Colorimeter measurements corrected from spectral readings

Late yesterday night, I completed the math and code to adjust my display high precision colorimeter measurements with spectral readings taken with a spectrophotometer.

During #MWC15 , I was going booth to booth with both a X-Rite I1 Display Pro colorimeter, that's particularly quick and a EFI-ES1000 spectrophotometer (same as a X-Rite i1 Pro), connected alternatively to the tablet running my software.

The spectrophotometer is about 3 times slower, but has the merit to see the light wavelengths intensities while the colorimeter uses some kind of RGB sensor that's able to tell the real colors only if their spectral characteristics match what it has been optimized for.

The correction logic was not implemented until now, so here are graphs from a +HTC One M9 unit, before and after correction.

Not only the spectrophotometer doesn't see the same thing, but these measurement confirm why you've heard reviewers mentioning a ''green tint" on the M9.
It's indeed here, and our eyes are very sensitive to an excess of green. We tend to be less bothered by wrong amounts of red and blue however. The M9 display also has too much blue and not enough red.

#progress #supercurioBlog #color #display #calibration #development

     

In Album Have correction from spectral readings now

Source post on Google+

+Charlie Demerjian, whom I had the chance to meet during #MWC15 at #Saygus booth while measuring the V² display wrote an article based on an investigation of what has been referred as the Qualcomm Snapdragon 810 overheating rumor

First of all let's establish something:
Declaring that a SoC "overheats" makes no sense today.
For several years, every single high-end/midrange SoC "overheats" during sustained usage and the only reason why the devices we use daily hopefully are not self-destructing and burning our hands is because of thermal management throttling the frequencies and voltages based on the actual temperature measured on the CPU, GPU, board and battery.
This is the role of power management and thermal management.

Now when the 810 overheating story came out initially, it was from Korean sources and linked to Samsung.
The first thing I thought then is that it was rather obviously a communication strategy to promote Samsung's 14nm incoming competing product.

Then after watching +Michael Fisher's videos about LG's G Flex 2: https://youtu.be/dkwsPf6KvR0, performance or optimization issues appeared more credible.

Then after playing with each Snapdragon 810 devices during #MWC15 , I mostly changed my mind on the subject:
– The demonstrated G Flex 2 performance was stunningly poor both on performance and UI smoothness
– The demonstrated Sony Xperia Z4 Tablet was even worse, being simply unable to run Android 5.0 system UI and launcher without massive dropped frames.
– The demonstrated HTC One M9 appearing not as fast nor as smooth as its M8 predecessor: dropping some frames when scrolling in Settings app, an UX regression impression increased by a higher touch latency.

So while I never bought the overheating rumor itself, since #MWC15 I concluded that it was likely that the Snapdragon 810 platform, both hardware and software at that time was suffering either from:
1/ noticeably poor performance tuning / frequency governor / power management optimization.
2/ a definite design issue, 4 A57 cores + 4 A53 + Adreon 430 GPU package on the same 20nm process leading to a power efficiency that's not good enough for satisfying performances in a smartphone.

1/ appearing a lot more likely than 2/

So while I still think that the overheating rumor was originated from an intent to promote as superior Samsung 14nm SoC, I also concede that Qualcomm made some damage by not communicating transparently enough on the fact their 810 platform power management and performance optimization was, or is still very much a work in progress.
Showcasing the LG G Flex 2 as proof the 810 had no issue was not the best choice.

We shall learn more about the Snapdragon 810 real world experience with the incoming HTC One M9 review, very soon.
The M9 review units have received a system update just a few days ago and hopefully it contains the latest and greatest Qualcomm tuning.

via +Stig-Ørjan Smelror via +Carlos-Cristian Radan

#supercurioBlog #critic



What is behind the fake Qualcomm Snapdragon 810 overheating rumors? – SemiAccurate
SemiAccurate has been following a massive FUD campaign for a few months and the rabbit hole has lead to some interesting places.

Source post on Google+

First article written for http://SpectraStudy.com is live!

Android Camera2 API promises to revolution digital imaging on smartphones or altogether with advanced capture and processing capabilities that have never been accessible to third party applications before.

Some of what becomes possible is manual controls, computational photography, RAW #DNG capture, full control over video recording, custom image processing. Limits are few.

Here's an analysis on how much of this new API is supported by the #Lollipop devices released during #MWC15 , including the highly anticipated Samsung Galaxy S6 and HTC One M9 flagships.

#supercurioBlog #camera #article #API #spectrastudy



Camera2 API on MWC 2015 devices: Galaxy S6, HTC One M9 and more Lollipop devices
Since Google announced Android 5 Lollipop at Google I/O 2014, mobile photography and video enthusiasts are hoping for a sizable upgrade of their smartphone’s camera capabilities. History The histor…

Source post on Google+

Here's the full list of the measurements files I got during #MWC15 over 5 days, as I just finished sorting and checking them

.chc files contains data formatted for HCFR
.amf is for "all measurements file" (YAML dump)
.3dm is for "3D measurements" (YAML dump)
-Reference=5 are the readings I make at the end with a spectrophotometer to serve as correction matrix for the (faster) colorimeter measurements.

For each measurement run, that's 602 colorimeter readings and 5 spectrophotometer readings (saving XYZ values and spectral data for later usage)

That's 24 runs in total, each taking something approximately between 25 and 40 minutes depending on the setup time.

It encompass also 13 different devices in total:

– Samsung Galaxy S6 (2 units, multiple modes)
– HTC One M9 (2 units)
– Nokia N1
– Gionee Elife S7
– Asus Zenfone 2 720p
– Asus Zenfone 2 1080p
– Saygus v2 (prototype)
– Huawei Honor 6 Plus
– Huawei Ascend 7
– Lenovo VIBE Shot
– Nexus 6 (I didn't have this one yet)
– Meizu MX4 Ubuntu
– Lenovo A7000
– Panasonic CM1 (2 units)

Now that's a lot of high precision data to analyze 🙂

http://pastie.org/pastes/10008275/text?key=yp0tsbbxyps64eqsqeavsq

#supercurioBlog #MWC #display #measurements



Private Paste – Pastie
supercurio / François Simond’s MWC 2015 display measurements list. Samsung Galaxy S6, unit 1, 100% brightness, adaptive display, 2015-03-01_20-02-03-Grayscale=256-NearBlack=11-NearWhite=11-Components-Saturations=17-3D LUT Generator=216.amf. Samsung Galaxy S6, unit 1, 100% brightness, …

Source post on Google+

Today, day 3 at #MWC15 was great!

I progressed with the measurements TODO:
– Huawei Honor 6 Plus
– Huawei Mate 7
– Asus Zenfone 1080p standard and with white point tuned to warmer, after the 720p variant measured in different modes yesterday as well
– Saygus V2: I spent so much time talking there too with a very open-minded product manager who selected most of the device's components
– Lenovo VIBE Shot
– Nexus 6, because I didn't had it yet

What didn't work out:
– Sony who refused anything I needed on all their Z3 and Z3 compact running lollipop, the Z4 tablet and M4 Aqua (not a very promising device BTW), I even got annoyed a little bit.
– LG GFlex 2: all of them were running a custom build for MWC, preventing WiFi to work properly which I need for my portable lab.

Then I had the chance to meet by accident +Jeremy Meiss​, +Chainfire​ and a nice +Paranoid Android​ developer I'm very ashamed to have forgotten the name.

There's still a lot in the list, but it was a productive day.

#supercurioBlog #MWC #measurements #display #color

 

Source post on Google+

Tonight I measured a Galaxy S6 unit in all its screen modes:

– Adaptive display
– AMOLED cinema
– AMOLED photo
– Basic

To anticipate some marketing or analysts claims, here are graphs and data representing the display of one unit, at 100% brightness.

Note: Those are based on preliminary results from a colorimeter only and I will apply some corrections based on the readings of a spectrophotometer.

If you look only at the 2D CIE 1932 gamut and saturations graph, the gamut and saturations seems to match rather closely to the sRGB or Rec.709 standards. It will affect the CIE diagram slightly but not the curves.

However it would be a mistake to claim that this display is color accurate to any existing standard, the reason being that the grayscale luminance and gamma response are wrong.
In fact, the average gamma ends up at 2.49 here which is really high: it makes things darker and more contrasty than they should (an approximate average gamma is 2.2)

So while the screen might look satisfyingly accurate if you look only at one particular graph, the Galaxy S6 display in Basic mode can't be trusted for color-critical work like video or photo editing.

Also, because of the correction required to reduce the saturation that's mechanically increased by a higher gamma, the overall appearance in Basic mode is inconsistent, and looking "off" to a trained eye.

There's more to say about the other modes but that'll be for later 🙂
The real #MWC15 starts in less than 8 hours!

Don't hesitate to point authors of a claim like "very accurate display", "most accurate ever" to the graphs attached.
They're making a mistake in their analysis: you can't look at only a fraction of the data, represented in a specific way and claim that it validates all the rest. But apparently it's a very common mistake.

#supercurioBlog #display #measurements #color #critic #Samsung

  

In Album About Samsung Galaxy S6 Basic screen mode

Source post on Google+