 4 red.
RL Photography
Forum Enthusiast
- Messages
- 407
- Solutions
- 1
- Reaction score
- 402
Perhaps you did not read the last word of my post: turquoise. The water in Peyto Lake is an unusual color of turquoise from the suspended rock particles in the water coming from the glacier. People can read all about it and see a (hopefully accurate) picture here: https://en.wikipedia.org/wiki/Peyto_Lake
Your alleged “got ya” really isn’t one, however. While the color of the water in Peyto Lake is somewhat unique, the problem of getting a third party to make an accurate color print is not at all. Whether it’s the color of water, the actual shade of blue of some man’s shirt, the yellow in some woman’s dress or the particular red in a fall leaf, people have always been at the mercy of both the skill and taste of the operator as well as the output gamut of the medium. It is not a digital-only problem. The answer to your question today is the same as when a tourist came back with a roll of Kodak Gold 30 years ago: it depends and good luck.
Last edited:
Mark S Abeln
Forum Pro
Look up “color appearance models”. It turns out that accurate colors in a photograph often look dull, flat, and dissatisfying. Under dim lighting—such as on a digital display or a print—both contrast and color saturation are diminished, and even some hues shift strongly, such as yellow becoming olive.
If you don’t take color appearance into account, color that is purportedly accurate in one relative circumstance will appear inaccurate in other circumstances.
There is a really good reason why cameras oversaturate images and have unrealistically high contrast.
I assume Hasselblad is the closest to "most realistic color images".
The short answer is to buy one of these for your camera, as far i know this is the only way to get accurate colors from capture to screen:
https://calibrite.com/us/product/colorchecker-passport-photo-2/
https://calibrite.com/us/product/colorchecker-digital-sg/
https://calibrite.com/us/product/colorchecker-display-monitor-profiling/
If one have the money and is obsessed with quality and colors buy a Hasselblad. They goes to great length to calibrate the sensors, an H-camera takes 6-8hr to assemble and 1/3 of that time is used to calibrate its sensor.
I agree to what this user shows:
Agreed, but to the OP question the SMI ISO17321 does not come into play as the RAW is not considered in this measurement and does not indicate how accurate the picture representation is for the end user.
So for accuracy i assume that HB is still the best but the H-cameras are clunky and slow but users as Annie leibovitz etc. dont care. These H-system are only made for the Professional users that are 100% focused on the end result and could not care less that the camera and lens is 8lb and slower than an constipated turtle.
The Hassy X1D series is also old tech, but its the very best camera i ever held in my hand its a much better design than my GFX100s. I went Fuji because of the lenses are more suited to my use and i am waiting only for the GF20-35 this autumn, ive already pre ordered it. The new Hassy X2D might be just as good or better than the GFX100s and hopefully HB makes a long zoom and a wide zoom as Fuji, but i can not shoot vaporware so i had to go with the Fuji :-(
Last edited:
bobn2
Forum Pro
Is there any evidence to back up this assumption?
This is marketing material. There is no reason to believe that the calibration procedures used by Hasselblad are any better than other companies, which have greater resources and produce optical metrology products for which accurate calibration is essential.
So far as I can see, he's comparing default out-of-camera JPEGs. I would take it as a given that anyone really concerned with colour accuracy will use a raw workflow.
That depends entirely on the workflow of the end user. I took the question to mean 'which camera is capable of producing the most accurate colour'. As above, I think the use of a raw based workflow is sensible for anyone trying to achieve something particular with respect to colour, including accuracy.
As I said, more evidence more than their marketing material would be needed before one can assume Hasselblad is best. Given that Hasselblad sources their sensors from the same supplier as most of the industry, and they have the same CFA characteristics, it's very likely that one can get the same results by setting up one's own colour profiles for most cameras, without the expense.
tbcass
Forum Pro
I had never heard of Peyto Lake so I looked it up. According to photos on the web, the color varies from blue to turquoise. Whether any of these is accurate is open to question.
Peyto Lake - Bing images
tbcass
Forum Pro
There is one reason to believe it. The cameras are very expensive and production is limited so the cost of accurate calibration could be figured into the cost.
bobn2
Forum Pro
I don't think that's a very convincing reason. 'Could' does not mean 'will be', and in any case, calibration is not expensive if automated, which may well be done on a higher volume line.
I'm also not clear on what they mean by 'calibrating the sensor'. It doesn't say what it means in the video that you linked, which is everywhere posed at a very low technical level.
Michael Fryd
Forum Pro
First of all, don't confuse "realistic" with "accurate".
Are you trying to create an image that scientifically measures as the same color as the original object, or are you trying to create an image that looks to a human to be the same color as the original object?
.
If you want the same color, then get a standard color checker target, create a profile for your camera for each lighting condition, shoot in raw, and process the raw file using that profile.
.
If you want it too look correct, then you have a different ballgame.
Assume you are shooting a scene that contains a bright neon pink color. Assume that this green is outside the color gamut of your output device. If you shift all the colors to bring that pink into the color gamut, then the image will look correct. If you leave the other colors where they are, and use the closest available color for the neon pink, the image will look wrong, but be more accurate.
You also have the issue that perceived colors are dependent on ambient lighting. both the lighting of the original scene, and the lighting for the print.
For instance, if you measure the color spectrum of the light coming from a white piece of copy paper in bright sunlight, you will find a peak in the blues. Move that piece of paper indoors, and have it illuminated by a traditional tungsten bulb, and you will find that the peak is now in the yellow. Same piece of paper, different light coming the paper, yet humans perceive both as being the same color.
.
Another issue is that the real world is four dimensional (length, width, height & time), and photographic prints are typically two dimensional. In the real world, the look of an object can be constantly changing as things (including the viewer) move.
Real world objects have texture. As your head move, the texture causes the look of the object to change as the viewing angle changes. That doesn't happen with prints. You need to find lighting that suggests the texture to the user, even though the look does not change with angle.
You have a similar problem with items that are reflective. The reflections change as your head moves. In order to make the image look accurate, you need to pick a set of reflections that suggest reflectivity to the viewer.
Imagine taking a photo of a product with a chrome bar. That bar is highly reflective. You don't so much see the bar, and see the things the bar is reflecting. If you surround the product with a solid blue background, you may end up with something that looks like a blue background. Generally, one would carefully surround the bar with white and black pieces of foamcore (or similar objects) in order to create reflections that make the viewer think they are seeing a reflective chrome bar. Obviously, the bar in the print won't be reflective, so we need to curate a image that makes the viewer think he is seeing a reflective object.
tony field
Forum Pro
The pictures are accurate. Colour varies according to nature of the glacier run off. It was a lovely place to go, particularly in the winter for ski touring. I never found it interesting for photography. Like all glacier fed lakes in the Rockies, the lakes are becoming less beautiful because of the receding glaciers.
tbcass
Forum Pro
I didn't link any video and my argument is every bit as convincing as yours. Neither of us knows. It's just speculation and that was the purpose of my response.
bobn2
Forum Pro
My apologies, I failed to spot that someone else had taken up the argument. I was referring to the video that the other guy linked.
You're a bit confused. All I was arguing was that it's just speculation. 'There is no reason to believe' is not the same as 'it is definitely untrue'. You might think that a high prices is by itself reason to believe that things are done better. I don't. Might be. Equally, might not be.
Typically, an outdoor scene is lit by lights of two different colours -- sunlight and skylight.
Iliah Borg
Forum Pro
Doesn't work.
Yes i would say i have evidence to back up the assumption.
The Hasselblad H6D actually dont have any diffrences on the 140 patch X-rite that is possible to observe by humans as shown further below by the Rijksmuseum.
Here are three independent tests with "X-Rite ColorChecker SG color target" the one used by pdnonline have fewer patches than the one used by heritage-science-journal and the Rijksmuseum. These three use mostly very similar products and also give up delta E numbers. I assume ! pdonline use the same standard dE as the industry standard.
Exploring the limits of color accuracy in technical photography - npj Heritage Science
The growing importance of publishing art collections online has led to increasingly strict tolerances on digital photography of art objects. There are two internationally recognized sets of guidelines for creating high-quality digital images, Metamorfoze and FADGI. These guidelines require using...
heritagesciencejournal.springeropen.com
This is with a Hasselblad camera:
"increasing the average for all patches from dE(CIE 1976) = 0.88 to 1.24. The strictest guidelines for camera profiling using the Metamorfoze guidelines demand for each of the neutral patches that dE(CIE 1976) ≤ 2.83 [12]. This is clearly not fulfilled when comparing data from different spectrophotometers, making it impossible that camera profiles will satisfy this demand. There is no problem yet with the Metamorfoze guideline that the maximum color difference needs to be dE(CIE 1976) ≤ 10.0 and the average for all 140 patches should be dE(CIE 1976) ≤ 4.0"
This is the best out of 19 different Cameras tested:
View attachment 3130351
The Rijksmuseum own measurement is extremely impressive, number values under 2,3 is not detectable by humans. Numbers less than 10 is considdered "archival".
View attachment 3130213
Hasselblads claims semms by the two examples above not to be without merit and their widespread use in the museum heritage industry etc would have quickly shown if it was only a "marketing material " trick.
This is probably what the OP and most users would experience so the example stands its ground regarding the OP question.
Is "evidence" above enough ?
Attachments
Similar threads
Keyboard shortcuts
- f
- Forum