Do you still render with fake colors?
Computer graphics are incredibly realistic today, but not a single color is really real.
Why is RGB fake?
RGB color values are neither producible nor reproducible. They just do not match real colors.
Real Digital Colors
Only spectral colors are lossless and respond to different types of lighting. They can be printed in full color, dyed and produced.
Welcome
[Google-Translation]
If you are interested in 3D visualization with real digital colors then you are in the right place. The goal is a CGI workflow that allows 3D visualizations to be created using a color space that corresponds to reality - without loss of color. CGI images should not only be displayed in full color on the monitor, but also be professionally printed and simulated under other lighting conditions. So that the presented virtual products look exactly like in the reality.
False colors
„Do you still render with fake colors?“
Computer graphics has developed rapidly in recent years. Drivers were and are the film and games industry. Therefore, it is not surprising that the tools are perfectly tailored for these purposes. Due to the increasing standardization of workflows and the relatively low entry prices, the CGI technology is now also used by many conventional industries. In contrast to the media and entertainment sector, whose products are purely digital, companies whose products are destined for the physical world are faced with a problem. Although the RGB color space used in all CGI workflows is able to reproduce unbelievably realistic and beautiful images, its limited color gamut does not reveal any color in the real world. This is because the RGB color space refers to the red, green and blue values of a particular input and output device. That is why the color representation of a monitor is often completely different from that displayed on another monitor, despite the fact that the image's RGB values are identical.
However, the exact interaction of the device-specific red, green and blue values can be specified in a "profile" which describes the color rendering of a device. For example, one speaks of a profiled monitor.
Not every RGB color gamut is the same size. If you want to map the color gamut of two devices by color profile to each other, this is possible only in the congruent intersection of the two profiles, so it is regularly lost colors. In addition, the color gamut achievable with the three color channels is generally much less than the color gamut perceptible by humans. Many are not aware of this fact. Others avoid the theme of color and manage with their calibrated monitors in their distress.
The fact is that all companies that use CGI these days consciously or unconsciously work with fake colors.
Color Loss
„Why is RGB fake?“
„Take a picture of a color and you've lost the color.“
RGB
SPEKTRAL
The RGB color space is omnipresent: photo and film cameras, scanners and monitors all work in it, and computer graphics also use RGB. If everyone works in the same color space is it a good thing, right?
However, as explained, the unprofiled RGB value always relates only to the interaction of the "individual", ie device-specific red, green and blue channels of a different color gamut. Unfortunately, each device has a different color capability when shooting or playing color images. For example, a photo shows different colors even on two identical displays. This problem can only be tackled with a regular color calibration (establishing the device's original state during profiling) and profiling. When displaying a profiled RGB image of a camera on a monitor, the color gamut of the camera profile is cropped by the colors that can not be displayed in the monitor profile. If you try to print such an image with correct colors you will notice that the color displayed on the monitor eventually once again will be cropped while transformed with the printer profile.
Professional colormanagement no longer helps against these color losses when using images across devices, because the description of a color behavior via RGB color space contains only color information, which colors the device can reproduce, but not which was the original, physically correct color. This is described with spectral values. A spectral value quantifies the energy of visible light wavelengths that are reflected by a surface. Thus, the entire, perceivable by humans color space can be described. A spectral value can be calculated according to the respective illumination to a correct color value. This is not possible with RGB values, since these are not based on physical reflection properties, but only on the "individual" red, Obtain the green and blue values of a particular device. However, since the scenes of a virtual image are calculated in the currently available renderers and shaders only with RGB values for surfaces and light sources, it is currently not possible to make a color-accurate calculation of a virtual image.
The digital photo reduced the color of the real world to the RGB color space of the respective camera or its color-managed description. The color loss is so final, since the actual color information as described above, by recording in the RGB color space is no longer included. The colorful images on the monitors play a perfect world for us - problems only arise when the correct color rendering of a product in a virtual image is expected.
The mistake is not due to lack of color calibration, but at the beginning of the chain - the color digitization.
RGB colors simply do not contain the necessary information about the behavior of a color, for example, to reproduce the correct color values of an object in a virtual scene with virtually inserted light sources.
Lossless
„Real Digital Colors“
With a device working in the RGB color space, such as a scanner or a photo, color digitization goes awry because important color information is irretrievably lost. Lossless color digitization goes beyond spectral data.
The sunlight is refracted at a glass edge and fanned out into its spectral constituents or wavelengths. The addition of the energy of the individual reflected light waves then results in the "color" perceived by humans. This is why we also speak of "additive color space" in all devices that produce color by superimposing light sources (eg projectors, monitors).
In order to digitize a color loss-free, the human-perceived range of 400 nanometers to 700 nanometers of the wavelengths must be recorded. For a realistic simulation of color in computer graphics, spectral color values as well as spectral color values are needed.
Only a spectral value is a complete color definition - a correct digital description of the color behavior of an object.
Render Pipeline
„Rendering with real digital colors“
In computer graphics, there are two components that are crucial for a color-accurate representation:
Materials and lighting
Ideally, one would have a Spektralrenderer that works directly with the color information in the form of spectra. Instead of wrong RGB color values, one would provide the materials and surfaces directly with their reflection behavior, ie spectral values. If the energy of the light sources used in a scene is also described spectrally, a color-accurate representation of the rendered scene is thus achieved. A pleasing exception is the renderer VRED from Autodesk, which can be operated in spectral mode, but unfortunately for various reasons is currently not usable for lossless color rendering.
Since all other renderers work in the RGB color space, the subject still does not seem to be feasible according to the state of the art today, although the subsequently corrected color representation in the industry devours immense sums of money.
Instead of only 3 channels (RGB), the renderings could also be implemented with 16 channels. Rendering on today's systems is no longer a serious problem. In addition to the classic offline renderers, implementation in real-time engines such as Unreal or Unity also makes sense.
It's not as easy as it sounds. On the contrary. There is a lot to consider and such work is very tedious. No reason not to start with it!
The possibilities of exact color visualization and full control of the types of light used would finally close the color gaps of the CGI. We support you in transforming your render pipeline.
When strong light hits colored bodies, a so-called "colorbleeding" develops. The reflected color of the object affects the environment.
Compositing Pipeline
Under certain conditions, it is possible to divide a RGB rendering into its components and convert it into a multispectral rendering. This can be particularly efficient if, for example, you can change the light at the push of a button without having to re-render 3D. Again, I have already created the first proof of concepts. Maybe this process is also suitable for your pipeline.
I support you in creating your compositing pipeline.
About HDRCOLOR.com
Here, two departments overlap: Color and CGI (Computer Generated Images)
As a Digital Artist I come from the field of computer graphics. My expertise covers the entire CGI creation process, focusing on lighting, shading, rendering, compositing, and VR / AR. For about 15 years, I have been creating high-end automotive visualization (Mercedes-Benz, Volkswagen) and follow the latest technical developments very intensively.
The topic "right color" has been my concern for a long time from the point of view of an artist. After several attempts to read me in, but I have repeatedly repressed it and dedicated to other exciting topics.
I only really understood the cause of the problem a few years ago. Through continuous cooperation with the color specialists of Caddon. Here I got to know the world of multispectral color measurement and the exact color representation on monitors and printers. Since seeing this workflow with my own eyes, I am learning the vast world of colors and can not leave the theme of color and CGI behind. That's why I created the website HDRCOLOR.com.
The future belongs to the real digital colors.
For me, digital transformation also involves the lossless recording, use and reproduction of colors.
With HDRCOLOR.com I want to make the color limitation of CGI transparent and understandable to non-color professionals.
Since both computer graphics and lossless color are highly complex interactions, there is currently no off-the-peg patent solution. Every company needs an individual approach and has special needs.
That is why I want to talk about CGI Color in addition to lectures on "CGI & Color" (Animago Munich, November 2019).
There are many ways to lossless color in the CGI. However, to go through these and to go through clean process steps is still pioneering work at the moment. But the good news is: who wants to can!
I like to exchange views on this topic and I am pleased about your interest.
Christian Frahm
Who is HDRCOLOR for?
Not every computer graphics company needs digital colors that relate to the real world. For example, the existing color workflows for media and entertainment are perfect. But who produces real products, the color can not be indifferent. From internal color matching between design and production to color-accurate representation in marketing, the right color can be a significant cost factor. For the ever-increasing digitalization of digital products, real digital colors are a must-have.
What does HDRCOLOR? mean
HDR stands for High Dynamic Range and is inspired by Paul Debevec's image-based lighting technology (IBL), which has long been used in computer graphics for realistic lighting. With HDR, it is possible to increase the dynamic range of a picture many times over - from the extremely bright sun to the deepest shadow areas, brightness information stored in this way, which is only available in the real world and which a monitor could not present in a rudimentary manner. This is exactly where the analogy to HDRCOLOR begins, because here you work with real digital colors, which are also not displayable, but whose lossless information is absolutely necessary for a color-accurate representation and light calculation. HDRCOLOR is based on lossless spectral color values.
Imprint
Responsible for content: Christian Frahm
Adress: Jasminweg 8 - 70597 Stuttgart
Email: christian at hdrcolor dot com
cell: 0172 94 180 95