Jump to content

Talk:DisplayPort

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

DRM as anti-feature

[edit]

I suggest to mark HDCP and DPCP as anti-feature Uis9936 (talk) 12:50, 3 April 2020 (UTC)[reply]

I suggest you give reasons for your suggestions, otherwise they cannot be evaluated and will most likely be ignored. GlenwingKyros (talk) 19:09, 3 April 2020 (UTC)[reply]
A feature that more people can use for something than not and that can be disabled with a checkbox in video drivers (because the DP standard itself only defines the method of transmission) isn't an anti-feature; it has to be used by software or it's unimportant. By this logic the viral nature of the GPL3 is far more an anti-feature since there's no getting rid of it and it can easily ruin someone's livelihood should they accidentally produce a binary that was statically linked instead of dynamic and make it available for download, but bring that up and suddenly there's a big ol' tent pitched for protecting IP from being appropriated and re-sold when it comes to a piece of software you checked in a one line comment fix and a bugged function that was reverted a year later to 10 years ago. :P A Shortfall Of Gravitas (talk) 04:17, 28 March 2023 (UTC)[reply]

Dear mother of "Bob", the endless tables...

[edit]

Honestly, do we really need this many of them? Monitor manuals don't contain this much information about supported display modes. They tell you what DisplayPort or HDMI version you need to be using to run at max resolution and color depth, which is where anybody who needs it will find this information.

To make matters worse, why the heck is there an HDR table that's really just a 10bpc table? Windows is happy to run HDR games and play back HDR10 movies at 8 (dithered), 10, or 12 bits (technically I think both of those are still dithered from internal 16-bit float representations too, but it isn't visible), so it really has nothing to do with supported resolutions. Normally you wouldn't have a displayport compatible display that was limited to 8-bit color, but I suppose a DP->HDMI adapter and a target display that only supports HDMI 2.0 combined with not wanting the desktop to look like garbage due to chroma subsampling might be a reason. HDR10 is foggy about how the content is actually played back. Lots of PC based players can tone-map it to SDR and avoid the need for a display with explicit support entirely.

Then there's the whole deal with at least 3 of the tables duplicating or providing needlessly similar information; we've got:
Refresh frequency limites for common resolutions: plenty of refresh rates nothing in existence supports, but whatever.
Refresh frequency limits for standard video: either we're talking about video or we're talking about monitor resolutions; if it's video there are too many listed and if it's resolutions it's missing about 20 more, plus there's a table of limits directly above that makes this redundant information... The non-"standard" video resolutions here are wide HD mode, 2560x1440, 4k (practically never encoded at 8-bit except by non-video oriented cameras and many phones), 5k (not standard at all, a couple of apple displays used it?), and 8k (not used very often in practice).
Refresh frequency limits for HDR video: explicitly starts talking about HDR10, which is a broadcast / video standard, and if we assume that's what it's meant to cover the entire table is incorrect because HDR10 video is also 4:2:0 chroma sub-sampled, or else it's the Dolby Vision variation and is 12-bit 4:2:2 subsampled... both of those have the same frequency limits in practice.

To top all that nonsense off we get arbitrary lists of supported resolutions and refresh rates in no less than 3 other different sections, as though someone had a severe OCD fit and had to calculate at least 5 variations for every resolution and wash their hands 20 times or they couldn't leave their house.

I wouldn't even mention this but I just tried to look up something about DP 2.1 really quick on mobile and couldn't even find a summary / history of versions section like there is towards the start of most other computer interconnect type pages on wikipedia. Then the tables mess wrecked the ability to scroll correctly without a stylus, I noticed the total lack of useful information being conveyed, attempted to will myself to die, failed, and decided to come point this stuff out.

I'm not fixing it because I'm actually kind of terrified of whatever mind decided to calculate all of this stuff and what kind of spree removing any of it might send them on. A Shortfall Of Gravitas (talk) 04:55, 28 March 2023 (UTC)[reply]

Hi, welcome to the DisplayPort talk page :)
Honestly, do we really need this many of them? Monitor manuals don't contain this much information about supported display modes. They tell you what DisplayPort or HDMI version you need to be using to run at max resolution and color depth, which is where anybody who needs it will find this information.
Yet one of the most commonly asked questions regarding DisplayPort is "what DisplayPort version do I need to run XYZ format" and "does DisplayPort version X support Y format?". Many people look for this information.
To make matters worse, why the heck is there an HDR table that's really just a 10bpc table? Windows is happy to run HDR games and play back HDR10 movies at 8 (dithered), 10, or 12 bits (technically I think both of those are still dithered from internal 16-bit float representations too, but it isn't visible), so it really has nothing to do with supported resolutions.
The reason there is a table for 10 bpc color depth is because people ask if DisplayPort will support X resolution at Y refresh rate at 10 bpc color depth. The most common reason people want to know this information is because it is used for HDR video, so the section heading uses this term to help people find the information they are looking for. Though yes, this table is also applicable to standard 10 bpc transmission without HDR as well. "Refresh limits for HDR video" is a somewhat dubious title, since "HDR" could refer to color depths other than 10 bpc, but in the general consumer market (at least, at the time when these tables were created) HDR essentially implied 10 bpc. Windows supporting 8 bpc transmission with source-side dithering to support HDR is a somewhat recent development. Certainly, the titles/organization could use a reevaluation.
Normally you wouldn't have a displayport compatible display that was limited to 8-bit color
There are many models with DisplayPort that only support 8 bpc color depth. And even if the display supports 10 bpc color, it is still useful to know, for example, that 120 Hz at 4K is only achievable by dropping the color depth down to 8 bpc if you're using an HBR3 link without DSC. Many people with GeForce 10-series cards encounter this limitation when attempting to use 4K 144 Hz monitors.
Then there's the whole deal with at least 3 of the tables duplicating or providing needlessly similar information
There used to be only the lower sections (one for 8 bpc and one for 10 bpc), then HDR/10 bpc versions where added. The table on top was added relatively recently, which contains mostly same information but in a more compact way. The redundancy is known, and this table should replace all of the lower ones, however the compact table currently doesn't have any information about the limits when chroma subsampling or DSC are used, so the other tables cannot be removed yet.
Refresh frequency limites for common resolutions: plenty of refresh rates nothing in existence supports, but whatever.
The original tables only listed "standard" refresh rates in order to stick to "nice" numbers like 60 Hz and 120 Hz when listing the maximum limits, instead of random arbitrary numbers representing the true exact maximum. But if we just list that number as the maximum, it misinforms people, as happened when monitor manufacturers started making 165 Hz monitors, and people would point at the table and say "No, DP 1.2 doesn't support 1440p 165 Hz, look, this table says max is 144 Hz". And when it was changed to say 165 Hz, monitor manufacturers started making 170 Hz monitors and the same situation repeats. So, to avoid this, the tables use a Yes/No format with each refresh rate listed individually, so it doesn't say "Max 165 Hz", it just says Yes 165 Hz is supported, and 170 Hz or 175 Hz or whatever other arbitrary new refresh rate they are using is simply left ambiguous because it isn't present on the table. But the consequence of this is that it makes the table very long, which you complain about. And indeed, you are not alone in this complaint, actually someone else decided it's really not important to stick to the nice round numbers, and let's just list the exact maximums, then the table can just list one number for each resolution and be much more compact, and that resulted in the first table you see on top. But then you also complain about that the table having random arbitrary numbers that "nothing in existence supports" (though actually, any DisplayPort source device with the proper output speed does support these, that is the whole meaning of the table). But it seems you will have a complaint regardless of the organization :)
either we're talking about video or we're talking about monitor resolutions; if it's video there are too many listed and if it's resolutions it's missing about 20 more
We're talking about the display resolutions that are most commonly used or discussed in relation to DisplayPort, which the majority of people will be looking for information about. And that's 1080p, 1440p, 4K, 5K, and 8K (5K and 8K mainly exist in academic discussion only, but since DisplayPort press releases in recent versions commonly reference these (example), it's relevant to the topic and it's information that people are sometimes curious about). And also 3440 × 1440 I guess, as the person who made the Common Resolutions table felt it's worthy of inclusion. I'm ambivalent about that one.
Refresh frequency limits for HDR video: explicitly starts talking about HDR10
I don't see where it starts talking about HDR10 in this section. I see where it name-drops HDR10 as an example of an HDR standard that specified 10 bpc color depth, prefaced with the phrase "such-as" but you seem to suggest that HDR10 is the main topic of this section, and I don't see that. I don't think anyone reading this is coming away with the impression that this section is specifically about the HDR10 standard alone.
and if we assume that's what it's meant to cover the entire table is incorrect because HDR10 video is also 4:2:0 chroma sub-sampled
This article is about DisplayPort, which is an interface for transmitting video data between a computer and a display. It is not about transmitting video files or reading or writing from disc or other storage media. HDR10 content and video files may be chroma subsampled, but that is not relevant here because displays that are running with HDR enabled and being used for viewing HDR content are not necessarily chroma subsampled, and in the case of computer monitors they rarely are; most computer monitors maintain RGB transmission when HDR is enabled, chroma subsampling is only used when the bandwidth is insufficient to transmit RGB with 10 bpc color depth is enabled (which, again, is why it is beneficial to know whether a format can be supported with 10 bpc color depth or not, so that you can know if chroma subsampling will be required; although this is less relevant now that Windows supports source-side dithering so can enable HDR with 8 bpc transmission).
To top all that nonsense off we get arbitrary lists of supported resolutions and refresh rates in no less than 3 other different sections, as though someone had a severe OCD fit and had to calculate at least 5 variations for every resolution and wash their hands 20 times or they couldn't leave their house.
This article is a collaboration from many different authors, each of whom have their own ideas, which is why such fragmentation exists. There is certainly some cleanup that can be done, if someone has time. You're welcome to share your ideas for improvement here.
I wouldn't even mention this but I just tried to look up something about DP 2.1 really quick on mobile and couldn't even find a summary / history of versions section like there is towards the start of most other computer interconnect type pages on wikipedia.
There is such a summary, towards the start, exactly like most other computer interconnect pages. It's the very first section of the article. [[1]]. It doesn't say too much, but there's not much to say about it.
Then the tables mess wrecked the ability to scroll correctly without a stylus
I'm really not sure what difficulties you encountered, I tried scrolling through on my phone just now in both mobile view and desktop view and did not find any issues with scrolling through these sections.
Regards  — Glenwing (talk) 08:10, 28 March 2023 (UTC)[reply]

Inaccurate feature comparison with HDMI

[edit]

Hey there, I just stumbled on this in the article. The article states:

HDMI can accept longer maximum cable length than DisplayPort (30 meters vs 15 meters).

While this seems fine at first glance, the comparison is inaccurate or "not fair". If we assume that with "maximum cable length" we are talking about the specifications of each, we have the following:

Now the question is should this point be removed entirely or clarified to include this disparity? Mihawk90 (talk) 11:19, 2 July 2023 (UTC)[reply]

Yes, this bullet point should be removed. As you pointed out, neither standard specifies a "maximum length" nor would it make any sense to do so, as there is no such fixed value.  — Glenwing (talk) 16:15, 2 July 2023 (UTC)[reply]
Done, thank you for the reply. Mihawk90 (talk) 10:25, 3 July 2023 (UTC)[reply]

Calculate what Resolution is possible with DSC

[edit]

How can we calculate, what max. Resolution on (for example) 10bit, 4:4:4 on 8k is possible with DP 2.1 and DSC ? So what maximum compression Ratio can DSC deliver actually?

How far down is it "visually lossless" ? I mean, the more compression, the more visual loss. So would 8K 240Hz be possible with DP2.1 and DSC (10 Bit 4:4:4)? 2A02:1210:8CF9:D100:D9A5:FB86:AC84:9418 (talk) 15:25, 21 February 2024 (UTC)[reply]

DSC compresses to some fixed bit rate, so it cannot compress to arbitrarily high ratios. The lowest possible bit rate is 8 bit/px. So with 8 bpc color depth (24 bit/px) this would be 3:1 compression, but if compressing from 10 bpc color (30 bit/px) to 8 bit/px it would be 3.75:1. For calculation simply multiply the data rate of the interface by the compression ratio, then calculate as normal.
 — Glenwing (talk) 08:51, 22 February 2024 (UTC)[reply]
and DSC can only use one specific bitrate/pixel or what rates are possible? Because if there is always only one bitrate, then there would also be always the same ratio. So i guess there are different possible bitrate/pixel. So which are they (all)?
And what is the maximum ratio then? Ok, you can't have arbitrarily high ratios, but what ratios can you have? Is 3.75:1 possible? And is it still "visually lossless" on the highest possible ratio?
When you look at the table, then for 10bit native it is 74FPS possible and therefore with a 3.75 ratio the maximum framerate would be 277.5FPS. So 240FPS on 8k 4:4:4 10bit should be possible with DSC? 2A02:1210:940B:1C00:6C8A:30FF:ACF5:B9DF (talk) 13:56, 7 September 2024 (UTC)[reply]
The DSC algorithm allows compression down to 6 bits per pixel up to 63.9375 in steps of 1/16 (I mentioned 8 bit/px as the minimum in my previous comment, as this is the lowest VESA recommends to maintain visually lossless quality, but the standard allows down to 6). So if you have 10 bpc (30 bit/px) input and compress down to 8 bit/px, that is 3.75:1. If you choose 7.9375 bit/px instead, it would be a 3.7795:1 ratio.
Theoretically the maximum ratio would be achieved with 16 bpc video (48 bit/px) with compression to 6 bit/px, which is 8:1. But VESA claims it maintains the "visually lossless" quality down to 8 bit/px only and they advertise the "3:1 ratio" heavily, so that's what we use in the table even though the algorithm can go further from a technical perspective. Higher complexity derivatives (VDC-M) can get visually lossless quality down to 6 bit/px, but this hasn't seen wide adoption/marketing yet.
So then for 10 bpc (30 bit/px) input, does it still maintain visually lossless quality down to the same bit rate (8 bit/px, which would not be 3.75:1 ratio), or does it only follow down to the same ratio (3:1, which would be 10 bit/px in this case)? As far as I know this is an open question, I have only seen studies that evaluated it with 8 bpc (24 bit/px) input. VESA never talks about DSC being able to achieve visually lossless encoding at 3.75:1 ratio. My expectation is it is probably still visually lossless down to 8 bit/px simply because the difference between 8 bpc and 10 bpc color depth is so subtle, but right now the article conservatively assumes only 3:1 is possible while maintaining visually lossless quality. Maybe there are newer studies; I haven't checked in a long time. I see an Anandtech article about VDC-M that claims DSC maintains visually lossless quality at 10 bpc -> 8 bit/px (3.75 ratio). I'd like to see the actual study though. https://www.anandtech.com/show/12759/vesa-and-mipi-announce-vdcm11-display-compression-standard-for-mobile
 — Glenwing (talk) 16:09, 7 September 2024 (UTC)[reply]
After further checking looks like VESA does claim 10 bpc -> 8 bit/px (3.75:1 ratio) is also visually lossless. I'll go ahead and update the table to account for this.
 — Glenwing (talk) 18:23, 7 September 2024 (UTC)[reply]

Questionable details about the physical layer

[edit]

In the table on the right there under “Electrical” are some voltages and currents specified with unclear meaning. DisplayPort doesn’t use 3.3 V for its signals. It uses differential signaling where the exact voltages aren’t specified, but the voltages (both the common mode bias and the differential voltage) should always stay way below 3.3 V. The DisplayPort connector has one supply pin though to power e. g. a DP to HDMI adapter. This pin supplies up to 0.5 A at 3.3 V. So this entry in the table could be renamed e. g. to supply voltage to make it correct. Additionally it is unclear to me, what the “Max. voltage” and “Max. current” entries mean. The latter could mean the supply current of the aforementioned pin, but then this should probably be stated explicitly. I don’t know of any voltage value that should reach 16 V, so this entry makes no sense to me. It should either be clarified or removed if there is no such voltage.

Another problem in this article is that at multiple places it talks about DP to HDMI adapters having to convert the signals from 3.3 V to 5 V. This is wrong as neither DisplayPort uses logic levels of 3.3 V (see above) nor does HDMI use 5 V logic levels. They both use differential signaling where the exact differential voltages aren’t specified, but are both way below 1 V. --Lukas.fink1 (talk) 21:26, 1 March 2024 (UTC)[reply]

Hi, thanks for the observations. Will review later today. I think DP specifies abs max 2.0V for Vcm and around 1.3V for Vpp, maybe this is where the 3.3V came from? (EDIT: Given the voltage listing as 3.3 V and max current of 0.5 A, this appears likely to be referring to DP_PWR) For the others, 16V etc. I have no guesses. This will need some cleanup. Thanks again for your notes.  — Glenwing (talk) 22:20, 1 March 2024 (UTC)[reply]
I've removed the electrical specifications from the main table since the categories themselves were quite vague and not of much value. I've also rewritten the sentence about dual-mode adapters.  — Glenwing (talk) 05:41, 2 March 2024 (UTC)[reply]
I notice also, many datasheets for DP dual mode level shifting ICs mention 3.3V (from DP_PWR) to 5V (for HDMI DDC) conversion, it seems likely this is the origin of that sentence, and something got lost in translation along the way...  — Glenwing (talk) 06:07, 2 March 2024 (UTC)[reply]
  • Just a comment here that this is another example of the article getting increasingly technical going beyond the scope of an encyclopedia. If the only source out there talking about this "limitation" is a whitepaper resource, such as the VESA Interoperability Guideline, then it probably doesn't have WP:DUE weight for inclusion on Wikipedia. The significance (and translation of what it means) really needs to come from a secondary source. This article is due for an overhaul, which I plan to address in a new thread. --GoneIn60 (talk) 03:28, 4 March 2024 (UTC)[reply]

Technical nature of the article

[edit]

Over the years, it has been mentioned on numerous occasions, such as here and here, that large portions of this article read like a technical whitepaper or manual. There is a gratuitous use of unexplained jargon and terminology used throughout that only makes sense to an expert or specialist. The first section is "Versions", which goes right into describing features like color space, link layers, Adaptive Sync, etc., which are never explained. Readers with zero knowledge of what these things mean will see this as technical jargon and find it useless. There's plenty more of that scattered throughout the article.

Per WP:TECHNICAL: "An article may disappoint because it is written well above the reading ability of the reader, because it wrongly assumes the reader is familiar with the subject or field..."

As noted in WP:UPFRONT, we need to push the highly technical portions of the article to the bottom and lead with a high-level overview that's easy to understand. A while back, we had an Overview section that followed the lead, which attempted to describe the basics of DisplayPort technology that appealed to a general audience. At some point, this was abolished and partially merged into the lead unnecessarily (which actually violates WP:LEADFOLLOWSBODY). We need to consider reinstating that section to help ease readers into the article. Eventually, I'd like to weed out as much unnecessary technical jargon as possible, but for now, pushing that to the bottom will suffice. If anyone has any thoughts, please weigh in here. I'll give it some time before making any cleanup attempts, thanks. --GoneIn60 (talk) 03:29, 4 March 2024 (UTC)[reply]

Forgot to add that during the cleanup phase, interpretations that are only cited to a whitepaper (meaning it provides unsourced analysis of technical specs) will be removed or flagged with a {{citation needed}} tag. Even if correct, these interpretations must be cited to reliable sources that are making that analysis for us. This does two things. First, it removes any doubt that the analysis is not WP:OR. Second, it shows significance that the detail in question qualifies as WP:DUE. --GoneIn60 (talk) 03:37, 4 March 2024 (UTC)[reply]


Not approachable

[edit]

Glenwing, I see that you reverted my edit. The YCbCr article describes YCbCr, Y′CbCr, or Y Pb/Cb Pr/Cr, also written as YCBCR or Y′CBCR as a family of color spaces. Is that description incorrect? Adding to the confusion is that the standards in the "colorspaces" section aren't strictly colorspaces. ITU-R BT.709, for example, is a very broad specification for encoding and signal characteristics that happens to include a color-space definition, for example.

The reference describes the "colorspaces" in this table as "colorimetry specifications". Is that more accurate?

The table isn't too approachable, and I think that also should be fixed. Linking to other wiki articles that explain the concepts being enumerated in the table would be helpful. -- Mikeblas (talk) 02:57, 31 May 2024 (UTC)[reply]

Primarily it didn't make sense to have two table sections both called "Color space support". Otherwise I might have just started a discussion here instead. "Color space" is a somewhat nebulous term which is used by different people to mean different things. My view is that YCbCr (YPbPr) is not a color space, it is a color model, like RGB. The page for Color space says:
Since "color space" identifies a particular combination of the color model and the mapping function, the word is often used informally to identify a color model. However, even though identifying a color space automatically identifies the associated color model, this usage is incorrect in a strict sense. For example, although several specific color spaces are based on the RGB color model, there is no such thing as the singular RGB color space.
I tend to agree with this, and the same with respect to YCbCr. I think that referring to YCbCr as a "color space" is incorrect, just as saying "RGB" is a color space is also incorrect. But other people may disagree, like whoever wrote the YCbCr article. But wikipedia reflects whatever is in its sources anyway; garbage in, garbage out. If misuse of a term is widespread outside WP, then it will appear here too. But at least in the context of monitors and video interfaces, "color space" generally refers to actual color spaces like sRGB. Whether you're transmitting in RGB or YCbCr is your pixel format (more correct I think) or color format.
ITU-R BT.709 doesn't give a specific name to just the color space that it defines, so there's no other way to refer to it. But I think it's fairly self explanatory that "ITU-R BT.709 color space" means "the color space defined in the ITU-R BT.709 standard", I don't think there's much risk of someone coming away with another idea of what was meant (I'm not really sure what other way it could be interpreted, honestly). It's in the same way that when someone says "Ultra HD resolution", even if someone was actually familiar with the Ultra HD specification (it defines other parameters like frame rate and video interface/HDMI requirements, not just resolution) it would be apparent that the person meant "the resolution used by the Ultra HD standard" which is 3840×2160. The standard defines other things but I don't think it will lead to any confusion about what is meant by "Ultra HD resolution". Same for "ITU-R BT.709 color space".
Not opposed to some wikilinking in the table, it can be explored. I should also point out we should be thoughtful about what links where; for example linking the words "color space" to lead specifically to YCbCr (rather than Color space) is a somewhat strange (perhaps WP:SUBMARINE) link.  — Glenwing (talk) 03:17, 31 May 2024 (UTC)[reply]
Not sure how to respond, since you think there's "no other way to refer to it" and "don't think there's much risk". I think you're more familiar with the material than most readers, and that familiarity prevents you from seeing the problems here. -- Mikeblas (talk) 15:30, 31 May 2024 (UTC)[reply]
You could respond with an alternative way of referring to the color space defined in the ITU-R BT.709 standard other than "the ITU-R BT.709 color space", which would show that my assertion was wrong. Or an explanation of how referring to "the BT.709 color space" could lead to confusion on the basis that BT.709 also contains additional things besides a color space. Above I was expressing my views as they currently stand. I'm open to new considerations, but it would require a convincing argument to be provided.  — Glenwing (talk) 15:41, 31 May 2024 (UTC)[reply]