HDMI Latency Myth Busted
Вставка
- Опубліковано 27 сер 2024
- Does HDMI add significant delays to your video signal? We find out...
Sony PVM-A170 OLED Monitor - djp.li/pvma170
Blackmagic Design SDI-to-HDMI 6G - djp.li/bmds2h6
Blackmagic Design HDMI-to-SDI 6G - djp.li/bmdh2s6
Blackmagic Design ATEM Mini - djp.li/atemmini
CrewAxis - djp.li/hdmidelay
Video editing by @wittyfilmgirl
--------------
To help support this channel, please...
Sign up for our video production management website:
www.crewaxis.c...
www.amazon.com...
Back us on Patreon:
/ djprod
----------
Follow us online!
DJP Website - www.djprod.biz/
Second UA-cam Channel - djp.li/youtube2
Email Mailing List - djp.li/email
Social Media Accounts - djp.li/social
Partial Equipment List - djp.li/gear
Free Production Tools - djp.li/tools
#hdmi #videoproduction #latency #blackmagic #videodelay #blackmagicdesignatemmini #sdi #sony #crewaxis #switcher #atemmini #liveproduction #livestream #djp #hdmidelay #oledmonitor
ATEM Mini input 1 is “direct switched” to HDMI out for low latency gaming use cases. Try measuring ATEM delay with inputs 2-4.
Also note that using higher frame rates like 60 FPS vs 24 FPS will give you lower absolute latency when measuring in ms instead of frames because a single frame buffer at 60 FPS is 17 ms (1/60s) delay but the same frame buffer at 24 FPS is 42 ms (1/24s) of delay.
Looks like you confirmed the delay! One frame of delay in the conversion, but that's a best case scenario since you're using a SDI-HDMI converter. The ATEM Mini not adding additional delay is encouraging though.
It's actually a lot less than a frame. I was able to confirm with a Blackmagic Design engineer that the conversion delay is only a couple milliseconds. If I had two identical low latency monitors, we'd be able to see that the actual delay is far less than a frame (but I don't). It shows up as a frame here because the monitor can't delay the second signal by anything less. Even with the processing delay of the ATEM Mini, it's still less than a frame.
@@djp_video ahh that's great news!
I'd be interested to see the same test, but with HDMI out from the camera converted to SDI, and both SDI and HDMI directly out of the same camera for reference. I'm also aware that Black Magic converters reclock the signal.
Great test. It would be amazing to have the manufacturers start generating some sort of video output delay test (kind of like what they do for computer monitors in mil-seconds). That way we would know roughly what the delay is for each HDMI camera, and then be able to resync audio if necessary. I know there will probably be some tolerance because each camera will still be slightly different, but 99% of people will only notice if its more than 2-3 frames out of sync in my experience.
I use 2 FS5's which have 3G-SDI outs. I was a bit bummed to figure out I could not get a 4k signal out of the SDI (learned about 12G/6G/3G etc from Doug). I ended up getting Atomos Shogun Infernos, which can take a 4k HDMI signal and convert it into 6G SDI, so now I have the best of both worlds. the HDMI cable is 1 foot to the 7 inch monitor (which is way better than the built in one), plus I have a super 35mm sensor with a servo zoom lens (28-135mm F4.0).
I'm working on a comparison spreadsheet for live video. It's currently got over 3,000 data points, 100+ cameras with 30+ features listed for easy comparison between models, frame rates, outputs etc. Will update you when it's done Doug! I'm using airtable, so it's super easy to filter, group, and sort. I want to possibly create a resource the live video community can open source/help update so we can compare gear better and faster, so we spend less time researching and more time shooting!
That's pretty amazing that you're taking the time to build a spreadsheet. That's some pretty useful information.
Try to keep a balance of both because when it comes to the real technical stuff on equipment your channel has been the most accurate and beneficial to me. If your beginner stuff is as good, there will be no need to watch anybody else. You have solved many technical problems for me. For example many channels tell, the technical max length for HDMI cables so I followed those guidelines. The cables never work properly I kept running into reliability problems: sound with no picture, picture no sound, both pic and sound but then signal drops, etc. I purchased several sets trying to get better cables, with each subsequent set costing more. It wasn't until I watched one of your videos, that I found the actual reliable length for HDMI and I switched to SDI for what I needed. I still do use HDMI but never more than 15 feet because with dongles for converting size and camera monitors and converters I do want to stay under the limit. Icing on the cake you told me the right kind of SDI cables to get. Glad to get more from you. Keep up the good work. I am definitely a fan and I have learned a lot from you, thanks.
If I want my channel to grow I'm going to have to cover more of the basics. The audience for the more technical stuff just isn't there. And in order to cover the expenses of creating the videos, I've got to get more views. Catch-22.
Other attempts to cover the costs just haven't worked.
in your test, you get one frame, because the converters add a line or so each.
that leads to a de-synchronization between input 1 and 2, the monitor thus adds 1 frame latency to be able to display both images side by side on the same screen
the real issue is not the hdmi itself, but the processing in various consumer level equipment.
DSLRs have between 4 and 6 frames latency, consumer tvs have the same (which includes useless "AI image improvement" and the like)
I wish I did have two A170 monitors so I could display both signals simultaneously without it having to delay the second signal, but as you're probably aware, these are very expensive monitors, and one is all I've got.
Yes, the issue is the consumer or gear meant for photography rather than video production. But somehow in a lot of online forums there are a lot of people blaming HDMI itself for the delay.
Doug Johnson Productions HDMI has plenty of other reasons to be labelled as “crap”, among which:
• HDCP requiring a couple seconds to synchronize
• HDCP being a pain in the rear end when you have legal uses and requiring the use of gray area devices to get rid of
• The whole “transmit the data on several pairs” thing that leads to utter uselessness of the tech for lengths of more than 10m or so
I'm not disagreeing -- I avoid HDMI whenever even remotely possible. But blaming HDMI for the latency when using consumer cameras isn't the right way to address that particular problem, and we're seeing a lot of that happening in online discussion forums and videos from other channels.
PhotoJoseph (UA-cam moniker) is my source of accusation of HDMI latency. His problem stems from audio delay that prevented him from monitoring audio while speaking as on camera talent. He solved this by switching to SDI on a Blackmagic Micro Studio Camera 4K, which solved the problem so that his voice was heard on ear bud monitors in real time.
I've seen the video (sigh). It's a camera problem, not an HDMI problem.
I’ve got a feeling Panasonic DSLR cameras are bad for latency, but photojoseph is a Panasonic ambassador so probably avoids blaming them. I’ve noticed it from a Panasonic GX80, but HDMI out of my Sony camcorder is fine. photojoseph channel is still great though.
We use a BMPCC4k for in-studio live streaming and the HDMI output delay into an ATEM TV Studio HDMI input is small enough that we can talk over our voice monitor live without disruption. Source mic is running audio through a mixer into the BMPCC4k and then bringing the combined signal into the ATEM. Once you throw in a cross converter to SDI and the delay can be distracting, but straight out of camera over HDMI, it's maybe a one frame delay at the worst. So the presenter gets a very slight reverb effect when listening to their own voice, but it's not going to derail the conversation.
Hi Doug. Great, real world test! May well have been said in previous comments, but I’m pretty sure a lot of this myth comes from disparities in camera output latency on mixed camera systems, particularly with a lot of folk using units like the GH5 for B-cam, which is well known for output latency. Always love your videos, BTW, so nice to be spoken to rather than shouted at! :)
What camera was used for this test?
Sony PXW-Z150
8:47 Yay! We're on the same boat! :D lol. I wish I wasn't on that boat, but whatcha gonna do? ;D
Thank you for this test and explanation.
I just performed some TDR (Time Demain reflectometry) tests on my “Amazon Basics” 10M HDMI with Ethernet cable - 32nS end to end. Hardly the end of the world
I have 5 the same camera's that run at 50frps. Three come to the switcher by hdmi (1 to 10 meters max) and two come to the switcher by converting to sdi at the camera and back to hdmi just before the switcher. The delay that happens is not noticeable different for the camera's that come in over hdmi as over sdi. There is 4 frames delay in the switcher compared to the direct sound that is coming to the switcher from my microphones through an audio mixer. My guess is that the camera causes the 4 frames delay and not the hdmi and/or the conversation to and from sdi. I always 'blamed' the camera for the delays but good to know it has nothing to do with hdemi :) I was rather surprised that the hdmi-sdi and sdi-hdmi conversion at the other end of the long cable, did not add any noticeable difference between the camera signals. :)
I remember ATEM Mini input 1 being special, and having the option of direct pass through to the output. Would be interesting to see if your setup could tell the difference. My guess is the other inputs sync to input 1, so that one stays least effected. Maybe only shows once multiple inputs are already synced. Really don’t want to confuse the topic, honest.
I would also like to see further research into this exact scenario
The test isn’t valid. Latency has nothing to do with the cable but rather the time the camera takes to convert the signal to HDMI. Using a cross converter won’t work either as those devices typically use chips that are much faster at converting. What you need is to compare a camera like the Sony FX6 outputting SDI against a say a Sony A7 series camera outputting HDMI pointing at a timecode reference. I’m not sure how much of a difference there would be but as others have pointed out, the GH5 has a significant delay on it’s HDMI output.
At the time this video was created there was a rumor circling the Internet that HDMI causes latency. It was not long after the ATEM Mini had be come popular, and multiple UA-camrs and others who are influencers in the video world kept spouting the nonsense that HDMI causes latency and that using HDMI was the cause of there being a delay on the Mini and other switchers. This video was meant as a way to dispel that rumor. The cause of the delay is the cameras that those people were using (typically DSLRs or mirrorless ILCs), not the type of connection.
Thanks Doug
1:40 in, it looks to me like the left insert is 1 frame behind your recorded camera shot, and the right insert is 2 frames behind, which is what I'd expect.
But I've been looking at video signals for 35 years or so... civilians won't notice, and I'd put either on an IMAG screen.
We have always run into issues when deploying video monitoring of a Maestro / Conductor so that a singer viewing at an off angle or from another location can sing in time with the orchestra. So, I never did this level of testing - and I totally appreciate the work here - but having tried to move our systems to digital without telling the musicians we get notes about how the singers are getting behind or off the beat. I can't quantify the frames-per-second but it has seemed that any latency at all is problematic for live classical musicians.
But ... if HDMI as a format isn't the culprit but rather the conversion for long runs of HDMI/SDI - SDI/HDMI ... can you do a video on the gear you recommend for a near zero latency digital setup?
The trouble with digital signals is that no matter what you do, there's always going to be some latency. If zero latency is a requirement, going back to old-school analog cameras and CRT displays is really the only option that delivers on that promise.
With that said, there are some cameras out there that have low latency, and some monitors can be put into a mode which reduces the latency significantly. Off the top of my head I can't think of any cameras that have latency less than about 50ms, and in terms of monitors you'd want something that's from a high-end professional line, or at a very minimum a TV or monitor designed for gamers.
Long story short, there isn't a great solution that I'm aware of for your situation.
Conversion is not the problem, try using 2 cameras, one on sdi one on hdmi on 2 inputs in ATEM and than look in the multiview from the ATEM.
You’d have to use two identical cameras which are synchronized for that to work. I don’t have any that have Genlock inputs and an HDMI output.
You and me are the same. Love live video production, video, equipment, and methodology. But hate editing video. Haha. Too tedious. I get more adrenaline by doing live TD (which, fortunately is now happening more and more)
Love the videos. Keep them coming.
So... How does cable length affect latency? Is there any more latency over a 50' 3G-SDI cable than a 100' or 150' cable? We do 50' cables mostly, but with more and more productions coming in, the need to expand our system will become more prevalent.
Not noticeably. Signals travel at a pretty substantial fraction of the speed of light, so you'll never notice it.
@@djp_video yeah, ok. I didn't know if there wasn't some latency induced by the medium itself. Obviously fibre-Optic is lightning fast, but I thought maybe copper would induce some latency. Good to know it's not much of an issue. Thanks Doug!
Dough what I have noticed is the problem comes when hdmi is directly coming from camera. In your test setup the feed is coming via sdi than its getting converted by BMD converter.
The purpose of the test was to see if HDMI itself causes any meaningful latency (which it does not), not to test how much latency there is on the HDMI output of a camera.
Every camera is going to have some latency, and less expensive models have more (as their electronics for producing the HDMI signal aren't as fast as higher end models).
From my own prior testing, I know that the cameras that I use have exactly the same latency on the HDMI outputs as the SDI, so taking the signal from HDMI instead of SDI wouldn't have altered these results at all.
Thanks for this Doug. I think I’m amongst the many people who’ve never worked with SDI, and therefore believed what we’d heard about HDMI vs SDI.
Correct me if I’m wrong though, but you did just demonstrate that BOTH have a significant delay, right? Which is why audio sync is an issue with external audio devices on the Minis.
And I’m not sure I understand your comment about games consoles. As I understand it, a delay of audio and video (together) of a few frames wouldn’t have any impact on a game... Only a lag in control input would cause that. So the delay caused by HDMI/SDI is no issue. Or have I missed something?
The delay that people are complaining about with the ATEM Minis is coming from the cameras. Neither SDI nor HDMI themselves have any sort of perceivable delay to them. (There's a tiny bit, but it's a few milliseconds at worst.)
Every digital video camera has some delay from the time the image enters the lens until the digital data comes out the output. It's just the nature of digital. Higher-end professional cameras keep this delay to an absolute minimum, as there's a pretty good chance they'll be used in a live IMAG situation (that's Image MAGnification -- or in layman's terms, cameras being displayed on a large screen like a projector). Lower-end models, including most consumer cameras, don't prioritize low latency output, as they're really designed to record video for future playback, not immediate display in a live situation. Keeping the output latency low requires fast processing, which is expensive, so lens-to-output latency isn't important to camera manufacturers making consumer-grade gear. As the ATEM Mini is rather inexpensive, it tends to get used with lower-end cameras, which has led to a lot of confusion in the growing area of brand-new live streamers. A lot of bad information is being spread around.
My comment about gamers is because as a general rule they're very picky about latency. Fast-action games really require an absolute minimum amount of latency from controller input to display output. Trying to play a game where your input is out of sync with what you're seeing is extremely frustrating. If HDMI had a significant inherent delay in it, many games would be near unplayable on any console connected to a TV or monitor with HDMI.
Gaming requires minimizing reaction time. If the audio and video are delayed relative to the human pressing a button -- regardless of whether the delay is introduced between the physical button press and the resultant on-screen change -- worsens reaction time. Especially for retro game speedrunners, a delay of 1-2 frames is enough to completely throw off the tight reaction times required.
I never heard that myth from anyone that knows anything about A/V but I'm pretty sure the myth comes from the influx of DSLR's being used on the lower consumer side. We all know a consumer-grade HDMI out on a DSLR can add many frames of delay. They then get their hands on a pro camcorder with SDI out and notice zero delays not realizing the hdmi out on pro-grade bodies will also have no delay but they in an A cam B cam scenario blame the hdmi out when it is actually their GH4 or something.
That's pretty much it. It accompanied the release of the ATEM Mini, and a whole bunch of UA-camrs got their hands on them. Some of them started telling people that HDMI was the cause of the signal delay. Which, of course, it is not.
Thanks for taking the time to make videos like this. It's appreciated.
Now that we know the delay of SDI to HDMI and back, it has me wondering what the latency is of other mediums? For example SDI to IP, where it's not just converting the signal for copper or fibre, but actually has to encapsulate the signal into packets, especially with hardware SDI/NDI converters becoming more common.
Going to IP complicates things greatly. There's the latency of the conversion (which is likely to be significant) plus any delays in transit, which can add up depending on the number and types of devices in the chain. Then you've got conversion back the other way as well.
Curious then... when i run HDMI from Canon G20 into my ATEM, the audio via XLR into the ATEM is out of sync by 200ms or so. I am using a behringer device to add audio delay to mitigate. What am i missing? (many thanks for this btw)
there is HDMI latency as demonstrated by his video when the clacker in his hand shot 7 frames in front of the HDMI and SDI signals...the analog audio has no latency...ergo you end up with a sync problem...
All digital video cameras have some output delay -- in this case it seems to be less than 3 frames (the monitor has some latency as well). Audio sent through a camera is given a similar delay to maintain lip sync. But audio mixed in later can be a problem. With my own setup sync isn't really much of an issue, and if it was I could delay it in my audio mixer.
@@djp_video Got it -- so basically there is latency, just not signficantly more so in HDMI as compared to SDI...
What about the cameras HDMI outputs? I use ATEM Mini with three different cameras, still cameras with HDMI out. Nikon D5300, Nikon D850 and Nikon Z6 ... The picture entering the ATEM Mini is not in sync. And if I put audio directly to the ATEM Mini or my computer. It will be out of sync. I do not blame the ATEM Mini ... but the fact that the three cameras take different time to process the video. The only way to get audio in sync is to go via one of the cameras (I chose the Z6), but then it is only that camera that has sync ... :(
It really helps to be using the same model of camera so the delays are the same. There isn't really an easy way to overcome varying delays introduced by different camera models. (That's why we try to use cameras meant for live work when we can.)
@@djp_video true, the day I see a used Z6 on sale here in Sweden ... I will get it!
@@djp_video That's right, you should be using one camera with an SDI output and an HDMI output, directly connected to the monitor via native inputs. Definitely not running through converters first.
@@djp_video If he is using a Blackmagic switcher can he try using AFV so whatever camera is live the audio feed is coming from that camera? That's assuming he can get audio feed to each camera from the audio board.
I don't have any cameras that output SDI and HDMI simultaneously. Nor do I have a pair of A170s, or a high speed camera. This is the best I can do with the equipment I own. But it clearly demonstrates my point -- that HDMI doesn't inherently add significant latency to a video signal.
Online forums are filled with "because you're using HDMI" as an answer to "why is my video delayed?" and that's what I'm trying to address.
There are myths around that there is a whole lot of problems with the atem mini not synching video and audio but that is just as true as that there is a delay within the HDMI signal. Cams and tech round cams what ever it might be of course adds delay it´s not magic what they do and the more stuff you add to the chain the more delay you get. If that wasn´t the case they would be time machines to look in the future with if you could tweak them to convert faster. I have done numerous streams even in 720 p at very low speeds from 800 - 2000 kbps at 24 fps and the video and audio is in synch. I have watched it on line and downloaded the files and uploaded them again and still no delays in any way.
Nice Video ! Greetings from Germany 🇩🇪
So if I go HDMI directly from the camera that should be the least latency possible, as I plan to run hdmi with one of those no latency hdmi over cat cable setups and into probably an atem mini pro!
The lowest latency possible with your camera, yes. But cameras all introduce some amount of latency, some (DSLR and mirrorless in particular) more than others.
HDMI definitely does add a noticeable amount of latency for certain things. Live audio production going through HDMI has proved to be impractical for me. Too much of a delay between me playing the note on my instrument and the sound coming out of the speaker. When I use S/PDIF for the same thing, the latency is so low that it is undetectable. Maybe it matters which HDMI cord you use though. The cords I used were pretty short, so I wouldn't expect the length to be a contributing factor. Maybe build quality has an impact?
It wouldn't be HDMI adding the latency -- it would be the connected equipment.
If HDMI inherently had latency, gamers would never use it because it would make games unplayable. But they do with the Xbox and PlayStation and gaming PCs and other gaming devices all of the time.
@@djp_video I use the same connected equipment and when the only difference is HDMI vs S/PDIF, there is a noticeable difference in latency. All professional quality live audio production software uses specially designed sound servers that can get much lower latency than what would ever be possible by default. Gamers would not possibly be able to benefit from them since games are not designed to use them. Live audio production deals with lower latency audio than what gamers need or are accustomed to.
It is the connected device that creates the hdmi, that is adding the the delay. If you feed a camera with analog audio, for example, it has to do analog to digital conversion and then mix it with the camera signal into hdmi. The better the camera, the less latency. I would never lead audio via hdmi if you need the signal live. The few frames latency is usually not a problem for video however at a concert, for example, when there are huge video screens next to the artist that show the artist, then you need expensive camera's and signal converters to keep the latency as short as possible.
Well I have a real world story. Video playing on a computer output into an HDMI switch into an HDMI splitter to a projector and TV in the sanctuary and TV in the lobby. audio going out to the mixing console to ceiling speakers amplifier blah blah. Visible bad! delay between audio and video in the lobby. Almost like a few seconds if I'm remembering right. HDMI has delayed because of the tv processing the signal tend to put a ton of processing on and dilley the audio to match. You usually have to dig around in the settings for a while to find some sort of game mode that disables it. Otherwise you have to delay the audio but that doesn't work for live things or things say on the other side of the wall of the lobby where you can hear the difference.
I think the main issue is when folks try to mix sdi and hdmi only consumer cameras and their are talking heads involved. Or even mixing different kinds of hdmi only cameras in a live production. I always thought that different manufacturers hdmi signals were delayed by varying degrees. Even if there’s a couple frames delay between sources, it can be noticeable. It would be interesting to see a similar test using different hdmi sources (mirrorless or dslr style cameras) going into your switcher and if there’s audio drift over the span of an hour or so. It looks like there isn’t any inherent delay with the hdmi signal itself, but is there still delay in practical terms when using mixed hdmi sources that are delayed from the camera side? I don’t know. I’m still on the fence on this one.
It's worse than just per-manufacturer -- different models of cameras have different delays as well.
Drift shouldn't be a problem, at least not with professional video gear. Many of the computer capture systems seem to have problems with that, though.
Is there any way to synchronize the delay that may exist in the video from different cameras in the atem mini or atem software? as they do with the audio
Can someone tell Blackmagic that the OBS guys have a solution for video lag not just for audio xd because they don't do anything about it
Very interesting video, but I have a question for you. If I make a multicam with three cameras, two entering the atem via hdmi, and one entering via a wireless transmitter like the mars400, how do I delay the two hdmi input cameras so that all three are in perfect sync?
There's no easy, practical, affordable way to do that. The better solution is to get a wireless system with no latency.
@@djp_video strange that there isn't an easy way, because I think it's a common problem, especially when you have a streaming system for events where you have cameras on stands connected via hdmi or sdi, and then a mobile camera on a stabilizer, where it's difficult to manage with wires and therefore you have to opt for wireless systems, but even good systems like the cosmi c1 bring some latency…. I saw that obs video input can be delayed but not on the single hdmi but on the whole ip video module by a maximum of 100ms, strange that with blackmagick cobe software you cannot set the lency as you do with audio.
Delaying video is the hard (and expensive) thing to do. Every frame of 1080p video requires a minimum of 8MB of memory (12 MB if you want to do it without significant CPU processing), so if you wanted to delay every source of an 8 switcher input by up to a half a second it would require 3 GB of fast memory and a CPU with a lot of bandwidth to do so... so beefier cooling systems are needed. It all gets expensive and requires more power and is bigger.
In addition to that, when you do that everything you're doing and seeing on your switcher is delayed compared to real time. And the audio has to be delayed too. If you're technical directing on-location where you're hearing and/or seeing the live action, you're likely to react early and cut between cameras before you should. And when you're giving direction to your camera operators, you're reacting late to what they are actually doing because you're seeing it delayed, after it has happened.
The easy solution is to not have any delay at all on your video sources. And good quality wireless systems have zero (or near zero) latency. They also let you use more than one wireless system at a time, which the Mars 400 won't.
Nice work Doug! kudos
Interesting video- as always. You're using an expensive SDI camera. Many people watching are using HDMI cameras (consumer / prosumer and not the more expensive professional-grade SDI). Consider running the test with an HDMI test camera. Thanks.
The only difference that would make would be how much initial delay there is from acquisition to display. It wouldn't affect the amount of delay from the SDI-HDMI conversion.
The amount of delay on the HDMI output of different cameras is going to vary. So testing one model also won't tell you anything about how other models perform.
@@djp_video I agree with you. I'm just highlighting that people (including me) are disappointed with overall HDMI delay. Sir, I would say it’s not a myth, but rather a point of perspective. I see a huge delay and I believe this may be caused by my camera (a $600 HDMI Sony FDR AX33). I understand the focus of your video is the transmission path and the converters (not the camera). However, you are giving a big sigh and blowing off the frustrations of many viewers ; ) Yes, the converters may only add a frame or so of delay, but the system is only as strong as its weakest link. I would at least acknowledge the problem of HDMI camera delay. That said, your test did strongly confirm that the problem is not in the BlackMagic converter and Switcher- great to know! Respectfully offered. Thanks again- I always enjoy your video’s.
I'm not trying to discredit the frustration. I'm just trying to point people in the correct direction as to the cause of the problem.
HDMI has a higher color bit depth than SDI right? So if you don't have a need for long runs isn't the color better slightly? Can you see the difference, or is it at the very least enough to affect chroma keying? I've really enjoyed your more advanced stuff, but there are so many misconceptions in our industry that going through the basic stuff can be quite enlightening sometimes. Looking forwared to more of your videos!
SDI is typically 10-bit, 4:2:2. HDMI can do 12-bit, 4:4:4, but in practice that doesn't really ever happen. (For reference, Blu-ray and consumer video is usually 8-bit 4:2:2 or 4:2:0.) You might be able to get 10-bit 4:4:4 out of a camera, but without going to a proprietary RAW format that's about the best you could do. I strongly suspect that the ATEM Mini and other HDMI switchers don't process in 12-bit 4:4:4 anyway, even if that is offered by the camera.
In terms of whether the difference is visible, for most content it won't be. If you've got high resolution, very small color text on a differently colored background (red text on a blue background, for example), yeah, you'd be able to see the difference (white, gray, or black text won't have a problem). And it likely would get blurred by final rendering/encoding so it probably doesn't matter once you get to final distribution.
HDMI is up to 48Gb/s now so faster than 3/6/12Gb/s SDI. For computer screens with static small text 4:4:4 is a win over 4:2:0 etc commonly used over SDI. I have no experience of chromakeying, but doesn’t that work well enough since the transition to HD if lit correctly? Greater color depth - good for HDR and heavy grading in post.
Unless you are pixelpeeping or doing heavy grading in post, I’m unsure if normal people can tell video signals apart. I sampled a bunch of 4K Blu-ray’s and pixel peeped them, they were all crazy unsharp no sharper than HD, but sitting in the couch a bit away HD and 4K all look fine.
Wait a second. Your test then is totally useless. Isn't it? Because you ran the HDMI with the Converter and then blamed the converter for that one second delay. A reliable test would be to run HDMI as it is from source to destination and just the same for SDI. Or am I missing something here? 🤔
It's a 1-frame delay, not one second.
The bottom line is that beyond the delay from converting to SDI-to-HDMI and then back again, there isn't any additional delay. HDMI is safe to use for live production in terms of latency.
reasoning is sound test however is not and would be torn apart by any detractors easily, see this all the time in audio crowd. HDMI is still a crap video cable from design to what they run through it to the Frickton of useless "Features"
It’s the hdmi cec that adds latency, aming feels floaty etc with it on
If that was true, game consoles like the Xbox and Playstation would be too laggy to be playable. Gamers would refuse to use it. But those consoles do support CEC.
CEC is a very lightweight layer added on top of other HDMI features and it shouldn't add any latency whatsoever.
@@djp_video you could try gaming for 5 minutes with Xbox with it on and off (it restarts every time you change this setting) and you’ll feel the difference… try with Xbox hdmi cec on and lg oled simplink on… then try with both off
@@djp_video LG CX has hdmi cec disabled by default… Xbox has hdmi cec disabled as default… many people don’t mind extra input lag, not many notice it…. It’s like many people use BFI (black frame insertion) on LG OLEDs at the cost of input lag and they still use it, they don’t notice it but it does add input lag, capish? Do you use black frame insertion? If you don’t notice the added input lag there that’s why you may not notice the added input lag when hdmi cec feature enabled on console and tv
@@djp_videoBFI (black frame insertion) it’s “not too laggy”… a lot of gamers use it and it does adds latency
@@djp_videoeven digital foundry recommends BFI (black frame insertion) for some games… and it does add latency, but still playable… 13ms with no BFI and around 30ms with BFI and many gamers consider 30ms latency as playable, many don’t notice it, i can tell you didn’t either… but it is a difference in latency… it’s not like 100~150 ms difference, that is unplayable… anything under 40ms is considered playable by many… but there it a difference between 10ms and 30ms etc… for it to be no difference would be to be the same number ms… like saying volume 10 and volume 15.. latency is exactly the same… but with console AND tv with hdmi cec enabled there is a difference… as well when BFI is enabled
I notice there's a 1-2 second delay from my Nikon D850 to the computer / OBS. Which is what brought me to this video. I've learnt from this video that the cable is likely not the issue, and the issue is likely with the output from the camera .. do i understand this correctly? is there anything i can do to reduce this delay? The HDRI is hooked into a CAM link 4K. Thank you.
The camera probably has a pretty significant delay in and of itself. The rest is probably being introduced by your capture device. I've used OBS with camcorder-style cameras going into both a Blackmagic Design DeckLink/UltraStudio or some ultra cheap HDMI-to-USB capture devices and haven't seen what I would call significant latency with that setup.
@@djp_video Thanks Doug, much appreciated. Do you feel there's anything i can do to reduce this on the camera side, or is this simple a nature of the camera design and typically something that cannot be changed without buying a new camera? Thanks again.
@@davidedwardsme There's likely nothing you can do without replacing the camera. DSLRs and Mirrorless cameras aren't designed with low latency outputs in mind -- it probably isn't even on the list of priorities when designing the camera, since they're far more likely to be used for cinema-style production (where latency isn't an issue) rather than being asked to handle live work. Camcorders are far better at this task.
120fps?
No
Good video, how can i reduce 8VSK latency (atsc) computer output via hdmi to ATSC modulator and that feeds TVs .. so much latency and all the tvs play the video and audio at different times even if they are the same brand and model
The very nature of ATSC requires some time to do the encoding, so there is always going to be some latency. If you can replace the ATSC solution with something else you'll be better off. SDI signals travel over RG-6 cables quite well, so you could swap out the RF distribution box with something meant for SDI, then use SDI-to-HDMI converters at the TVs for perfect quality and sync.
Great work Doug,
but what about cable lenght, does it matter? SDi can do 100m +, HDMi can't, what if you try for examble 20m SDi, 20m HDMi (where HDMi starts to have quality issues etc.)...
Thanks and please keep up uploading 🤙🏾😋
SDI is a better way to go whenever you can. And, no, cable length doesn't matter to latency in any kind of practical terms.
cable length issues in HDMI are due to the fact it uses multiple pairs in parallel, which leads to bit-level desynchronization after about 10 meters or so. the receiver can't make sense of the information
Sdi sends everything through 1 wire, which makes it impervious to those issues.
it does have signal level loss issue though, which means the receiver is not sensitive enough to detect the voltage variations. that is easily fixed by using fiber transmitters.
note: optical hdmi cables multiplex the signals for all hdmi pairs onto one fiber, and carry the return signal on a second fiber, correcting the issue, but they get much more expensive than sdi transmitters at that point.
Amaury Jacquot hmm, Wikipedia and common lore reference signal attenuation and cable quality issues, not TMDS breaking apart. Not saying you are wrong, desynchronization on a high speed bus seems possible. Very unsure why anyone would design a cable standard using 8/10 coding *and* parallel *and* a dedicated clock, it’s clearly a very weird cable design. They just took a lot of different cable/signal technologies and put it in a bizarre mix. Any other cable would be serial or parallel, and no one would have a clock on a high speed cable (static jitter is emi possibly disrupting other electronics, such as the data wires...)
I would like to see that test, in my experience, I have seen the delay at very long distances.
One question: i recently livestream a lot and use 2 SDI OUT camera’s and 2 sony alpha with HDMI out. The Sony’s defenitely have a lag compared to the panasonic SDI out cams. Is this due to the fact that the HDMI ports from the sony alpha’s lag? I’m confused.
We also use a blackmagic ATEM
Yes, the HDMI ports on those cameras do have a delay.
You’re using cameras meant for a film-like production workflow. They weren’t designed for live use. So having a minimum amount of lag (latency) was not a priority when they were designed. If you need a minimum of delay it is best to use cameras that were designed for live production.
@@djp_video okay, thanks for the info. Bummer that they do that.
Best part at 7:19 perfect music =D
My question is: how many fps do we see on the monitor?
60 FPS
Great job. Looking forward to more instructional videos!
it feels like the black level of my monitor affects my delay, is this true?
It shouldn't. But higher quality monitors with better black levels probably also have lower latency. It's a correlation, not a causation.
@@djp_video alright thanks, and yeah i have a pretty good 240hz monitor
16ms 🤷🏽♂️
Likely less, but yeah...
Why would you do do a latency test by running sources through converters... smh
If you can figure out another way to provide simultaneous, guaranteed to be as close to in-sync as possible -- SDI and HDMI simultaneously, I'm all ears.
@@djp_video you could use 2 of Z150s, one on SDI and other on HDMI, and if their outputs have identical delay, by randomly turning them on and off it would show a 1 frame delay randomly on any monitor input
Dislike due to improper test condition/setup. A test with hdmi cable vs sdi would be a proper test. As stated in the video he added fail points but still originated with sdi feed. A pure hdmi feed is the “myth” of added delay.
It's still an HDMI signal. It doesn't add meaningful delay. I confirmed that with BMD.
Finding a camera that outputs SDI and HDMI simultaneously and to have two identical low latency monitors that both support SDI and HDMI makes doing this test properly quite problematic.
The myth of the added delay comes from people using cameras that aren't meant for live use and don't make outputting video in real time a priority. It's very common on DSLR-style cameras so people erroneously associate HDMI with the problem.
I use HDMI all the time for live production with my Z150s when I'm shooting in 4K (as they won't output 4K via SDI) and there's no noticeable delay to the signal.
👍🏾🙏🏾