Some background first, before going onto the PhaseONE IQ3 100 Megapixel Trichromatic Sensor/Camera. Never mind the contradiction of using gamut-clipping AdobeRGB color space for the pink jacket and pink suitcase woman—I’ll assume this is just ignorance in the marketing department.
Update: skepticism abounds on the trichromatic claim by PhaseOne. I’ve received four emails from 3 different readers expressing skepticism about the PhaseOne claims that follow below, including one comment about the diagram leading to loss of color discrimination where the sensitivity cuts out, and a general feeling that the claims to mimic human vision are rubbish.
I had taken the diagram as conceptual not literal (obviously there can be no such linear hard cutoff as depicted), and I would be very surprised if PhaseONE (long recognized for excellence) is going to bat on something that is less than stellar. The actual results will show us... the actual results, which is all that counts. Reader comments follow below. But the reader comments show that PhaseOne really needs to offer up something proper, like a spectral transmission graph.
What do I want in a Nikon D850? The best possible color, the best possible dynamic range and the best possible detail—even if that means ISO 32. That’s why Nikon’s description of the D850 as having balanced performance bothers me by using the word “balance”: I do not want to trade off optimal performance at low ISO for some “balanced” performance across the range. Here, “balance” means “compromised” as far as I can tell in marketing speak.
The Nikon D850 allows landscape photographers to capture a diverse range of scenes in sumptuously rich detail. It is the first Nikon D-SLR to use a backside illumination sensor, which allows incoming light to reach photodiodes more efficiently. Together with the camera’s low-noise performance, this enables it to achieve ISO 25600 despite its high pixel count. What’s more, it strikes an optimal balance between sensor sensitivity and the volume of light information accumulated in photodiodes, yielding images with a wide dynamic range even at ISO 64 (expandable to ISO 32 equivalent) — the lowest native ISO setting offered by any camera manufacturer. Copper wiring is used to cut electrical resistance, while the backside illumination structure allows a flexible wiring layout, reducing stray capacity. These measures enable 45-megapixel FX-format images to be captured at continuous shooting speeds of 9 fps*1. And because the sensor is designed without an optical low-pass filter, it can harness the sharpness of 45 megapixels when combined with the high resolving power of NIKKOR lenses. The D850 yields pictures that can be enlarged as massive
So what I want in a Nikon is the best possible image quality at base ISO, not a jack of all ISOs. What happens at ISO 200 on up, or even ISO 100 is uninteresting—I just want the best possible low ISO quality. Note that this very idea is called out below in the PhaseONE discussion below—kudos to PhaseOne. But why cannot Nikon (or Canon) run with similar ideas? Even Pentax at a distant 3rd has its limited-use-case but awesome super-res pixel shift capability.
That’s why the PhaseONE IQ3 100 Megapixel Trichromatic Sensor/Camera is intriguing and disappointing in this sense: why do DSLR users have to settle for “good enough”?
PhaseONE IQ3 100 Megapixel Trichromatic Sensor/Camera
I have to also wonder if the PhaseONE IQ3 avoids the nasty violet cut filter that most all cameras employ, which chops off the violet end of the spectrum, including some flowers. See Spectral Transmission of Digital Sensors.
The IQ3 100MP Trichromatic Digital Back is different because of the hardware. Designed around the concept of mimicking the dynamic color response of the human eye, we have physically customized the Color Bayer Filter on the 101-megapixel sensor to tailor the color response. This allows the Digital Back to capture color in a new way, unlike anything else.
Commissioned by Phase One and built by Sony, the Phase One Trichromatic sensor is only found within the IQ3 100MP Trichromatic Digital Back. This pioneering technology, allowing for the capture of vivid and vibrant color close to the dynamic color ability of human vision, sets the IQ3 100MP Trichromatic in a league of its own.
The technology inherent in the IQ3 100MP Trichromatic starts with the hardware. We have physically customized the Color Bayer filter on an all-new 101-megapixel sensor to tailor the color response. This allows the IQ3 100MP Trichromatic to capture color in a new way, unlike anything else. In essence, it’s designed around the concept of mimicking the dynamic color response of the human eye.
The customization of the Bayer filter material in the IQ3 100MP Trichromatic allows the sensor to capture, and thus produce, cleaner color separation of the red, green and blue pixels, particularly at lower wavelengths. Separating color at a sensor level from the time of capture, with little to no color contamination, provides improved latitude in the final image render. This separation allows for richness and control that otherwise cannot be achieved. Recorded RAW data is, therefore, able to supply a more faithful representation of color in the final file, making it possible to easily and accurately achieve natural results.
Prioritizing Image Quality – ISO 35 The ISO performance of the new IQ3 100MP Trichromatic sensor has expanded to a range never-before reached in a medium format CMOS sensor. While others may concentrate on higher ISO’s, Phase One concentrates first on image quality. A base ISO of 35 means the IQ3 100MP Trichromatic can produce the cleanest 101- megapixel image possible, and reaching to 12,800 provides the flexibility required by the world’s most demanding photographers.
- The filter array of the IQ3 100MP Trichromatic remains as a standard RGBG Bayer filter mosaic. As this sensor is customized specifically for color, the filtering material that defines the color response of each individual pixel has been customized to produce cleaner R, G and B values.
- The new sensor of the IQ3 100MP Trichromatic is designed to provide accurate colors more easily as the sensor can ensure the purity of color measure for each individual pixel. This is easiest to see in richer chroma as there is less contamination (and therefore improved additive color control) from neighboring hues.
- The IQ3 100MP Trichromatic, along with the existing IQ3 100MP and IQ3 80MP Digital Backs, produce 16bit color RAW files. The IQ3 100MP Trichromatic, however, is able to capture, control and therefore deliver these colors in a way that mimics the color response of the human eye, giving a more natural, pure result. This provides improved color performance, efficiency and post processing flexibility.
See the full description in the PhaseOne Trichromatic tech specs PDF.
- 101 megapixels at 11608 X 8708
- Up to 60 minute exposures.
- 15 f-stop dynamic range, 16 bit opticolor.
- ISO 35-12800
- 4.6 X 4.6 micron pixels
- Electronic shutter.
Kathryn R writes:
My apologies for not making my point more clearly. My point is that Phase One claims that their new sensor more closely matches human vision because it clearly separates the spectral sensitivity of the three color sensors. This is categorically NOT true in the human visual system. In fact, the spectral sensitivity of the green and red cone cells has major overlap.
You are correct that the visual system in the human brain/eye is NOT analogous to optical sensor in the camera. In fact the human visual system involves processing at both the retinal level and the brain. This accounts for many capabilities in the human visual system (see “Weber effect”, color constancy, etc.) that current optical sensors don’t begin to duplicate.
In fact, the best current data suggests that color perception in the human system responds to the relative difference in stimulation of the three types of cone cells rather than the absolute value of the cell outputs (which accounts for color constancy). Luminance data is instead supplied entirely by the rod type cells and therefore is integrated into the image at a higher level. Note also there are areas of the retina that are only populated by cone cells and more peripheral areas populated by only by rod cells. Again, we are in "violent agreement” that the human visual system has very little in common with the typical camera sensor …. but it is the Phase One marketing department that is claiming the superiority of their new sensor which allegedly “mimics” the human system which is rubbish (even though it may be a better sensor).
In fact, there is an argument to be made that because of the spacial separation of the color sensors there may actually be an advantage in some spectral overlap to help avoid certain types of color artifacts.
P.S. My background includes two engineering degrees and extensive research in neurophysiology in addition to my medical degree so it grates on me when companies misrepresent the science.
DIGLLOYD: this all makes sense, and I wish PhaseOne had chosen to supply a spectral transmission chart.
Philip S writes:
Laughably stupid, IMO. The PhaseOne blurb, especially the illustration, implies no overlap in spectral sensitivities of the red, green, and blue photosites. If that were actually true, the camera would be severely limited. For example, suppose the following sensitivities: Red: 700 - 580 nm Green: 579 - 490 nm Blue: 489 - 380 nm.
Then, monochromatic red lights of say 660 and 600 nm would be indistinguishable, since both would excite only red photosites. There is a good reason why the spectral sensitivities of human cone cells (and the red, green, and blue photosites of camera sensors) have overlapping spectral sensitivities. The overlap makes it possible to distinguish very fine differences in color.
You are being very kind to PhaseOne. To me that press release (and PDF) is a perfect example of marketing BS that fails to provide any real information and is positively misleading. And couldn’t they have come up with something better than “Trichromatic”. Hardly a distinguishing characteristic.
What, for example, does “cleaner color separation of the red, green, and blue pixels” really mean? Or, “separating color at the sensor level from the time of capture”?
What I think PhaseOne is trying to say is that they have worked with Sony to adjust the spectral sensitivity functions of the filter array in a way that increases the accuracy of color reproduction.
A clue is here: “In essence, it’s designed around the concept of mimicking the dynamic color response of the human eye.” But why the emphasis on “color separation”? Sorry for the rant. And, yes, I know all other camera companies are guilty of the same sort of advertising-speak. Perhaps I’m holding PhaseOne to a higher standard.
DIGLLOYD: I’m beaten at my own game of being skeptical of camera vendor claims!
Jason W writes:
What the simplified spectral response diagram suggests to me is we're getting a sensor that probably behaves more like Velvia film, which has more isolated RGB spectral response curves.
If this is true, we should probably take issue with PhaseOne's claim about a sensor that mimics the color response of the human eye, which has significantly overlapping spectral curves in the red and green cones. In this sense, the original MP100 might objectively produce images more akin to human eyesight. That being said, people often subjectively prefer rich, saturated Velvia-esque color images even if they are not accurate.
DIGLLOYD: PhaseOne’s communication has resulted in four skeptical emails, and zero believers. Seems like they have a problem on their hands in terms of marketing.
Roy P writes:
I read the comments from your readers about the new P1 Trichromatic, including someone who didn’t like the name(!). P1 called a b&w back “Achromatic”, so I guess “Trichromatic” was a handy marketing name. At any rate, maybe the P1 marketing blurb was not accurate relative to how the human eye works physiologically. But what it does seem to deliver are colors that are purer. I see it more along the lines of a Sigma Foveon sensor, although the technologies are not at all the same. (I have heard that Sony is working on a new generation sensor that takes the Foveon concept further – but that’s not what’s in the Trichromatic.)
Here’s the first comparison I’ve seen between the IQ3 100 and the new Trichromatic. Assuming they didn’t fudge anything, and the exposure and RAW conversion were identical, the Trichromatic image shows more vibrant, saturated colors. But this is something that is easily done with the older IQ3 100MP by simply bumping up saturation in Capture One. The Trichromatic shows more highlight details in most areas, even in the white areas, which is impressive. However, inexplicably, there are some bald spots in the Trichromatic image with very little detail, but the same areas show more details from the older IQ3 100 back. Now, that should NOT be happening – that makes me wonder if the two exposures were really identical.
The piece de resistance for this comparison is the one triangular cut glass that looks very blue in the Trichromatic image, but decidedly purple in the regular IQ3 100 image. When I saw this, my first reaction was, wow, that is pretty dramatic. My second reaction was, hey, how could it be this dramatic? I would have expected to have broken a little sweat studying and comparing the two images, before convincing myself that the Trichromatic was indeed better. With this one, the incremental improvement in the color seems too much relative to the incremental improvement in the sensor technology. Net-net, I do think this is an improvement, and I would expect it to deliver sharper images to the extent there are abrupt color transitions in an image (e.g., from red to green). If I were Phase One, that is probably the angle I would be emphasizing, not pitching that the new back is closer to becoming the human eye!
DIGLLOYD: while I see the color difference, it’s hard to understand without seeing the original scene.