10 bit Gradient Test Patterns - Page 3 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 32Likes
Reply
 
Thread Tools
post #61 of 73 Old 08-03-2016, 09:39 AM
Newbie
 
Join Date: Aug 2016
Posts: 1
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 0
Quote:
Originally Posted by WiFi-Spy View Post
There have been a lot of people asking me how they can tell if their UHD TV has a 10bit panel or not, so I made some 10bit Gradient test patterns. These will show discreet steps on an 8bit panel.

The test patterns are 2160p 10bit HEVC gradient grayscale patterns. There are two different clips, one is a gradient from black to white, and the other is a gradient from black to about 75% gray.

The clips are only 10 seconds long, so you might need to pause them so you can get a good look up close to the screen.

I have included both .MP4 and .TS versions for compatibility reasons. Both versions have the same elementary streams inside, they just have different containers.

Here is the link:


Hello Tyler,

I'm a post-production artist wondering if I could find any before ( raw frame originated from camera like RED (OR) a 16 bit test pattern created in Photoshop) and after ( same frame rendered out in Rec. 2020 but in 16 bit) versions of any single frame or a clip. Since this is what happens in Post Production, and the encoding happens after the fact depending on OTA, OTT or LOCAL. These before after frames would be great help for Post Production artists to at least get the technical aspects in place. One of the main problems we have is lack of Reference Monitor with full Rec. 2020 support and no proper info on what Working LUT we would be looking through to judge the colors while working in Color Grading software! I have spent considerable amount of time to search for solution before posting here...

Any help or suggestion would be greatly appreciated!


Thanks
SrKay
trapcode is offline  
Sponsored Links
Advertisement
 
post #62 of 73 Old 12-20-2016, 03:38 PM
Member
 
ultrasilent's Avatar
 
Join Date: Aug 2010
Posts: 159
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 94 Post(s)
Liked: 16
Both clips from first post looks like some kind of fake for me.... I test with them my 8 bit player and 8 bit TV and got smooth gradients where they should be smooth for 10 bit setup only.
How it is possible?

Last edited by ultrasilent; 12-21-2016 at 10:06 AM.
ultrasilent is offline  
post #63 of 73 Old 12-21-2016, 07:28 AM
Member
 
ultrasilent's Avatar
 
Join Date: Aug 2010
Posts: 159
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 94 Post(s)
Liked: 16
I got this 10 bit test from another source... please test it with your systems...

https://cloud.mail.ru/public/6zy1/AZA9g884s
ultrasilent is offline  
 
post #64 of 73 Old 12-21-2016, 02:18 PM - Thread Starter
AVS Forum Special Member
 
WiFi-Spy's Avatar
 
Join Date: Feb 2004
Location: Seattle, WA
Posts: 2,809
Mentioned: 58 Post(s)
Tagged: 0 Thread(s)
Quoted: 521 Post(s)
Liked: 966
Quote:
Originally Posted by ultrasilent View Post
Both clips from first post looks like some kind of fake for me.... I test with them my 8 bit player and 8 bit TV and got smooth gradients where they should be smooth for 10 bit setup only.
How it is possible?
Your TV could be dithering

Tyler Pruitt - Technical Liaison - CalMAN

10 Bit Gradient Test Patterns (HEVC) - Free Download
WiFi-Spy is offline  
post #65 of 73 Old 12-21-2016, 02:57 PM
Member
 
ultrasilent's Avatar
 
Join Date: Aug 2010
Posts: 159
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 94 Post(s)
Liked: 16
Quote:
Originally Posted by WiFi-Spy View Post
Your TV could be dithering
Sooner, PC is dithering, because the TV gets only 8 bit signal. Anyway, if so, this test not fulfilling it`s main task - to identify real 10 bit video path.

I have real problems in Premiere pro to encode in proper file this test:

https://cloud.mail.ru/public/GQEp/R6su4jbi9

Can you please help me? Can you do same job as you did at the beginning of this topic.
ultrasilent is offline  
post #66 of 73 Old 01-10-2017, 09:14 PM
Member
 
moheban79's Avatar
 
Join Date: Jan 2014
Posts: 27
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 11 Post(s)
Liked: 14
Is Samsung F8500 10 bit?

Anyone know if F8500 plasma is ten bit?

Thanks
moheban79 is offline  
post #67 of 73 Old 04-10-2017, 05:45 PM
Newbie
 
sf49er's Avatar
 
Join Date: Apr 2004
Posts: 4
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 10
going to plead ignorance here, just curious if the panel is 10bit, what I should be expecting to see on these tests. I ran the gradient and spears test and in both instances, the gradient was smoother for the 10bit side, the 8 bit side had banding whereas it was noticeably more nominal on the 10bit side. As I understand it, does this reflect the possibility it is 10bit r indicate against that? hould there have been banding on the 10bit side?
sf49er is offline  
post #68 of 73 Old 04-10-2017, 09:52 PM
Member
 
Quad5Ny's Avatar
 
Join Date: Apr 2007
Posts: 196
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 47 Post(s)
Liked: 34
Quote:
Originally Posted by SnipeUout View Post
How do we know that the gradient was smooth because of a 10-bit panel and not because of dithering?
The difference between 10bit and 8bit+FRC is academic at normal viewing distances. They should both look the same.

Stick your face directly up to the screen and watch a solid color. Depending on if you get a dithered color or not, then you will see a very light (moving) dithering pattern.
Quad5Ny is offline  
post #69 of 73 Old 07-22-2017, 02:10 AM
Newbie
 
Join Date: Jul 2017
Posts: 3
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 0
Quote:
Originally Posted by WiFi-Spy View Post
There have been a lot of people asking me how they can tell if their UHD TV has a 10bit panel or not, so I made some 10bit Gradient test patterns. These will show discreet steps on an 8bit panel.

The test patterns are 2160p 10bit HEVC gradient grayscale patterns. There are two different clips, one is a gradient from black to white, and the other is a gradient from black to about 75% gray.

The clips are only 10 seconds long, so you might need to pause them so you can get a good look up close to the screen.

I have included both .MP4 and .TS versions for compatibility reasons. Both versions have the same elementary streams inside, they just have different containers.

Here is the link:

https://drive.google.com/file/d/0B68...ew?usp=sharing

Stacey Spears created a 2160p HEVC Rotating 8bit and 10bit Quantization Artifact test pattern. He gave me permission to post it here.

https://drive.google.com/file/d/0B68...ew?usp=sharing
Alright, I just joined the forum for this purpose, To educate myself about the subject, after searching and searching about bit depth and how to determine it I found your post which was helpful, but here is the problem
Can someone tell me that if the bit per channel number matters at all? Correct me if I am wrong but it's only used to determine your actual bit depth, you can have higher bit depth by adding another element (eg red green blue white, red green blue amber pixels) and you'd get a 32 bit display if you store 8 bit for each subpixel because you have 4 channels instead of 3, which means you'd get 1.07 billion shades still, WITH 8BPC! (Or maybe there's something wrong about this?)
And that is exactly what is confusing me
My laptop has RGBW pixels, it indicates it has 32 bit display, I played the gradient test and there is a huge difference between the upper and the lower segments, no banding or whatsoever at all on the upper segment
Does that mean my display actually has 32 bit color? Does that also means that it is equal to a regular RGB with 10bpc because it has 32 bit colors? Please help me I am really confused
Ghostface009 is offline  
post #70 of 73 Old 07-22-2017, 09:59 AM
Advanced Member
 
cswiger's Avatar
 
Join Date: Aug 2015
Location: Sunnyvale, CA
Posts: 624
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 304 Post(s)
Liked: 149
Quote:
Alright, I just joined the forum for this purpose, To educate myself about the subject, after searching and searching about bit depth and how to determine it I found your post which was helpful, but here is the problem
Can someone tell me that if the bit per channel number matters at all?
Yes. Bits per channel matters about as much as resolution, at least once you get over 1080p.

Quote:
Correct me if I am wrong but it's only used to determine your actual bit depth, you can have higher bit depth by adding another element (eg red green blue white, red green blue amber pixels) and you'd get a 32 bit display if you store 8 bit for each subpixel because you have 4 channels instead of 3, which means you'd get 1.07 billion shades still, WITH 8BPC! (Or maybe there's something wrong about this?)
Yes, you can add another element such as a white subpixel.
However, unless the color chosen is a unique primary color, you aren't adding additional shades.

A dark grey in 8-bit RGB might have a value of x090909. In a Pentile RGBW display, 0x09090900 would be the same grey. But it could also be expressed as 0x08080801, 0x07070702, 0x06060603, etc to 0x00000009. These would all be the same exact shade of grey; the main difference is that it is a bit more power efficient to use the W subpixel for greys and whites.

Normal human vision is trichromatic so three well chosen primaries:

https://en.wikipedia.org/wiki/Trichromacy



...are all that is available.

Quote:
And that is exactly what is confusing me. My laptop has RGBW pixels, it indicates it has 32 bit display, I played the gradient test and there is a huge difference between the upper and the lower segments, no banding or whatsoever at all on the upper segment. Does that mean my display actually has 32 bit color? Does that also means that it is equal to a regular RGB with 10bpc because it has 32 bit colors? Please help me I am really confused
If you can see a huge difference and lots of banding, that indicates your display is probably 6-bit+2-bit FRC.
If one can can see some difference and medium to mild banding, that indicates a true 8-bit display.

On a 8-bpc+2-bit FRC display, I can still see faint banding, most noticeable on the darker portion. Finally, on a true 10bpc display, I can tell that the upper and lower potions are not the same, but I don't see banding per se, more like the grey has a subtle but different "texture".

Computer systems routinely store 8bpc RGB values in a 32-bit format and call that a "32-bit display"; the remaining 8 bits might be unused, or they might hold an alpha channel, shadow mask, etc for textures. 10bpc RGB formats also fit into a 32-bit format, this time with 2 bits left over.

What format is used in the graphics card, or is sent over a cable, does not change the capabilities of the display. If the display can only do 8bpc, then you will get banding even if you send it a 10-bit signal, although better displays will take more advantage of a higher quality signal when dithering down to 8 bits.
galonzo and ConnecTEDDD like this.

Primary (2.1): Thiel CS 1.2 / Velodyne F1000 sub / Adcom GFA-555 Mk2 / Rotel 970BX / Sony CDP-CA9 ES
Secondary (5.1): Paradigm Mini-Monitor v.3 / CC270 v.3 center / ADP-170 v.2 surrounds /
SVS SB1000 sub / Marantz SR-6010 / AppleTV + Win7 PC / BenQ BL3200PT

Last edited by cswiger; 07-22-2017 at 10:08 AM.
cswiger is online now  
post #71 of 73 Old 07-22-2017, 03:53 PM
Newbie
 
Join Date: Jul 2017
Posts: 3
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 0
Quote:
Originally Posted by cswiger View Post
Yes. Bits per channel matters about as much as resolution, at least once you get over 1080p.

.
First of all thanks for your informative response, It helped me to understand quite a few things.
Secondly since I am a new user, I am not allowed to post links, so instead of using "." I will replace it with "(dot)" it may look stupid, but Im trying my best so I hope you just go along with it and appreciate it :-) Anyways
Unfortunately I am still lost and confused about the subject.

Main questions: 1. Is my panel true 32 bit afterall? if not then why?
2. how would consumers be able to test if their panels have the true bitdepth that the panel's manufacturer claims without spending ~$5000 on a spectrophotometer to analyze gradient test shade by shade since not even a gradient test (!!!) would be able to tell if the panel has that bitdepth or not due to frc? and how to test if frc is there or not? because after searching most tech-review websites..actually use gradient and artifact testing in a similar manner to the one here.

Anyways what I meant in the previous comment is that why bits per channel even matter when it's in the end just used to determine the actual bit depth of the display? if we assume a panel with 4 colors per pixel and 8bpc can achieve the same bit depth of one with 3 channels and 10bpc? (Actually 2 bit more deep since 8x4 =32 and 10x3=30) ? why shouldn't we look at the bit depth because it's the one that actually tells us how many colors and shades the panel can produce (2^24=~16.7 million)?
but then your explanation was that the additional channel has to be a primary color, but isn't white a primary color?
maybe you meant a primary color as in RGB (trichromacy) as you mentioned, but again there are pentiles with RGBG, Are they true 10 bit while rgbw isn't?

Also I didn't understand why these close shades of grey would be all infact a single shade of grey that gives the same output, the main counter argument about all this is the fact that W isn't a normal alpha channel and the resulting colorspace isn't the same as rgba (a for alpha), it still emits visible white light which can be combined with the remaining color using the 8bits stored of the w channel for control along with the 24 bits from the rest of the channels (rgb) to create more shades of the same color, as said here www(dot)mobilebeat(dot)com/understanding-rgba-vs-rgbw-in-two-simple-steps/
and here www(dot)reinhard-kietzmann(dot)de/rgbw_color_space(dot)html

Trichromacy and biovision seems like a weak excuse to me against the fact that you can combine more colors to result in more colors, it's like saying if we use an orange colorpencile over a yellow we won't be able to see the result.
what I meant is rgbw can be used to produce true 32 bit output, unlike what you suggested.

I'm sorry If I sounded mocking your observasion anywhere, I just want to educate myself about the subject and try to be as polite and mature as possible since it's my first appearance here, but it worths to mention that your explanation to the subject wasn't convincing enough and I didn't understand some of your explanation

and regarding the Pentile technology, why you assumed directly that a rgbw panel was automatically pentile? don't regular rgbw-rgbw exist at all? or you just assumed it is pentile because most pentiles use additional channel? don't 4 channel per pixel systems exist outside pentile or that's a pentile proprietary for Pentile? confusing.
Again sorry If I sounded mocking your informative observation but it wasn't convincing enough, add to the fact that there are true deep color rgbw large panels out there.

regarding the difference I didn't mean there isn't "banding" at all , there is a visible gradient but it isn't as noticeable as the bottom part, I think it's obvious that English isn't my native language, the difference was huge as I felt the upper part had more "depth" to it.

"your display is 6bit + 2 bit frc" sorry what does that even mean?
you mean that instead of saying it's actual 32 bit, it's 6bpc + frc for 24 bit content, and 6bpc + frc for deep color content, but again 6bpc for 4 channels would mean 24 bit, which is true color and 8 bpc would mean deep color, which would mean it's true 32 bit again (+ frc? sign me in for that panel, or no, because I consider dither to be nothing more but downsample)
excuse me, but does that even exist? (even if we assumed it was 3 channel rgb? I mean why in the hell a 8bpc RGB panel use 6 bpc + frc for true color content when it is actually 8bpc frc?)

More honest p.o.v: it all sounds illogical instead of saying a certain panel is truly 10bpc or truly deep color, I know about advanced frc and advanced dithering algorithms, I know how dithering and frc look like in comparison to native, and if my panel was using frc I wouldn't be able to notice the gradient of close, random sequential shades of random colors at close distances like I did, we are talking about a difference of x64, or 10^3 to the original, it would appear way smoother than native with less color depth and less details just like the example here if you display it on frc:
tiamat(dot)tsotech(dot)com/24-bit-color-sucks
KEEP IN MIND: I am not saying frc can't compensate to view a similar result to the target, but the target will always be better and frc/dithering will always fail to 100% fully represent it, if frc can compensate all these shades in your theory then we wouldn't need higher bit depth at all, we can just use frc to compensate for any given number of shades but it's a well known fact that no frc can't do that:
( en(dot)wikipedia(dot)org/wiki/Frame_rate_control , emphasis on the second paragraph)

you might say that I am underestimating frc afterall, maybe, because it's nothing like what people on the internet say or describe it as some sort of voodoo black magic made by the devil and used by evil manufacturers to trick customers (same goes to bayer filter, Pentile , etc) , I see people everywhere on tech-related websites chanting "Boo fake deep color, true color display in disguise with frc and dithering!" reaching the point of turning bit depth into a very controversial subject, more controversial than it ever was and leading to unnecessary misunderstandings and vortex(es) of twisted logic, and even the point of strong probabilities of falsely accusing true 10bpc/deep color panels as faked with frc unless it has a huge pricetag and fancy marketing terms attached to it, we don't know what's the truth anymore and which is which, between manufacturers who approach goals by cheating and the folks at the ultra end market who just sounds like the folks at headfi who use words like "PRAT" as a justification to them spending lots of money on average products.

Again I'd like to thank you for what you posted and for what you'll post in the future, it was helpful but as you can see there has to be a misunderstanding by either of us or a wrong point in your informative wonderful reply, in short terms I appreciate what you've done and for what you'll do here
Ghostface009 is offline  
post #72 of 73 Old 07-22-2017, 09:34 PM
Advanced Member
 
cswiger's Avatar
 
Join Date: Aug 2015
Location: Sunnyvale, CA
Posts: 624
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 304 Post(s)
Liked: 149
Quote:
Originally Posted by Ghostface009 View Post
Main questions: 1. Is my panel true 32 bit afterall? if not then why?
You haven't told us what panel it is you're asking about...?

Quote:
2. how would consumers be able to test if their panels have the true bitdepth that the panel's manufacturer claims without spending ~$5000 on a spectrophotometer to analyze gradient test shade by shade since not even a gradient test (!!!) would be able to tell if the panel has that bitdepth or not due to frc? and how to test if frc is there or not? because after searching most tech-review websites..actually use gradient and artifact testing in a similar manner to the one here.
Most people test their displays with the Mark 1 eyeball.

However, you can use a recent cellphone or DLSR camera if it can do video capture at, say 1080p @ 60 Hz to see temporal dithering on a frame-by-frame basis. Beyond that, you could also run an EDID utility against your display, which will give results like:

Code:
Monitor
  Model name............... BenQ BL3200
  Manufacturer............. BenQ
  Plug and Play ID......... BNQ8017
  Serial number............ 68Exxxxx
  Manufacture date......... 2014, ISO week 33
  Filter driver............ None
  -------------------------
  EDID revision............ 1.4
  Input signal type........ Digital (DisplayPort)
  Color bit depth.......... 10 bits per primary color
  Color encoding formats... RGB 4:4:4, YCrCb 4:4:4, YCrCb 4:2:2
  Screen size.............. 710 x 400 mm (32.1 in)
[ ... ]
Color characteristics
  Default color space...... sRGB
  Display gamma............ 2.20
  Red chromaticity......... Rx 0.653 - Ry 0.332
  Green chromaticity....... Gx 0.305 - Gy 0.626
  Blue chromaticity........ Bx 0.146 - By 0.047
  White point (default).... Wx 0.313 - Wy 0.328
  Additional descriptors... None
Quote:
Anyways what I meant in the previous comment is that why bits per channel even matter when it's in the end just used to determine the actual bit depth of the display? if we assume a panel with 4 colors per pixel and 8bpc can achieve the same bit depth of one with 3 channels and 10bpc? (Actually 2 bit more deep since 8x4 =32 and 10x3=30) ? why shouldn't we look at the bit depth because it's the one that actually tells us how many colors and shades the panel can produce (2^24=~16.7 million)?
The number of unique shades is what actually matters to the eye.

There isn't a fourth primary color channel lurking on your DVD and BluRay disks. All pre-recorded movies, streaming video, etc are recorded in 3-channel RGB formats (albeit sometimes lossy compressed into YCrCb 4:2:x), and that remains true if you look at the TDMS wire format that is sent across the DisplayPort or HDMI cable.

Again, for a specific panel, you can run an EDID utility and look at the VSDB block:

Code:
CE vendor specific data (VSDB)
  IEEE registration number. 0x000C03
  CEC physical address..... 1.0.0.0
  Supports AI (ACP, ISRC).. No
  Supports 48bpp........... No
  Supports 36bpp........... Yes
  Supports 30bpp........... Yes
  Supports YCbCr 4:4:4..... Yes
  Supports dual-link DVI... No
  Maximum TMDS clock....... 270MHz
If your panel doesn't say Yes to supporting at least the 30bpp format, then it is limited to 8-bit per channel only. If it supports 30bpp only, it can at least handle 10-bit per channel signals, but the panel itself might dither that to 8-bits using FRC or it might be native. If it says Yes to 36bpp, that means it can handle 12-bit per channel signals, etc.

Quote:
but then your explanation was that the additional channel has to be a primary color, but isn't white a primary color?
Nope.

Beyond that, I don't claim to be an expert on RGBG or RGBW displays. I think they're pretty much all under the Pentile branding, but I have no idea whether anyone else has cloned the notion or not.

While I am happy to point you to a tool like:

http://www.entechtaiwan.com/util/moninfo.shtm

...and I am also happy to point you towards information so you can do your own research, your remark here:

Quote:
your explanation to the subject wasn't convincing enough
...indicates a misunderstanding of our interaction: I'm not here to convince you of anything.
galonzo and Ghostface009 like this.

Primary (2.1): Thiel CS 1.2 / Velodyne F1000 sub / Adcom GFA-555 Mk2 / Rotel 970BX / Sony CDP-CA9 ES
Secondary (5.1): Paradigm Mini-Monitor v.3 / CC270 v.3 center / ADP-170 v.2 surrounds /
SVS SB1000 sub / Marantz SR-6010 / AppleTV + Win7 PC / BenQ BL3200PT
cswiger is online now  
post #73 of 73 Old 08-04-2017, 03:14 PM
Newbie
 
Join Date: Jul 2017
Posts: 3
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 0
Quote:
Originally Posted by cswiger View Post
You haven't told us what panel it is you're asking about...?

Most people test their displays with the Mark 1 eyeball.

However, you can use a recent cellphone or DLSR camera if it can do video capture at, say 1080p @ 60 Hz to see temporal dithering on a frame-by-frame basis.
Hi again,
Although I wish you gave this helpful reply from the 1st time, I'm still thankful for the useful info, unfortunately I don't have the laptop with me atm.
I did mention it was a laptop, the panel's name or the laptop's name isn't important atm, I just wanted a method to test and I found it.
By stating "It wasn't convincing enough to me" I didn't mean I need someone to convince me, I just needed a factual/objective point of view, like this reply you gave when i first got here.
However, I find taking shots with fast shutter or 60hz video and frame by frame analysis and checking for temporal dithering to be the best way to reach our objective, which is what I'll do.
Ghostface009 is offline  
Sponsored Links
Advertisement
 
Reply Display Calibration

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off