Some more information on WDR courtesy of Carl from HDZero! ♥ There are two types of WDR cameras available: True WDR cameras that work with a shutter speed technique and two images. Digital WDR (DWDR) cameras which work with a software compensation technique. True WDR cameras work by capturing two images with different exposures to light and merging them into a third one, and therefore works more intensely but at cost of latency, whereas the digital WDR works only with one image to make a digital exposure compensation, with only several line latency which is neglectable. RunCam and Foxeer confirmed that most of cameras they have made use digital WDR not only because of less latency but also less cost.

@aphinion

Жыл бұрын

Which cam did you use in your tests then? I was pretty surprised by the 40ms latency claim since that's clearly not what I was feeling with my runcams/foxeers vs. dji. but who knows!

@ChrisRosser

Жыл бұрын

@@aphinion I used Turbowing Cyclops, RunCam nano, RunCam Eagle Micro, and Foxeer Falcor

@udyfpv

Жыл бұрын

@@ChrisRosser I'm guessing the latest Falkor 3?

Wow HDZero is amazing! And DJI 120FPS not to far behind. Analog with WDR was a huge surprise to me! Thanks Chris!

@KOrnhOliO1

Жыл бұрын

Yeah, now if it could only penetrate objects, we'd actually have something! 🙄

@spikester

Жыл бұрын

@@KOrnhOliO1 I'm not sure if hdzero uses LTE but that might explain why DJI/WS get far better penetration, however LTE requires 2 way comms for realtime link quality monitoring from both ends to know which packets to prioritize resending, it isn't just due to the H264/H265 compression those systems use.

@spikester

Жыл бұрын

Now... this is something hdzero could benefit from via an RC link callback of link quality via the hdzero backpack, can hop or boost power output of vtx in realtime when received vtx signal degrades.

@KOrnhOliO1

Жыл бұрын

@@spikester why not just fly it on full VTX power? You're describing what dynamic power does for Crossfire receivers. 😎🤘

@Vousie

Жыл бұрын

@@KOrnhOliO1 Well if DJI would only stay at that 31 ms latency instead of dropping to 60-100ms when it has objects in the way then "we'd actually have something"... 🙄

Just saying thanks for this thorough testing and analysis. I've watch this video a few times now and keep learning new stuff each time. You're truly a treasure to the FPV community.❤

@ChrisRosser

Жыл бұрын

Glad it was helpful! Thanks for the support :D

Keep it up the good work Chris! Thanks for all the time you spend doing videos! Thanks a lot! 😃

@ChrisRosser

Жыл бұрын

My pleasure!

I think a clarification is in order about delay from drawing in scan lines. This is a penalty for a frame based video system, but not a scan line based video system. Drawing to the OLED in scan lines is in fact the fastest way to present new visual data continuously with analog and hdzero, and is not a latency penalty as it sounded like in your description. As long as the camera is rolling shutter (and it is here) and the data is streamed instantly with no whole frame delay/sync (which it is). Thanks for the data, I will test the hdzero goggles latency soon and we'll be able to compare notes 😉

@ichi1082

Жыл бұрын

So how is the rolling shutter synced to the refresh rate of the goggles? (pls dont say sync pulse)

@ChrisRosser

Жыл бұрын

You are spot on. This is covered in a lot more detail in one of the other latency testing videos for Analog, DJI and HDZero.

@RyanQuellet

Жыл бұрын

@@ichi1082 a phase lock loop

@hristiantodorov3923

Жыл бұрын

I am guessing the refresh rate of the goggles is synced to the first scan line received and then displaying data as it comes ... maybe ?

@ichi1082

Жыл бұрын

@@hristiantodorov3923 No, as goggles don't display interlaced video.

Excellent presentation of the data Chris! I look forward to more!

@ChrisRosser

Жыл бұрын

Glad you enjoyed it!

Thank you for putting this info into such a great package to help us digest. The way the cameras work was very interesting and something I hadn't understood was effecting what I see in my analog system. Fantastic testing.

@ChrisRosser

Жыл бұрын

Glad it was helpful to support your own testing! 😁

Great comparaison ! Thanks Chris.

Thank you Chris. Much needed comparison.

GREAT JOB. THANKS FOR PUTTING THIS TOGETHER FOR US

@ChrisRosser

Жыл бұрын

My pleasure!

Excellent review! Thank you Chris.

I love this testing for sure, it actually shows what these systems are actually doing not what’s advertised. I’m glad that the market is getting competitive because it brings more options for us pilots to use. Thanks 🙏 👊 for the work. Now real world 🌎 use is always a slightly different experience with the way the systems handles rf environments and break up.

@ChrisRosser

Жыл бұрын

You are absolutely right. With DJI and Avatar systems the latency can go up with interference which doesn't happen with HDZero/Analog. But the tradeoff is much better penetration because of the retransmission.

@Vousie

Жыл бұрын

@@ChrisRosser Yeah, that's a *massive* difference. Avatar especially seems to go over 100 ms (even 300?) as soon as it struggles a little bit. Personally I'd like a system that does the retransmission, but cuts it off much earlier than either DJI or Avatar do - eg never gaining more than 20 ms when there's interference - once it's at that added 20 ms, it should just show whatever it has instead of continuing on trying to resend while showing us the old frame, thinking we won't notice on a quad going over 100 km/h....

so badass, thanks for the breakdown

Kudos to our work for the FPV community with all your knowledge.

Great comparison and in depth analysis of the latest systems we are using. Hope Walksnail is doing more updates to make the system even better. 👍

@ChrisRosser

Жыл бұрын

I hope so too!

Great analysis, thanks Chris.

Thanks Chris. This is the type of information we need.

@ChrisRosser

Жыл бұрын

Thanks!

Bravo! Great testing as always, GJ Chris! Interesting to see if Walksnail can fix the latency, because the hardware should be able to be faster than DJI and if a theoretical 2nd version of the goggles will include 120 or 144hz displays. Cant wait to see how the HDZero goggles will perform.

Nice test, I just started in FPV, and I'm waiting to have more experience to switch to HD, I always think that HD-Zero has the potential to be the future HD. Thanks for sharing

This man is the professor of the FPV scene. Time to give him is PHD in FPV Tech!

You broke some mantras. Very well done!

Looking forward to Chris replying to some of the questions raised in the comments. Good healthy discussion happening here 👍🏻

Well done ! Nice video

How cool?! I love your test setup.

I already have the DJI system and a few analog. The thing that this testing doesn’t address is variable latency due to the quality of the signal. From what I have seen, The Walksnail system is more consistent but has a little less range than the DJI. The HDZero system is really consistent at even greater costs to the image quality. When the latency is that close the consistency matters more. It would be nice if at least one system eventually had modes to support more consistent vs better range or quality. I wouldn’t want to fly the 1080p mode all the time but it would certainly be worth flying for some flights.

@ThomasAndersonFPV

Жыл бұрын

DJI does... 25 mbit will carry further and with lower latency and more consistency. Quality obvious in 50 kbs. Also, you can choose latency or quality. But maybe a more mapped out option set ;)

@TechnicallyTom

Жыл бұрын

@@ThomasAndersonFPV At the best setting for latency, it still varies a bit. If they added a standard definition option, you could probably make that work with more consistent latency and it would still look better than analog. Making this an option, you would still be able to choose the higher quality options for most situations that do not require as low and as consistent latency.

I don't agree with your WDR assumptions. With the limited computational power, i think the latency is just a result of expensive full frame based tone mapping + maybe a more expensive sensor readout. Reasons against WDR = combining frames: a) Racing cameras are also marketed as WDR and they just rely on sensors with a higher dynamic range + tone mapping. The RCSchim reviews show clearly that they do not have a latency penality. b) combining two image does only work for (almost) still images.

@ichi1082

Жыл бұрын

It would be odd to have a analog camera do whats essentially deinterlacing with dual ISO readout, only to then convert the signal to analog again. But then again, thats almost how split cams work, and they are known to have a delay of more than 30ms.

That's a really great and detailed comparison! Good job and thanks you! I'm really impressive with HDZero, for sure it will be the best option for competitive racers, specially with the new googles to be released.

@ChrisRosser

Жыл бұрын

New goggles promise to be even faster. Can't wait to test them.

Thanks Chris! Nice to see an honest comparison with real test data. Blows most of the complaints about DJI right out of the water. I started with DJI, then built an analog quad with high end components and I hated it. I’ll be ditching the analog equipment I bought as fast as I can. Steeles setup was OK, but expensive and impossible to purchase anymore. Even Steele said HD zero was better than his analog setup. I trust engineers over the KZreadrs out to just make a buck.

@ChrisRosser

Жыл бұрын

Yeah I'm definitely a fan of just showing test data comparing the systems rather than flying it and giving a subjective opinion.

Can't wait to see the HD Zero results, great presentation! thanks

@KOrnhOliO1

Жыл бұрын

That's just what he showed us.

@northerntierfpv8947

Жыл бұрын

Yeah I was thinking of another video, all good

@ChrisRosser

Жыл бұрын

The new HDZero goggle promises to be faster than the external vrx. I'll test it if I can when it's available.

@northerntierfpv8947

Жыл бұрын

@@KOrnhOliO1 see Chris's reply. I am not nuts afterall, i was referring to the newer HDZERO. Cheers!

@northerntierfpv8947

Жыл бұрын

@@ChrisRosser yeah thats what i thought you said. Thanks for clarifying that. I just picked up the new VRX 4 interesting results so far.

Excellent and very well explained 👏 👍 👌 😀

Fantastic testing, Chris! 😃 Thanks a lot for all the analysis and explanations! Stay safe there with your family! 🖖😊

Knowing this I'll try my analogue camera with WDR off and see if I can notice the difference

Now this is usefull. WOW!!! I mean... WOW! I'm shocked from the results.

@ChrisRosser

Жыл бұрын

Glad it was helpful!

I enjoyed reviewing the test. For me I will continue with my DJI goggles. But if I was a new pilot getting into the hobby. I could see getting the Avatar system

@YotaGuy_Kyle

Жыл бұрын

@@LyroLife if I was a new pilot I’d go avatar, the lighter form factor goggle. In this test 720p 120fps at 100hz of the avatar beats out the DJI. Plus the OSD is better. But I’m not a new pilot, I’ll keep my 2 sets of DJi goggles and 6 vista units. I also love my Fatshark HDO2. For micros

@ChrisRosser

Жыл бұрын

I agree the form factor and screens in the avatar goggles are the biggest draw for me. I believe that image quality etc. is likely to approach DJI fairly quickly with firmware updates.

Surprising results all around. HDZero looking more and more an appealing proposition. As a longtime DJI user, I'm very pessimistic about the future of the hardware I already have. If DJI lock me out of their roadmap, I'll upgrade to HDZero.

@rcrdps8144

Жыл бұрын

I was pretty pessimistic too. But I read a lot of comments, and there have been several occasions that folks that claim to have sources say that the "Goggles 2" will be backwards compatible with Vistas, Link, and AirUnits, and that V2 goggles will be forwards compatible, though lacking features, with v3 air unit. I could care less about foward compatibility with v3 as I would only buy those air units if I had Goggles 2. But as long as they keep making Runcam Link( Vista with new vendor), and provide it with firmware that won't lock out v1 Goggles, then I will be happy,....and probably upgrade to Goggles 2/v3AirUnit eventually to get 1080p. If all of that is true, then DJI will be in a much better position to keep us from buying Avatar. I keep HDZero in the running too, but really, the only reasons I want HDZero is the little whoop board, and OSD/Canvas mode. I look forward to 1080p in Avatar and DJI, and so far HDZero hasn't really hinted at 1080p any time soon.

@ChrisRosser

Жыл бұрын

I also want the DJI camera to remain available. It's a step above the cameras from Runcam or Caddx.

@rcrdps8144

Жыл бұрын

@@ChrisRosser Yeah, I find the Nebula Pro's( micro and nano) adequate during fairly good lighting conditions, and I'm actually finding that I like the softness of the micro over the DJI,...but,.... when the lighting is tougher( bright blue sky through the trees), DJI shines. The Nebula Pro's get that hazy look.

Looking forward to the HDzero goggles which is announced with 4ms glass to glass.

@ChrisRosser

Жыл бұрын

Me too. I hope they'll lend me one to test.

Very cool and interesting. I've seen lots of people commenting on the "noticeably" lower latency of the avatar system in comparison to dji. Obviously a placebo effect. I think displaying a lower latency on the avatar goggles is just a marketing bluff.

@ChrisRosser

Жыл бұрын

You're right I don't think the latency of the systems is noticeably different. I think the latency reported by Avatar is accurate. It's just not glass to glass, it's glass to framebuffer! DJI are probably doing the same but in DJI's case framebuffer to screen is ~0ms.

did not expect these numbers, really cool. thanks to avatar for bringing back interleave, hahaha. good luck for me, dji not that bad and i also have no problem with my 60fps polar cams. i hope hdzero will continue, because i never will buy something from fatshark.

I used to test analog cameras for latency (of just the camera itself), and there was a big difference between them. The Foxeer Predator had a consistent near zero latency (around 1ms) for first pixel showing light, where a camera like the Caddx Ratel would have up to 30ms. CCD cameras were obviously tied to the frame-rate since it waited for the entire frame at once. Which Analog camera did you do these tests with? Great job. I appreciate the data.

@ChrisRosser

Жыл бұрын

I tested with an Turbowing Cyclops (consistenty 2 ms latency) a RunCam Eagle (consistently 18ms), Foxeer Falcor (consistently 18ms) and Runcam Nano 4 (consistently 2ms). The fast cameras had noticeably worse dynamic range than the slower cameras.

@RecursionLabs

Жыл бұрын

@@ChrisRosser Thanks. I appreciate the additional information. I'm pretty sure the Foxeer Predator had WDR enabled when I did the testing, but it's been a few years now.

@GRANOLAFPV

Жыл бұрын

@@ChrisRosser Given the large discrepancy between the analog cameras tested, what do the numbers shown for "Analog" in the video represent?

@ChrisRosser

Жыл бұрын

@@GRANOLAFPV Recently I've moved to showing the theoretical latency of analog in lieu of the specific latency of a particular camera. This is 2ms pixel to pixel and 19ms to full frame.

@GRANOLAFPV

Жыл бұрын

@@ChrisRosser Thanks for the reply. Does this mean you’re using the theoretical minimum latency for analog? In my case I’m using a Foxeer Predator V5, TBS Unify Nano, and RapidFIRE on FatShark HDOs so I’m wondering what the approximate latency of that analog setup is vs. HDZero in particular.

Hdzero is gonna get good!

The hdzero latency really shows the latency of the goggle hdmi input and screens. I would expect the walksnail module to be at least 12ms slower than the integrated avatar 720p60. So your looking at 41ms latency first bright pixel vs 2ms for analogue. For a full bright frame your potentially looking at 60ms with the external avatar module.

I’m curious to see if the new DJI goggles will do full frame or line by line like the Avatar goggles. Can Walksnail change the way the screens refresh or is this a hardware limitation? Is this the same as progressive scan vs interlaced?

@Vousie

Жыл бұрын

1) full frame - show entire frame at once. 2) progressive - draw the frame line by line 3) interlaced - draw every second line, then come back and draw every other second line

I had been wondering if the VRX component of the Avatar system is connected to the display component over something like an HDMI style link. It seems like the rolling refresh might suggest something like that. Like, WS developed basically a standalone VRX and then integrated it with a standard HDMI type display component in the goggles.

have you tested the googles 2 or integra latency with o3 end to end?

Can you make a test, where you measure/show jitter? Walksnail should have some jitter, because of 120fps video on 100Hz Displays. (This is probably also the reason for higher latency compared to DJI.)

Do lcos displays that are in skyzone 04l have a lower delay than lcd?

So analog rules ! YES!😉

I thought the hdzero v2 cam was the better one is that not the case should I get the foxeer

Does latency compound exponentially with any of the HD Digital systems?

👏👏👏👏

super interesting so basically the walksnail could shave approx 15ms off their latency just by updating the entire screen all at once like the dji? perhaps then they could actually send a less compressed image and add clarity and a higher quality image, by using some of that latency for data processing and transfer instead of using it to draw the frame from top to bottom

15:30 Lol. Amazing how real science just continues to show how much "feel" is such a b.s. measure. Pilots be like .... "Whaaatttt!!!"

I can't wait the battle between Avatar and Avata cause dji V2 is outdated soon. My bet is for dji V3 and then I will upgrade from goggles racing edition if they will dji drones too. If not I will not spend 1 cent in any new fpv product, only buy used air units and cameras for the dji goggles racing edition, the big red ones with touch control. Still works great with mavic 2 pro.

For flying in bandos and behind trees is walksnail good or dji?

Does that Latency increase as the distance between VTX and goggles increases?

Which DJI system is the video referring to? O2 or the older air unit.

Hey Chris, how has the latency changed over the past year? Hopefully they've made some improvements? Thanks

Really interesting insight into video systems here, especially the impact of WDR into latency. This really shows that consistant latency seems to be more important than the absolute value when peeps (including me!) claim, analog feels just so much more immersive despite the really shitty image. I'm gonna check those WDR settings ;D This also means a bit of tweaking and optimizing of Avatar and launching that pre-teased fixed latency mode might actually do a whole lot of the trick and could really be a strong and competitive feature, even if this ends up in a slightly higher absolute latency. This also might be a strong feature against DJI who likely won't do that for their more image quality focussed target audience. Still... now I'm damn curious how non-WDR feels against active-WDR... What camera did you use for this test? I could imagine this WDR-impact will not be the same on every camera. This might actually spread quite surprisingly between models even of the same manufacturer.

@BumbleBeeFPV

Жыл бұрын

I don't think that a constant latency is very important. If we just talk about a jitter of 3-5 ms it will not be perceivable. The human brain is limiting factor 😉

@aphinion

Жыл бұрын

@@BumbleBeeFPV I'm talking about DJI consitently dropping frames or dropping to 60-100ms the moment there is a tree in the way, not jitter.

@Vousie

Жыл бұрын

Constant is one thing, as Enno noted, we probably won't be able to notice 3-5 ms jitter. But that's not what DJI & Avatar do. They have eg 31 ms in optimal conditions, but drag out to over 100ms at times if you're out of range. Analog and HDZero stick to their latency - whether 29 ms or 35 ms and don't *triple* in parts of the flight. You're definitely gonna notice that.

@aphinion

Жыл бұрын

@@Vousie Indeed, that's exactly the issue. While Jitter free transmission (= real time = single clock synced) improves things (and trust me, it's noticable), the elephant in the room is the massive latency drop during frame recalls. I really wonder how constant latency mode in such a compression based system would look like. Lets see!

@Vousie

Жыл бұрын

@@aphinion Constant latency would essentially look like HDZero, but probably more "fuzzy" instead of "blocky". And it would have poor penetration like HDZero has. People keep complaining about HDZero's lower penetration but that's specifically because HDZero doesn't resend frames which is why it gets so much better latency. People want to have their cake & eat it... I'd love an option of something like DJI/Avatar but where it has an absolute max latency increase of say 20 ms and after that it just displays whatever it has & puts black squares wherever it doesn't have the data. Both of those systems currently seem to keep going until they get the image to show - even if they show a lower resolution version, they still want something, and that's the issue. Though I wonder whether it's a technology issue - whether they need the full frame in order to be able to display it... An alternative would be HDZero but just using the previous frame to fill in any holes from the current one - might be hard without a new chip, but it would greatly improve HDZero's "penetration" without adding any real latency.

If someone made drop-in replacement OLED screens for LCD Fatshark goggles, they'd make a pile of money.

Can you test Walksnail Goggles X? Would make a great video ❤

Digital systems that use compression should be tested with pictures and not just with a led turning all pixels to the same color and intensity because the compression is very efficient if all pixels have the same value. In real use cases the compression is far less efficient and introduces much more latency due to all pixels changing value all the time.

@timmolteno8021

Жыл бұрын

Totally agree with this. When compression introduces artifacts, do these modify latency (blocky images etc)? How about placing 16 blue LEDs in a grid over a monitor playing recorded gopro footage of FPV flying in a situation with rapidly changing detail, e.g bardwell video through trees :). Then you'd get some measure of latency in different regions of the screen. Eg DJI may update the central area more often under situations where compression is struggling to reproduce the rapidly changing video... Despite this criticism, I totally appreciate your content. Thanks.

@ChrisRosser

Жыл бұрын

I love that not only did you provide constructive criticism but you also suggested a sensible and feasible way to implement an improvement. That wouldn't necessarily allow a time to first pixel measurement but could be a step forward for time to first field/frame.

I fly HQ mode today. I'm only interested in Avatar for 1080p 50mbit mode. So your testing makes me even more reluctant to buy that system. Kinda hoping the DJI Goggles 2 does 1080p properly. I've always thought it funny how folks felt analog was lower latency when my favorite analog cams matched DJI in HQ mode. Variable latency,...ok,... but how often are you flying proximity at the ranges the latency gets bad.

@ChrisRosser

Жыл бұрын

This is a great point. For me the variable latency of DJI is only noticeable in situations with such poor signal that an analog link would no longer be usable. One of the huge benefits I found with DJI was the increase in signal penetration.

@rcrdps8144

Жыл бұрын

@@ChrisRosser Yep. I remember when first getting DJI, I felt I could "fly the whole neighborhood".

Can you please add DJI Avata with goggles 2 to this comparison?

It is tehnicaly possible that oled screen in this gogles is a "multiscan" thing Oleds generaly have much lower multiplex ratio than for example LCDs - you need more lines light up at same time or you will ned to push crazzy amount of pulse current througch ol;eds to get bright image - same thing happen in large format LED screens.

@ChrisRosser

Жыл бұрын

My understanding is you need to PWM oleds to reduce brightness without seeing colour drift due to the different responses of the coloured oleds.

I want to fly and you need a PhD to understand the latency details. Full frame counts - fake self reports are proofs of fooling customers. Lot of work you had done and success you had achieved . Thx a lot

So, at 16:19 in, doesn't that show that HDZero is just 2ms faster to draw a full frame than DJI? And that Avatar is right behind those two at only about 2 ms slower than DJI? It also seems to show that people saying that analog has significantly less latency than digital does, are actually incorrect, and it has more latency to draw a full frame. It appears from your chart, that analog WITHOUT WDR actually is about 6ms slower to draw that first full frame than DJI appears to be, correct? And, if a person is using the very popular WDR camera on their analog quad, they are actually getting far more latency than any of the digital systems. Seems to me like DJI is the only company truly rating their latency very, very close to the tested correct amount. Hmmm, Am I correctly reading your chart that you've clearly shown that the 3 digital systems have less latency to draw a full frame than the best analog non WDR camera's for analog? If so, the whole ballgame has changed with science! Thank you! :)

@flybyirwin

Жыл бұрын

Don’t forget that DJI and Avatar’s latency increase more when signal is poor. What analog pilots say is they would rather have the analog static and consistent latency than skip frames while DJI retransmits a badly received frame.

@KOrnhOliO1

Жыл бұрын

@@flybyirwin And, what digital pilots say is "Damn! That penetration of objects is amazing! I can fly completely behind big buildings with no loss of picture quality!" As the analog and HDZero pilots say "I can't even attempt to fly behind that building without complete loss of signal!" Yeah, analog and HDZero are great for racing, but there's a huge portion of FPV pilots who couldn't give a crap about racing through gates in an open field. I know many more would like to fly places analog and analog2 can never dream of ever flying, and with gorgeous image quality! 🤔👍

@npcfpv9542

Жыл бұрын

@@KOrnhOliO1 😆 he said, analog and 'analog2' ROFL

@KOrnhOliO1

Жыл бұрын

@@npcfpv9542 Well, if we all really think about what HDZero really gives us better than analog, it's: Clearer image quality - not as clear as FatSnail or DJI, but it's still clearer than analog. Same snow and fuzz break up as analog. That weird as Fu%K vertical Rainbow lines black out. So, when we really give it some thought, it's got better image clarity than analog, but the same snow and this Rainbow signal loss crap. To me, that's a better version of good ole' analog = Analog2 ... Or, we could call it: Anal log2 LOL Now, I digress.... 🤣🙊

@ChrisRosser

Жыл бұрын

I think the best comparison is time to full field. That puts Analog without frame combining WDR fastest. HDZero slightly ahead of DJI which is slightly ahead of Avatar.

Analog with WDR vs all Digital systems is my eyes and brain was lying to me?

HDZero it's the 1 most people have already got a set of goggles anyway it's by far the cheapest option as long as you're on your own set of goggles

If hdzero shaves off the display buffer then that 4ms to first pixel will be comparable to analog.

I finally found out why I feel strange watching your videos. Your camera is placed underneath eye-level. It might be a psychological thing but I have the feeling to be looked down on and stupid. You should try recording at (at least) eye-level. Maybe then I won‘t feel so „dumb“😅

@ChrisRosser

Жыл бұрын

Sorry you feel that way! I'll see what I can do for future videos, to be honest my whole camera setup needs a bit of an overhaul!

caterpillars dancing up and down.

Almost 60ms glass to full frame at 1080p60... that's like lagging 4 frames at 60fps....

@ChrisRosser

Жыл бұрын

Yes we can really see that h265 is not a codec optimised for latency! It's designed more for high compression rates. I expect the 1080p latency to improve as WS and Artosyn refine the encoder settings.

очень интересное видео. получается, низкая задержка аватаров это ложь??? получается, что даже аналог более медленный??? как же так...

I need a high speed camera.

@ChrisRosser

Жыл бұрын

You can pick up a Sony Xperia XZ that can do bursts at 960fps for $100 nowadays.

Analog for me, not by choice.

@ChrisRosser

Жыл бұрын

Nothing wrong with that. Still the most flexible, available and cost-effective solution. 👍

I'm not sure it's fair to compare two interlaced fields with a full frame.. The second field might complete a normal frame's resolution, but it is from a full time step into the future compared to the first field Edit: then again, frames also represent a time span, not a moment :)

@ChrisRosser

Жыл бұрын

Agree, that's why I prefer time to full field.

The worst part about you giving full statistics for first pixel, first field, and first frame is that now I have to decide what is most important? It's like there's no absolutes these days! 🤡 JK, I flipping love this science-- awesome hard facts.

@ChrisRosser

Жыл бұрын

Haha! Great response 😁 I think I lean towards full field rather than full frame. I don't think first bright pixel is particularly useful but is the only sensible comparison to the reported pipeline latency.

@CapnBry

Жыл бұрын

@@ChrisRosser I agree. I'm sure my brain can't tell when the first scan line has changed, but at first full field I'm pretty sure I can tell what is going on ("you're crashing, dummy"). I really appreciate you supplying all the data to let users choose instead of just declaring something as best, don't worry about why. Now tell me which ExpressLRS packet mode is the best!

I'm dusting off an analog system that I never put in the air. It will suite me just fine while HDZero & FatShark/WalkSnail HD systems hit v1 maturity, possibly holding out for their respective v2s.

Great info..... your data shows that HDzero is the clearly the best in this test and HDzero fans are still butt hurt I don't get it OMFG I don't know why KZreadrs even talk about HDzero because there is clearly no correct way to review, test or compare the system

What camera and what shutter speed did you record 960fps with? The strips could be a result of said factors as well as the rolling shutter of the camera used to capture the screens. I suspect this is a result of the OLED vs. LCD panel response times mixed with the rolling shutter... OLED panels have response times quicker than 1ms which is faster than you are capturing them in action.... and average LCDs are around 10ms....

@itsallsobothersome4537

Жыл бұрын

This has nothing to do with it, look at the previous videos.

@TheTomJon

Жыл бұрын

@@itsallsobothersome4537 I did, and he doesn't know what it is which I was I offered up this concept for him to investigate and think about.

@ChrisRosser

Жыл бұрын

The camera is a Sony Xperia XZ. A single back strip is common for OLEDs like the sky04x a zebra Stripe is uncommon but the camera is capturing it accurately. The frame rate is high enough to resolve it scanning down the screen. Black bars are not visible at lower frame rates on eithe oled panel as they blur out over the longer exposures. 👍

- 6:54 Scanlines only appear on CRT screens drawing interlaced video. - You are conflating PWM rate with refresh rate. These are two entirely different things. And calling it "60hz scanning" and "100hz scanning" sends shivers down the spine of any person even remotely familiar with modern display technology. - The image starting at 8:10 is just completely incorrect and I have no idea how you came to these conclusions. - Again, 17:20, your very own Skyzone Sky04x displays fields as frames at half resolution with a one pixel shift up and down between fields. This argument makes no sense because using this particular type of deinterlacing, your image is ALWAYS going to have half vertical resolution and fields are always displayed as frames.

@jimjones3639

Жыл бұрын

You seem to be an expert you should make your own video and show chris how its done. Im sure it would be entertaining

@ChrisRosser

Жыл бұрын

I agree, I'd be interested to see you make a video explaining the topic. There's too much to unpick in just a comment reply.

This have no sense to be honest, you are testing analog video as digital,.which have not many sense. I think you are mixing video information, with signal digital processing. I recognize I didn't watch the full video but the conclusion cannot be right. Why you measure time for full frame , time show the two scans? Or you say it works 30fps? That is ridiculous, you can cannot measure a analog video as a digital video. 60hz are 60fps, no 30fps, you receive 60fps, with half info and your brain mix the info for full frame. The info that you lose, is not less in any case than the info that you lose in areas with digital system. Your are mixing terms and technologies. I can measure with oscilloscope 2-6ms of latency in my cameras and 14ms end to end. Not sure what are you doing here. For any pilot , and in practice, wont have many sense your results either, except the HD zero one that will pair with decent quality analog cameras, as phonenix v2 but still won't catch others low latency as Predator v5, CAT2 etc Quite miss leading this video

@ChrisRosser

Жыл бұрын

If you're interested in this topic I've done a couple of other videos which cover the points you've raised in more detail: Analog video, interlacing, and frames vs fields: kzread.info/dash/bejne/pa5-sbeRmdHLido.html DJI behaviour with partial bright frames and analog vs digital latency testing: kzread.info/dash/bejne/a4d2xZVug8bJdMo.html

@Flamenawer

Жыл бұрын

@@ChrisRosser yes I watched them, I really think you treat the video as static images, in the moment you put a brain there to joint the things, your way to see the things is not valid nevermore. In analog your brain make the part that in digital make a processor. I really think your way of treat this topic is misleading, in example how you try to pair in latency analog with digital by the need to wait for two scans to get the full image, that is not how a brain process video, is how a brain process a magazine. You cannot double the latency of analogue by your way to measure the things. The analog latency, is what it is, with half of info in everyframe, the image processor will scale it and double the lines to fit the screen, and can be faster or slower and that is not analog video topic,

@ChrisRosser

Жыл бұрын

@@Flamenawer I tried to be very clear about this point in the video. I think the best comparsion between analog and digital is time to first full field not time to first full frame for exactly the reasons you suggest. Time to first full frame and time to first bright pixel is provided for completeness.

Walksnail,fatshark,or DJI are overpriced massively, definitely NOT buying and I would advise others not to buy unless you love wasting money or your one of many boring KZreadrs who get given the stuff free