-
FSI – trustworthy or not? Maybe plasma better?
-
Robert,
Attached is a screen shot of a simple Cal-Man report of the LM-2461W in Rec 709 mode vs Rec 709 color gamut reference as measured by a 5nm spectral bandwidth spectroradiometer. As you can see the Red primary is not inside or short of Rec 709 at all. However, if you put this next to an LM-2140W (our 8 bit White LED monitor with Red primary at exactly the same chromaticity coordinates) and compare 100% Red vs 100% Red on both displays most, though not all, observers see the White LED backlight unit’s red primary as perceptually redder (and see the white balance as redder for that matter as well). The issue is different spectral distributions and the limits of standardized color science models to truly predict/provide perfect metamers when dealing with such disparate spectral distributions. Search around and you will hear the same type of feedback with respect to OLED monitors and other new display technologies in that perfectly matching (according to very high quality probes) displays still look different when comparing different display technologies. In the days of CRTs all Grade1 CRTs had essentially the same spectral distribution so things were indeed simpler.
When FSI customers discuss perceptually matching to their high-end projectors we are talking about matching when operating in the same color space (for example Rec 709 to Rec 709 not Rec 709 vs DCI P3). The LM-2461W matches these projectors very well, the LM-2340W and 2140W less so perceptually. However, these less expensive White LED units do look like reasonably well calibrated consumer devices also using white LED backlights. If you are unhappy with our selection of backlight technology in the LM-2461W accounting for its spectral distribution and ability to perceptually match such projectors that is something I can live with, but what we won’t do is intentionally make the Red primary objectively wrong by pushing it out even further in Rec 709 mode.
Bram Desmet
FSI (Flanders Scientific, Inc.)
http://www.FlandersScientific.com -
I am very sorry if I in any way made you feel you were doing something wrong.
I WANT to buy a Flanders, and I want to just be able to say to myself I can simply trust it and not worry. i want the Flanders to be that legendary Sony monitor from before my time.
$5000 is not a lot to pay for that kind of piece of mind. I also like to support small businesses who push the envelope like you. We are a small business, and try to do just that.
But… Steve was referring to objective measures with a probe, using Lightspace CMS – not perceptual observations.
I have met you guys at tradeshows, and I have rock-solid instincts form an entire adult life spent in show business – there is no possible way you are dishonest – and I can smell a bad vibe right away, very, very precisely. Whatever you are saying here is your direct true experience- I have zero doubt.
But I also have trouble accepting that Steve is just wrong. So I am confused. How can both realities be reconciled?
Could this be a Calman vs Lightspace issue? Does Lighstapce’s LUT algoythm maybe better emulate human vision than Calman’s and somehow account for perceptual differences rather than just observe “technical correctness”? Isn’t perceptual color what we use to make grading decisions and thus what is most important? Have you tried Lighstspace? Do you use Lightspace? What are Lighstapce’s results on your monitors?
Maybe Steve can chime in here: Let’s assume (and I think we can safely assume) that Calman report of the is true. Does your Lightspace software work differently from Calman? Can it observe things that Calman misses, or does it somehow look at displays in a different way from Calman? (Assuming both are using adequate probes).
I can tell you that I just tried growing a white box on a number of high-en plasmas – the ABL effect makes them unusable for grading – it just does. So on that you and Steve both can agree – and we can conclude that we absolutely need a Grade-01 LCD to do our work – a plasma won’t do.
Like I said, I WANT it to be yours.
It may seem like I am prodding unpleasantly, but my goal in doing so is to assuage my concerns, mine and those of others here. I want the end conclusion of this thread to be that the informed user, regardless of their budget, wants to buy FSI.
Side question: Can I send a YUV HMDI signal to your monitor vis a DV-HMI simple/cheapie adapter?
-
Steve, can you post a chromaticity diagram of yours that shows the FSI red to be off-axis and under gamut?
Bram, can you clarify the diagram you posted? Does the diagram describe the CIE xy space or Rec.709? (I see the x and y values don’t go near 0.8 and 0.9 as I’m familiar with in CIE diagrams.) If it is CIE, do we see just one triangle because of the close overlap between the FSI and Rec.709?
So far it seems to me that the different opinions arise from different approaches to measurement.
Color is probably where the interaction between science and video production is the closest. If anyone wants to resolve these debates (which exist online by the thousands), there’s no feasible way but the scientific method, no matter how boring or daunting it is to us production folks.
-
I have tracked down a friend in London with a Hubble, Lightspace and a Flanders 2461.
This is his report of a 2461 – using Rec709 factory preset (he did it very carefully, and works in an extremely high end context):
As can be seen the green is over gamut, and blue is off axis.
Also, the white point is too blue, and was visibly so compared to a known reference.The green is folding back on its self as Luma increases, as is Red to a lesser extent.
This could mean the internal calibration is controlling the peak gamut, but not the lower gamut levels.
It would seems to show that the technical level of calibration being used by Flanders is problematic.
I hope Steve can elaborate on why Lightspace can show problems that Calman does not seem to be able to, based on FSI’s reports, as this is beyond my knowledge of how Calman and other software work.
He has since recommended removal of 2461s from grading suites.
Since he has no vested interest in knocking FSI I see no reason to mistrust his findings.
-
As I mentioned much earlier, but perhaps did not clarify enough there was a change when we released the 2461W as a replacement for the 2460W. We upgraded to several much higher-end and precise spectroradiometers than what we had been using before that time. Anyone with a 2460W could (still can) have their units updated with the new LUTs and have their unit aligned with these higher end tools to obtain an even more precise result. To put this into perspective below is are graphs of the old calibration result as measured by a Minolta CS-2000 and the new calibration result as measured by that same Minolta CS-2000. By the way, these measurements were taken by a third party using there own probes (TV Station evaluating display).
However, I don’t want to confuse the issue here. Even with the newer calibration there will be a perceptual difference between a CCFL backlight unit and a white LED backlight unit, even if they objectively measure 100% the same. Again, I think the video we posted early in this thread explains this as well as we can explain what is ultimately a rather esoteric issue. Long story short we offer different display technology options to meet different needs (and even preferences), but the reality is that we live in a world where there are an ever growing number of spectral distributions that even given identical calibrations as can be measured will lead to perceptual differences.
What is even more troubling is that different people can perceive the same spectral distributions differently. Best case I ever saw of this was walking into a room where two guys were looking at an OLED monitor with a flat white field on it. The one guy kept saying it looks green to me, the other keep saying it looked magenta. Same screen, taking turns sitting perfectly on axis, two very different perceptions. FWIW it looked green to me, but to my colleague (4th person to walk in, it looked magenta as well).
Bram Desmet
FSI (Flanders Scientific, Inc.)
http://www.FlandersScientific.com -
Again, let me say that if we could use an inexpensive probe and get accurate results we would. There is a reason we use higher end probes and as the CS-2000 measurements results I posted show the results with that probe are entirely different. Same thing when we measure with a 5nm PR-655.
Bram Desmet
FSI (Flanders Scientific, Inc.)
http://www.FlandersScientific.com -
Sorry, i see those earlier images may be hard to see. Hope this is better:
OLD CAL
NEW CAL
Bram Desmet
FSI (Flanders Scientific, Inc.)
http://www.FlandersScientific.com -
I understand these perceptual differences – but that does not explain the Lightspace report I posted which you have not responded to.
Your eyes do adjust to slight backlight temperature issues when they affect the whole image unilaterally, as long as that issue is subtle, but not to a particular color being under-gamut to a varying degree based on its brightness level.
This was done with a high-end probe – it is not just perception. The fact that the probe’s findings match observed perception does not take take anything away from them – on the contrary.
We grade based on observed color. EIther a monitor “looks right” compared to a very carefully calibrated rec 709 standard Dolby used to grade 100 million Hollywood features (the reference here) – or it looks different. If it looks perceptually different it is not much helping the colorist, regardless of why. If you are using some form of backlight tech that distorts perception then you should stop doing so. When all is said and done I grade with my eyes, not Calman and probes – so it’s what those eyes see, for whatever reason, that matters.
I guarantee if anyone here probes their Flanders with Lightspace and a high-end probe they will get the same report. Many of the world’s top facilities rely on Lightspace (much less often Calman)
If you do a probe on a calibrated Dolby at Company 3 you will get a virtually perfect report – not one like this.
They will ALSO perceptually look right – a properly white-balanced Alexa/Epic/F65 etc shot will show a red dress and stop sign known to be red as properly red, not kind of orange.
FSI, please respond to my London feeind’s calibration report of your new 2461, recently purchased.
BTW – here is the FSI monitor with the internal CMS turned off (i.e. the “automatic calibration feature” that lets you pick Rec 709, P3 etc.) turned off rather than set to Rec 709, as was the case in my post further above.
As you can see, it is still off, but less bad. Something is going wrong with the CMS electronics, the measuring method, or both, or something else altogether.
If anyone else has an adequate probe, a high-end profiling software (Lightspace, Cinespace, etc. – Calman I cannot comment on as to me it is more of a home-theatre application, but I could be wrong) and a recent 2461 please feel free to post your results here. I and my London friend are not making this up.
Steve Shaw, if you could please explain how these great calibration reports made with other software seem so different from those made via Lightspace – and why Lightspace seems to better align with observed color than probe reports from other software (such as Calman by F.S.I shown above – not sure what the TV station used). It would be much appreciated.
I for one am learning extremely valuable information from this thread and would like to continue doing so.
-
Robert, is it just me or are your diagrams in xy space while Bram’s recent diagrams are in u’v’ space?
-
A cie chart is always x-y.
Reply to this Discussion! Login or Sign Up