TechTalk Compiles All the Android News You Need in One App

If you’re as obsessed with mobile devices as the vast majority of our forum … more

Android App Review: Google Inbox to Improve Your Life – XDA TV

Recently, Google+ exploded with the announcement of Google Inbox, as … more

Damp the LG G3 Thermal Throttling

As our mobile devices grow thinner and more powerfulwith each passing generation, it’s inevitable … more

Google Play Developer Distribution Agreement Due Tomorrow

Every once in a while, Google updates theGoogle PlayDeveloper Distribution … more
Post Reply

[ANALYSIS] ASUS Eee Pad Transformer screen measurements

OP supercurio

25th July 2011, 12:28 PM   |  #1  
supercurio's Avatar
OP Senior Recognized Developer
Flag Chambéry
Thanks Meter: 5,072
 
3,529 posts
Join Date:Joined: May 2010
Donate to Me
Hi guys!

As you know EeePad Transformer has quite a capable screen, with excellent viewing angles and good response time.

I just wrote an application to ease screen measurements, Free on market and Open Source, named Screen Test Patterns

But you might ask why its color look desaturated despite its correct contrast ratio.
I measure mine at 593:1.

All this explains when looking at the measurement graphs.
Quote:



Luminance

Target 2.2 Gamma is the white curve.
Almost every color are displayed brighter than it should, which results in washed out colors.

Red curve is also brighter than Green and Blue ones, that makes the colors looks a bit reddish.

At almost white there is some clipping, fortunately quite minor.

Quote:



Gamma

Consequence of the luminance curve, the gamma curve is very far from following 2.2 reference line as it should.

Average on this screen is gamma 2, which explains by itself the washed out colors (even if it's not the only factor)


Quote:



Near Black



Near white

Near black there is no clipping but the screen uses a common trick to increase the visual contrast, by lowering the luminance of the darkest values (its called black point compensation)
That's not pure 2.2 nor fits to the sRGB model, but it's a nice detail improving the visual result IMHO.

Near white thing's are not right, there is some clipping. Means values 252, 253, 254 and 255 gives the same output. Not terrible in its amount but never welcomed.
Last edited by supercurio; 25th July 2011 at 01:51 PM.
The Following 4 Users Say Thank You to supercurio For This Useful Post: [ View ]
25th July 2011, 12:29 PM   |  #2  
supercurio's Avatar
OP Senior Recognized Developer
Flag Chambéry
Thanks Meter: 5,072
 
3,529 posts
Join Date:Joined: May 2010
Donate to Me
Now about colors:

Quote:



RGB Levels (compared to 6500K)

Here is individual Red Green and Blue channels values compared to what they should be to get a 2.2 gamma response at D65 / 6500K which is the industry standard for content display (web, videos)


Quote:



Color Temperature

Color temperature is mostly between 7500K and 8000K, but that's just numbers.
The most important in this graph is its linearity.

As you can see the white balance shifts to blueish in low between 0 and 15% gray.
This is because the black point of this screen (which is not black obviously) has not the same color characteristics as the other grays.
This screen doesn't really shines in this area, but its the case of most LCDs so there's nothing to worry here.
Still, it explains why dark grays look a bit blue.


Quote:



Saturation Lumiance

Saturation Luminance graph allows to detect artificial color boosting like by an image processing
This graph shows that the luminance of colors are fairly linear. Not perfect due to screen hardware limitations, but okay.
Only notable is Greens and Yellows which are often lighter than then should. Once again it reinforce the lack of color punch, especially since its the colors our eye is the most sensitive to.

Quote:



Saturation Shifts

100% Saturation is pure Red/Green/Blue/Yellow/Cyan/Magenta
This graph shows how much saturated are colors which are not at 100% saturation level (every non primary or secondary colors)

This graph shows that Red and Green colors lack saturation the most.
Blue saturation is a bit higher than it should in general.

I must admit I have no explanation why Cyan saturation is measured so high. It may demonstrate the presence of some sort of processing, but probably I don't understand the graph well enough yet.

Quote:



CIE Diagram

And here is finally the famous CIE Diagram.

The latest explanation for the dull color is here: you can see the gamut is a bit small compared to the reference sRGB behind.

A smaller gamut has for consequence less vibrant colors.
Blue is okay, still lacks some extend, but Red and Green extension is lacking, as perceived by naked eye.

The additional dots show the screen is not affected by any color conversion distortion.
Last edited by supercurio; 25th July 2011 at 01:28 PM.
25th July 2011, 03:30 PM   |  #3  
Senior Member
Thanks Meter: 537
 
2,499 posts
Join Date:Joined: Oct 2010
More
Wow, nice analysis there! I'm planning on getting a Transformer this fall for light school work stuff for ultra portability and long battery life.

So does this mean you are working on a Voodoo Color for the Transformer IPS panel?
25th July 2011, 03:43 PM   |  #4  
supercurio's Avatar
OP Senior Recognized Developer
Flag Chambéry
Thanks Meter: 5,072
 
3,529 posts
Join Date:Joined: May 2010
Donate to Me
Quote:
Originally Posted by jamesnmandy

Wow, nice analysis there! I'm planning on getting a Transformer this fall for light school work stuff for ultra portability and long battery life.

So does this mean you are working on a Voodoo Color for the Transformer IPS panel?

You're welcome.

Transformer Kernel has no trace of any drivers nor hardware capable of color correction.

It's not impossible Tegra2 display controller has a LUT but still unlikely as no Tegra2 device I saw offer "color vibrance" control or things like that integrated in their PC hardware and drivers.

At the contrary, Galaxy Tab 10.1 which is also Tegra2 adds another screen controller with a very advanced color correction engine (implementing even things like sharpening, noise reduction, dynamic contrast, chroma channels adjustments), not available in Transformer hardware.

Tegra2 is not especially documented.
Honeycomb is closed source (for a potential OpenGL engine)
So no not planned so far, it's just an analysis in order to better understand what/how/why, and have fun with the tool I just wrote.
Last edited by supercurio; 25th July 2011 at 03:50 PM.
25th July 2011, 06:22 PM   |  #5  
dennishylau's Avatar
Senior Member
Flag Hong Kong
Thanks Meter: 16
 
180 posts
Join Date:Joined: Sep 2010
More
Well written analysis. Although I don't have a transformer (maybe wait for newer generations?), I really appreciate your approach in the refinement of sound and visual on our android devices. Your measurements and analyses are some of the most, if not the most, detailed and informative ones that laymen like me will be able to obtain and comprehend. Thank you.
The Following User Says Thank You to dennishylau For This Useful Post: [ View ]
25th July 2011, 07:52 PM   |  #6  
Member
Thanks Meter: 3
 
85 posts
Join Date:Joined: Jan 2008
More
Great analysis.

Just a question: did you run the analysis on a B60 or B50?

In my experience, having owned the two, B50 has a very cold display, (as in your finding) while B60 is much warmer and pretty close to my calibrated monitor.
26th July 2011, 01:40 AM   |  #7  
supercurio's Avatar
OP Senior Recognized Developer
Flag Chambéry
Thanks Meter: 5,072
 
3,529 posts
Join Date:Joined: May 2010
Donate to Me
Quote:
Originally Posted by wis38

Great analysis.

Just a question: did you run the analysis on a B60 or B50?

In my experience, having owned the two, B50 has a very cold display, (as in your finding) while B60 is much warmer and pretty close to my calibrated monitor.

Oh really there are several screen revisions with different color characteristics? (only backlight or more profound?)

I'm not sure what you cakll B60 or B50. If it's about the serial number, mine is partially wiped from the bottom transparent sticker, but its seems its:
B40KAS057442 12

The screen on my device is a bit cool but looks pinkish despite what the T° in Kevlin would say.

I'm not totally confident in the temperature reading of my colorimeters BTW, in some conditions.
Please tell me all that you know about different screen revisions

Anyway it makes sense there are several screen revisions because my measurement don't match in terms of temperature DisplayMate ones.
Last edited by supercurio; 26th July 2011 at 01:59 AM.
4th August 2011, 07:56 PM   |  #8  
Senior Member
Thanks Meter: 627
 
3,558 posts
Join Date:Joined: Apr 2011
Quote:
Originally Posted by supercurio

It's not impossible Tegra2 display controller has a LUT but still unlikely as no Tegra2 device I saw offer "color vibrance" control or things like that integrated in their PC hardware and drivers.

Only just found this thread. Man, if accurate, that's seriously disappointing. Did you ever get any reply on Twitter from Nvidia re LUT info or documentation?

Post Reply Subscribe to Thread
Previous Thread Next Thread
Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes