Digital FPV Latency & Frame Delivery - DJI - HDZero - Avatar HD - A Lot To Understand.

[@MadRC]

Hi everyone, there has been a number of common questions on this video I wanted to answer. 1: Why did you not test Avatar HD Race Mode ? A: Racemode on Avatar HD Remains 100fps and while it reduces resolution to improve overall stability the overall minimum latency is with in error of being the same as 720p 100. Racemode does appear to offer a little more latency stability in flight but that is even questionable. Overall it’s the same as 720P100. 2: Why did you not test Avatar Goggles X. A: This was tested in later data as shown later in the video. The Goggles X do actually show a little higher overall latency than the original Avatar goggle however again this is very minimal and I would not class this as a definitive issue. 3: Why did I not test HDO2 Goggles on analog A: I don’t have any to test. I can only test the products I have in hand and that consisted of the goggles shown. 4: Why did you not test in real world like inflight? A: Today this is close to impossible to do in real world situation to show the actual variable latency effect of DJI and Avatar. Avatar does log the latency however O3 does not and even then those numbers are based on the systems internal calculation and is not absolute fact. 5: Why did you not stress more about this being the best case scenario and that DJI and Avatar are variable latency ? A: Did you even watch the video ? 6: You have made my favourite digital FPV system look bad and I think you’re a shill for the alternative system I don’t like! A: Thank you for your constructive feedback

[@tk-maker]

I was thinking conduct the test somewhere behind a wall, make the signal at weakest and see how the latency go.

[@kithng]

For HDZero, I think not only for racing but also great for long range because of it's similarity to Analog which give you a much better idea of your video signal strength.

[@MallyFPV]

This is top notch quality content, well done! Glad to see that the Avatar VRX performance on HDZero goggles is very similar to the Avatar goggles themselves.

[@mikerich32]

I'm glad to see that too! When I decided to upgrade from analog to HD, I went with the HDZero goggles. Now I know that if I ever want to do Walksnail, I don't have to worry about buying the Walksnail Goggles, and that an external module on the HDZero goggles is plenty good enough.

[@MallyFPV]

@@mikerich32 yes! the only downside is that the goggles becomes pretty heavy, but its manageable, I run this setup most of the time unless only flying hdzero, then I remove the vrx. Its great to have best of both worlds!

[@bernardfouche5241]

Thank you so much for this video. What an impressive amount of testing! My 2 cts comments: 1) why not show the standard deviation value for each set of the measurements ? This value is the consistency value. 2) A digital camera recording on a SDCard is a very complex system of its own. How to be be sure that the camera really records frames at 1KHz, that it is really functioning in "real time" ? (Photoresistors and an oscilloscope instead?). If you already have checked the digital camera timing accuracy in a previous video please ignore this comment.

[@palethiusfpv]

Oh man, so nice to have it confirmed…. I could never fly DJI as well as analog. I always blamed the variable latency, but having the image displayed in all sorts of random orders HAS to be a big part of it too. It always just feels a little off when doing fast/precise freestyle stuff. 😅 Glad to hear it’s not just me…. 👍

[@zonewolf]

this has been plaguing me since I only have two systems at the moment (v2 03, v2 dji fpv drone), and this quenches all of my questions and curiosities in latency between systems and frame rates! thank you Mads!

[@joeztan]

Great work! Will be interested in seeing this kind of testing on the next generation of digital video systems. DJI O4 vs O3 is certainly a topic I would be very interested in.

[@elxero2189]

The dji 04 will likely not come to FPV more likely for the standard commercial phantom style drones

[@joeztan]

@@elxero2189Yes 2024 will be interesting. O4 on the Mini 4 and Air 3 would indicate that it is cost effective on low cost drones, while delivering better performance. Would think that a future air unit (DJI O4 or whatever it is called) would likely leverage a common hardware platform and improve on what we have now.

[@nickgrushetsky4625]

Thank you for this comparison, it is very informative. I was just planning to switch from analog to digital next year and I understand that O3 as a replacement for an external camera still does not cover my needs. It turns out that there is no point in overpaying for a dji system.

[@MCsCreations]

And looks like there's a 4k camera coming for Walksnail... We just don't know how good or bad it's going to be. But who knows, I'm crossing my fingers.

[@j005u]

@@MCsCreations It will inherently have the same restriction. If you want smooth 30fps you're going to need to fly at 30fps, which will definitely have worse latency. There's just no way past this with a single sensor.

[@dronepilot260rc]

I’m still rocking DJI for quality image and analog for long range 🎉

[@tk-maker]

This is the most amazing data oriented reviews and again almost 1 hours of watch. highly recommended for anyone who wish to get a digital system. great job!

[@Quadwrex]

I had a hard time flying dji, so I went to hdzero. Bought the avatar goggles on sale and was surprised how smooth the video link was compared to my dji setup. Nice and informative. Your test results explain a lot. Thanks for the great content

[@BogHopperFPV]

Ian in his niche and at his very best- nevertheless I also value the full range that the channel covers. Thank you

[@testpilotian3188]

Oh I need to go lie down in a dark room, information overload lol. (Really good video, thanks namesake).

[@Lowbattery_fpv]

Top content ! personally i can't prefer image quality or best penetration to low latency. I sold my walksnail stuff, and switch back to analog and hd zero. Smoothness feeling is my priority.

[@1337plans]

@madrc Thank you for your content!!!! Amazing and very helpful!

[@RyanQuellet]

Very comprehensive results - tons of work! One thing I've been thinking of is how to make a test that captures link variance over time while flying. Maybe use a radio link to send a LED pulse to a camera or something while flying with another camera. Really hard to do, so probably not going to happen - haha 🙂

[@MadRC]

Yea doing it in “real time” is close to impossible, in avatar we do have their latency numbers in the srt however that could be total nonsense. Evernote if DJI are looking real data we don’t see again it’s not actual facts. About the only thing you could do is what Chri did is introduce artificial attenuation. However the issue is actually sorting the data as it’s a crazy manual task to do it as you have to physically look at each run. We really need a AI model trained to look at the screen with a camera identify and log the changes for both full frame and first pixel and then log it say 1000 times per mode. I don’t have the skill to make that 😂.

[@RyanQuellet]

@@MadRC Exciting ideas - AI would be interesting for sure

[@canigetachannel]

Thank you, fine sir. 👍 That's why I'm here.

[@Mupshot]

So when will they deliver on quantum image transmission where we see in the goggles before it's in the camera?

[@MCsCreations]

Fantastic testing and comparison, Ian! Thanks a bunch!!! 😃 I was watching the video earlier, but I had to stop it... Because I won a giveaway... 😂 Anyway, stay safe there with your family! 🖖😊

[@tendrugi2093]

Thanks for this great comparison and latency testing. Would love to also see a latency for HDMI-out when connected to TV and capture cards. Black bars on OLED screens are probably used for dimming image brightness. You could check it by changing the brightness setting in the goggles.

[@babc142]

This is an amazing resource, thank you!

[@Vousie]

12:30 I vehemently disagree with that. HDZero has the *most* link resilience of any digital FPV system. It holds on to the last bit, and even if you lose it, you can punch out and *immediately* get your image back. DJI/WS "look nicer" for longer, but once they fully drop the link, you've lost it for quite a few seconds (or until power cycle) and you've almost certainly crashed the quad. This is something that people doing mountain surfing have learned the hard way/ (Read: They usually used HDZero but went mountain surfing with WS to try it, went just behind a hill, popped up out of the hill, thinking the video would come back immediately but it did not. So the quad crashed and they were in for a few kilometers' worth of walking, which took about 4 hours.)

[@johnrosner8964]

Love your channels way better than bardwells

[@M-H433]

Agreed,Ian really deserves many more views than the other person.

[@bobtruck1594]

Oh damn. I just ordered the Avata Explorer for 890$. 300$ off. Anazon won't ship until Dec 10th. So i expect them by the 15th. Then add the 30 days return. So i have till Jan 15th to rerurn. Hope we ger more news

[@starsstripesjacket]

Awsome info. Would love to see the avatar race mode tested also 😊

[@dvsfpv8957]

this was Great Ian.

[@BaldmanFPV]

Waddd. Now I have to try the 03 at 100fps on G2 before retiring it on a whoop. 😦

[@kandredfpv]

Hey Mad, awesome analysis of these digital goggles, but did you miss testing the HDMI latency of the sky04x with the 2 VRXs? You had said during your last live stream that you would include this data. Thanks

[@Vousie]

I think we really need some of these tests done where you have a concrete wall or two between the VTX and the VRX antennas. Comparing WS/DJI's *absolute best case* with HDZero's *always* case is totally misleading, and not representative of what the latency for WS/DJI will be like when you're actually flying.

[@vertigo2893]

exactly. Many people will come away from this video thinking the variable latency of WS and DJI (which was NOT tested and barely even hinted at) is the same thing as variability in the test when screen and camera run at different refresh rates. Extremely misleading.

[@ibeginge113]

great video Mads , loads of info to take in. One question ref DJI 03 on the goggles 2, as you said and tested lowest latency is at 100 fps, would i be right in thinking this is 1080p 100? or does this cover all resolutions at 100 fps modes with the g2 and 03?

[@offtheground9152]

Thanks Ian!

[@zbluga4418]

Well done, really this must take lot of time to do it. Only think iam curius about is why you dont include result for the new Goggles X

[@unRealityFPV]

I have a feeling the discrepancy with frames not finishing on time or not starting at the top of the frame is the difficulty in having the first frames of the camera recording and the goggles start at the same time. I’m not sure how this could be accomplished with the current setup, but more accuracy would be achieved if shooting in 2000fps 4000fps 10,000fps etc

[@flymoretime]

Thanks for the great work you've done! And have you tested racemode avatar 540p 100fps. And meanwhile it has more stable latency than 720 - 100fs

[@gmivisualsjason3729]

Excellent video

[@kenb6565]

There is nothing better than a new, super detailed, and informative video from Juan Pablo...who? Juan Pablo😊

[@attackfpv]

Lol

[@JustInTime0525]

Great tests and video! For Goggles 2 running O3, does 4K/ 2.7K/ 1080p 100fps changes the latency? I do notice a FOV change when switching between these resolutions, perhaps you can make another video diving into these and whether recording or not also changes it, thank you!

[@rotormasher]

I believe he's already made a video on this

[@typxxilps]

I like the evolution of the channel over the past 5 years if I remember right cause I am a subscriber who subscribed when I got my red goggles racing edition and you were the first to build and show a head tracker which then became my major goal to build too. So quite a long time and I still remember the early Corona days and your goal to buy an atem mini or so which then in april 2020 or so had gone by 100 GBP due to lockdown and Corona implications. And I mention that cause in these days , those of the goggles racing edition RE you started to measure latency of goggles and so on - and now you finally got the camere that provided 1000 frames. I am still wondering what will happen with the whole goggles market considering that only 3 or 4 have made the jump from analog to digital and not more following. Will they all catch up to dji digital fpv or can dji stay ahead as in the past decade ? Will there be a major jump with dji fpv goggles V3 ? Currently it feels like all of the manufacturers have their own niche and try to improve their strength not so much hunting the competitor.

[@typxxilps]

and it is time that you get the 50.000s subscriber as a major mile stone on the 100 k hunt.

[@simonkrause9550]

Great

[@MrHeller55]

If I am understanding this data correctly with HDZero@90 you get a little over 71 frames of usable data per second, with walksnail and DJI you only get about 30 frames of usable data at best. Am I understanding this correctly? By usable data I mean that if something changes in frame due to the latency you only can see that change at fastest 14ms after it happens. Edit: I come from a PC gaming background so I am thinking about this in the same way framerate would work. In pc there is added latency on top of frame rate but since this is a total system latency test technically the frame rates of cameras don't really give you that rate in your goggles due to the latency. As an example 60fps (16.7ms per frame) on a PC game with a really fast monitor is usually 40-60ms response time between action and change on screen.

[@norquad]

Solid video as allways sir! But why didnt you add racemode on WS?

[@MadRC]

The latency is basically the same just more stable. It’s not any lower than 100 it’s just less variable and as such on the chart will look the same as 100.

[@norquad]

@@MadRC Ihave read the same just very curious if its sliiighty better. It sure feels better but that might be just the feeling since its more stable

[@he214abc]

how about built-in receiver goggles latency like ev800d

[@seb-depp]

Funny, I fly O3/G2 (and seldom Avata) with 50fps (so almost the worst) and before that Vista/G V2 and never notice anything. So I guess latency doesn't mean much to me.

[@FpvBerget]

Is there any reason u didnt test the 1080p 30fps mode on hdzero? Would be cool to see how mutch latency it really got.

[@cedricgrandseigne]

I would be interested to have the values for HDZERO 1080P30 and actual refresh rate: it seems like it is not even 30i/s.

[@califpv]

pick & enjoy the goggles you buy. theres a goggle out there for everyone.

[@P-Bee]

I'd love to know what latency I'm currently getting using the SharkByte VRX on my SkyO4X goggles. Also what would I get if I bought an Avatar VRX. Does anybody know of any sources for that data? I have a very narrow IPD and the O4X goggle optics are the best I've found, quite a bit better than the HDZero goggles.

[@RickLaBanca]

I don’t understand why you can’t use o3 as a GoPro replacement if low latency? Isn’t the onboard recording mode independent of the goggle video?

[@MadRC]

No the latency is tied to the recording frame rate.

[@RickLaBanca]

@@MadRC I thought it was 4k60 or 2,7/120 no matter the goggle rate. How are they tied together, if you run higher frame rate you can’t do 4k?

[@wisdomfpv1499]

Curious why you didn't test the race mode 540p for walksnail if the. test was about latency. Also fatshark hdo2 with module for analogue latency

[@MadRC]

Avatar race mode is the same latency as low latency more from a lowest latency point. CaddX say it’s more stable and my race mode flight tests suggest it is however the actual latency is the same. As for hdo 2 I can’t test what I don’t have.

[@wisdomfpv1499]

@@MadRC fair enough and thank you for your Work. I personally see you as the pinnacle of yt fpv content atm. Keep it up!

[@Vousie]

Why is the G2 03 100fps so low (34ms full frame), whereas on the same setup, all the other modes - higher and lower framerage, i.e. G2 O3 60fps, G2 O3 120fps, are all nearly double that? (62-66 ms full frame) That's a *huge* difference. And I noticed you just have the framerate, not the resolution you chose.

[@MadRC]

Resolution had no impact on latency. As for 100 vs the rest it’s because G2 has 100fps screens. Move off the native resolution and it turns to crap.

[@Vousie]

@@MadRC Thanks. That's quite nuts though - that it doubles - even when you've picked a higher frame rate... I would've expected to see more people saying "always use 100fps, not even 120fps".

[@jean-nicolasmenu-galinier8221]

How come the G2 O3 at 120fps has a much higher latency? Shouldn't we have a latency lower than 100 fps?

[@MadRC]

No because the G2 screens can only do 100fps so it ends up using 60fps for the link and doubling the frames it seems. Again 120 is not the native frame rate of the G2 goggle.

[@jean-nicolasmenu-galinier8221]

@@MadRC ah ok that's why vertical synchronization only works at 100fps! is the latency the same whether in 4/3 or 16/9? there is also a subject on the crop performed on the image according to the resolution, fps, and 16/9 -4/3. The crop is different depending on the fps as well.

[@gprov590]

Your TX is charging for too long 😂

[@zeno78200]

I don't quite get... why did you use the old walksnail instead of the new goggles x? Is the difference so high that you did not want to show it?

[@MadRC]

No the testing was done before the goggles came out. This is months of testing. The result is with in 1-2ms. It’s negligible.

[@zeno78200]

Thanks a lot for clarification!

[@vertigo2893]

I really wish you would have at least touched upon the effects of signal degradation on latency.. People are going to come away from this video confusing "variable latency" with the variability in your test results which stems primarily from camera/screen not being synced giving perfectly predictable random variations. These are very different things. HD zero latency is not going to be affected at all by range or signal strength, DJI and WS absolutely are and you only showed the ideal world perfect condition for them. I understand it would be difficult testing this thoroughly, but some indicative tests with attenuated signal would have been very helpful. Also, I think you should stop saying low latency only matters to racers. It matters to everyone pushing their limits, even (or especially?) people flying whoops indoor, and especially people who arent as skilled as bubbyFPV and similar pros that can do tricks blindfolded, or drunk because they have the muscle memory and can compensate for latency. People like me need to react to what they see in the goggles, and 30ms extra latency means ~30 degree pitch/bank angle difference at full stick.

[@MadRC]

I did. I specifically mentioned that it will be affected. How much more you want to me to stress it’s variable. This is best case scenario. I said this clearly enough

[@slappomatthew]

Walksnail is better than the other systems. all things considered. and its only getting better while DJI is getting worse.

[@ydnubm]

first

[@johnrosner8964]

The latency from your brain to your fingertips is approximately 50 milliseconds. So any latency lower than that isn't going to matter. So my advice is to buy the d j I goggles

[@nwfpv8681]

That’s not really how it works. If the latency is 35 milliseconds from glass to glass, then 50 milliseconds from your brain to your fingertips, then those two things add on top of the latency from the radio link to the quad. It all stacks up, so reducing latency at every part of the process is going to make a huge difference.

[@git8430]

Exactly so you end up with the sum of those two latencies. That is why I am not really sure why you are recommending dji.

[@johnrosner8964]

@@nwfpv8681 ok🙂

[@johnrosner8964]

@@git8430 🤑

[@elxero2189]

If I had a penny everytime someone tried to sue that argument... Sigh. The simple fact is it does matter and some people ie racers depend on it period end of story. Just because some can't feel it doesn't mean it's the rule for everyone even the great JB had to conceded in his hdzero 90fps video. Sorry but it matters .. not to everyone obviously