-
FSI – trustworthy or not? Maybe plasma better?
-
When I received my 2461W, I was very surprised at what the colors looked liked. In short, it felt colder.
I then took the monitor around in quite a few film-rental firms here (Brussels), to compare the FSI and other top-of-the-line monitors. I was always surprised at the results, FSI vs. the rest of the world. Then I compared to one of the best DCI projection in town (Cinémathèque Royale), and guess who was right.. the FSI.
But this was almost an insult to me, so I dug into this quite a bit. How come my carefully expensively calibrated projector was wrong when the factory-calibrated FSI was right ? I exchanged dozens of emails with Bram, but I took nothing for granted, I had to check for myself. The answer “you need a high end spectro” did not satisfy me.
So in the end I rented a CS-2000, and compared it with my Hubble.
I then put the Hubble for sale (https://forums.creativecow.net/thread/277/11184#11184), and kept the FSI (I don’t own a probe anymore, I rent high end spectrometers every now and then).
With the CS-2000, you can aim a very very precise portion of your monitor : I used a 1 pixel cross to be sure to be at the exact center. What I found out was that by moving 30px away from this cross, you could have variations of several .001 xy (which translated into several delta-E -calculations by calman-).
I was not doing all this for the sake of science. I just needed monitoring I could trust. FSI won my trust through their support (which you’re experimenting right now), through comparison with the best digital screens here, and through measurements with the mother of all reference probe which showed a tolerance inferior to the tolerance I would get by being 30px off axis.
More simply than metamerism failure, I found out you can have good looking calibration data, and a poor image. So I became less crazy about Delta-E.
A few figures :
the spectral output of my FSI :
see how spiky it is ? Very difficult to see the color for a colorimeter or a low-end spectra.
the graphs of my measures. Please note I’m not claiming this is the truth about FSI (be it to defend or attack the product). This is what I measured, read above to why measures should be taken with caution when not done in a scientific lab.
They are not perfect, there are some quite big dE on the gamut. But overall I trust the image I get, I never have unpleasant surprises when our work is going in nice screens outside, and the FSI match my new projector (Projectiondesign).
Both projector and monitor are used without a LUT for Rec-709 monitoring, without gamma curves, and without playing with Color Hue Tint Saturation.
That said, more than a year has passed since my purchase of the FSI and the measures I post here, and I can see looking at my test pattern (https://www.charbon-studio.be/resources) that in the darks there has been a little drift, so I’m going to send it back for recalibration for free + shipping costs.I can feel the heat in this subject ; I’m both a client of Steve and of FSI, and I’ve started a few threads on this board that went all wrong with some people bashing and contempting. So apart from this post, I will stand aside and watch, hit me if you must 😉
Michael Cinquin
Final Cut Pro – Avid Media Composer editor
DaVinci – Color – Baselight colorist
Color profiles for Color
http://www.michaelcinquin.com/tools : tools for FCP | Color | RED | subtitles | Cinema Tools | Timecode – Keycode calculator -
Look again at the two charts below – top one is with Flanders CMS “off” and bottom one is with Flanders CMS “on”.
The “on” has all kinds of odd kinks in it – the off version does not – yet both were done by the same probe, on the same day.
it is my understanding that Calman misses a lot of issues in the way it looks at color – issues that have real practical implications for high-end work – it can fail to catch a lot of problems. The green gammut folding over itself is a clear example that can be missed by Calman when calibrating any monitor, not just the Flanders. Calman was designed to do 10 point grayscale calibrations for home theater – it does that very well. The other features, like gammut evaluation, are brand new, and work to a degree that has not yet been well-evaluated by anyone. People just assume they work – but I don’t know if they do or not – perhaps someone has tested them on this forum?.
it is my clear knowledge that an “inadequate probe” to the degree that anyone would call a Hubble inadequate – will make little difference – maybe a few tiny misreads here and there off by an almost or fully imperceptible amount – not this kind of kinked up mess that very coincidentally matches my eyes looking at a Flanders, and very coincidentally also matches the tell-tale signs of how a calibration using software like Calman can fail (although Steve probably knows way more about this than I do, so I hope he can chime in).
It’s not the probe that really matters, it’s the CMS circuitry to which the measurement is applied, and the software methodology used to create those measurements, and how the two work together.
A probe’s quality level, beyond a certain point, and the Hubble is well beyond that point, (you’ll see above I was questioning whether maybe even an i1Display Pro is good enough) – will only make an extremely subtle difference – and that difference will be located almost entirely in the darks. What you see above is not subtle, and is not contained only to darks. It can’t be a probe issue (if the probe was wonky, it would have delivered wonky data on both the “native” mode and the “Rec 709” mode – but only the calibrated 709 mode is squiggly – the native mode is actually in some ways better, at least, it’s perfectly linear, even though the primaries are off – in other words increasing luma arriving from the input creates smoothly increases luma being displayed without a color jump back – it doesn’t fold over itself the way the “calibrated” mode does. This folding over itself would mean that at some point, a brighter pixel in your source would be wrongly displayed as not just brighter, but another color using the Flanders “rec 709” setting).
Still don’t believe me?
Try this – make a gradient in photoshop of the pure color green – 20% luminance on the left, 100% on the right- but with hue and saturation left the same from left to right – Now look at it on your Flanders in the Rec 709 setting. Does it look right? I guarantee not. Now switch to “native” mode. See how it’s smooth with the processing turned off? (Even if the color itself is the wrong gamut, it’s now transitioning smoothly from left to right, as it should, as it is a linear gradient.)
It is possible that the Flanders is not properly calibrated due to poor software methodology used in measuring data and sending that data to the CMS. I hope that is not true, but it is looking that way.
I don’t think the Flanders guys are bad people and I’m sure they use the best probes available with superb intentions toward their customers, but until I see an evaluation done with truly high end software here (Lightspace being a good example, and I’m sure there are other examples, Calman no) I am reluctant to believe that their product can meet the very high standards I have for our gear.
Flanders probes are great – and they are great guys for sure – their software methods I have real doubts at this time.
-
It seems to me that $5k is fairly inexpensive for a monitor and that a simple way around these calibration issues for FSI would be to make a new firmware which allows for external calibration LUTs to be loaded into their monitor. That way an end user could have the choice of using the FSI calibration or loading a LUT from Light-Space, Cinespace or Truelight, etc.
On another note has anyone done a side by side of the FSI and the Sony PVM-2541 OLED? They are in the same price range.
-Rob-
Robert Houllahan
Director / Colorist
Cinelab Inc.
http://www.cinelab.comMAHC-PRO 6-Core 3X GTX285 20Tb SAS Wave Panel Panny 11UK SDI Plasma. Light-Space CMS + Hubble
-
Maybe I don’t know, and that’s cool, but as I understand it, the 2641 is a “wide gamut” display… and rec709, is again as I understand, a much narrower gamut than say P3/DCI/XYZ… So if you turn off the mechanism that controls the color space, you are going to get a “wider gamut” report.
And though I don’t have the spaces committed to memory your non CMS report looks very close to the P3/DCI/XYZ spaces.
So to me, the comparison of the CMS on/off is a moot argument. Because you’re arguing clamped vs unclamped.
It is also worth mentioning, there are a bunch of people on this forum that use FSi for Rec709 work, and I have not heard any similar complaints. I would mention some names, but I don’t want to drag anyone into this.
And it seems to me, that you are discarding answers that don’t align with your findings, which if that’s what you want to do that’s great. But after reading through the entire thread, it seems to me that FSi has gone to great length to answer your questions, though maybe not the answers you want.
-
i love my 2461w! i am doing mainly color correction for cinema, and i never had an unpleaseant surprise in the cinema, on the contrary… i grade in rec709 and let the post-facility do the dcp.
-
Lots of people trust Flanders monitors. The question is “Why do they?” I do not know of anyone who still trusts them after having a look at their actual performance via Lighspace or Cinespace.
In most cases, they trust them because they believe that they have been “super calibrated” at the factory and because the company gives great service and is run by quite obviously great guys. They do not check this “super calibration” in a technically rigorous manner, if at all – because part of what they are paying for is the impression that they do not have to.
But technical evidence points toward Flanders monitors being highly flawed. Now these flaws will not be obvious at every turn – only when you hit those particular colors that are in the worst parts of the kinked response curves, highly saturated reds and greens being among them. Maybe your clients won’t later notice an issue, maybe they will. Maybe you will never land on a highly saturated red color in your work. But to say that you can “absolutely trust” a Flanders monitor all the time is not actually true, according to technical, measured evidence.
It might appear to be true if you test just one image – an image that does not contain the problems we discuss here – and compare it to a more high-end offering. That image, free of the “problem areas” of a Flanders monitor, will look great “compare to a Barco” etc. and thus the reports. But when you compare it using the “wrong” image – well, thus the less stellar reports also found on this thread. In that way, they do not contradict each other. Both sets of people are truthfully reporting their experiences
Flanders has not actually addressed my concerns at all. They just keep telling us how expensive their probes are, and showing us charts from software such as Calman that, while very good at the specific things they do, are also lacking in any ability to reveal many serious potential color issues. (One could say that the philosophy behind Calman predates the digital display era, when kinked gamut curves were impossible to create, even on the cheapest CRT tube, as they contained no advanced digital electronics to potentially create them, and primaries were impossible to correct for the same reason – you could only control the mix of red/green/blue not the accuracy of those colors themselves as they were color-gel hardware placed over the three tubes. If the blue tube was actually closer to teal – well, there wasn’t much you could do)
As mentioned, even million-dollar probes mean nothing if you are not applying their measurements within a good software and good CMS circuitry context. So Flanders responses are not actually responses, they are topic switches.
The sad truth is, it is my current opinion that while many of us here very much want to believe that you can buy a “true” reference monitor that is $5000 and does not require any knowledge of calibration to use, that is simply not true. You can’t have it, not yet. Many others here very deeply DO NOT want to believe that their treasured $5000 Flanders is not as reliable as they thought it was, and that to some small degree they have let their clients down. Nobody wants to believe unpleasant news. The first stage of grief is always denial. I think this is why so many here do not want to accept simple technical observations, even if they happen to match their perceptual ones.
The good news is that the second stage of grief is bargaining – and my guess is that bargaining with Flanders will be easy.
I genuinely believe that the guys at Flanders have excellent intentions, and will do everything they can to make this right. The Penta guys, I have learned, made exactly the same error until recently, also in 100% good faith – they weren’t aware they were making the mistake, because their calibration software was not up to the task of informing them. But there were frequent reports that certain colors “seemed to look wrong” to which they replied, because this is what they firmly believed at the time “That’s because you’re not used to seeing a truly calibrated monitor.” or “LCD technology inherently looks different” Sound familiar? Yes, maybe those statements are true, but they should not actually have THAT much noticeable effect.
Truth is, the only response from Flanders that would satisfy me is at this stage is “We will look into this. We will fix it.” I mean, if I can afford Lighspace or Cinespace, so can they, although maybe fully fixing the problem would require new CMS ASIC chips – not sure (hopefully not, as I wish them well and wish them easy solutions to the problem). Whatever they are using now is evidently letting them and their customers down. I doubt they’ll want to continue on that path – even though even they, for a while, will not want to accept that they have made this mistake. They are obviously good people whom I expect in no way would find it fun to realize they’ve been misleading their customers, even if it was due to a 100% honest mistake.
P.S. “Clamped vs unclamped” yes that’s what it is. But here you are seeing unclamped – revealing/proving the ability of the probe to generate smooth readings, vs badly clamped and full of kinks and errors. A well implemented CMS system, based on good probe readings interpretation and good programing of LUTS into the circuitry would clamp the colors to something aligned with Rec 709, but just as smooth. Using a very cheap probe to do the job, but still otherwise good methods, you would also get a smooth clamp, just somewhat mis-aligned from the ideal – even a cheap probe would not generate kinks like this if the rest of the process was working properly. Again, Calman can’t properly show you problems like this.
-
I think there is a question being asked here and it is not how good the actual FSI monitor is but how good the FSI calibration is. From what I have seen the FSI is a well made LCD monitor for a reasonable price. I suggested that they allow for externally created LUTs to be loaded into their monitor so that end users could either use the FSI calibration or load a LUT from another calibration system, Cinespace, LightSpace, Truelight. FSI could call it a feature something like an “open” LUT standard.
Easy…. probably a firmware update as their monitor already supports 64x64x64 LUTs of FSI’s making….
-Rob-
Robert Houllahan
Director / Colorist
Cinelab Inc.
http://www.cinelab.comMAHC-PRO 6-Core 3X GTX285 20Tb SAS Wave Panel Panny 11UK SDI Plasma. Light-Space CMS + Hubble
-
I was saving money to get a FSI monitor… now I’m confused and concerned.
Confused about the monitors and concerned about FSI reaction.
I would love to see the FSI guys asking for time to check the situation and come back with a real and honest answer.
Never love a filmmaker…
-
I don’t think there is anything at all wrong with the FSI monitors, the questions are about profiling and calibrating them. I got the impression that FSI is trying to get the best our of their products. Remember a Dolby is about 10x the price compared to the top end FSI.
-Rob-
Robert Houllahan
Director / Colorist
Cinelab Inc.
http://www.cinelab.comMAHC-PRO 6-Core 3X GTX285 20Tb SAS Wave Panel Panny 11UK SDI Plasma. Light-Space CMS + Hubble
-
[Robert Houllahan] “I don’t think there is anything at all wrong with the FSI monitors, the questions are about profiling and calibrating them. I got the impression that FSI is trying to get the best our of their products. Remember a Dolby is about 10x the price compared to the top end FSI.
“Well if you have to profile them yourself, then a large part of what you are paying for is not there. You can profile any 10 bit LCD – and the Flanders does not have an internal way of sending your own LUT to it.
A Dreamcolor is just as good a 10 bit panel, and has internal LUTS – as do many other panels that are all way cheaper. WIth Flanders, you are supposedly paying extra for the “calibration certification” – saving you the time and expense of buying and learning software, and buying and maintaining a probe. That’s a lot of time and money. If Flanders delivered on their promise, it would actually be a really great deal.
But genuinely scientific evidence, as well as perceptual observations which concur with that evidence, show you that they do not deliver this. When I get both together, I tend to trust both.
Unfortunately Flanders has not said “We will immediately investigate this”, let alone “We will fix this” and none of their answers, which are their usual pitches about how great their probes are, along with displays of Calman charts (which cannot reveal many real potential problems) have directly addressed the issues our Lighspace examination revealed.
Reply to this Discussion! Login or Sign Up