iPhone LiDAR accurate drift over 200 feet

[@trollenz]

Wow, you really did put a lot of efforts in this test ! Congratulations, state of the art ! The results are actually not that bad for such a distance. I share your conclusion. Thanks.

[@Megaaleh]

iPhone: US$ 1.100,00 Professional Equipment: US$ 50.000,00 - US$ 200.000,00. The iPhone has a 0,5% margin of error + rendered texture and colored the surface accordingly. That is IMPRESSIVE. Conclusion: The iPhone LiDAR is not meant to be used in SOME professional setups. But it’s superb for everyday use for 99,99% of people and SOME professional uses: When the 0,5% margin will not like kill someone, for example. Great job Apple. Really really impressive. 👏🏻 And great video. You did an awesome job. Very professional, very informative. Just all around amazing. 👏🏻😁

[@liamcarper6517]

nice! yes, the iphone and the integration of measurement tech is absolutely amazing. in 5-10 years, it will be mind-blowing to look back and see where we've come kzbin.info/www/bejne/j3fRZWt6ir-oldU my gig is supporting these products with Trimble....

[@1371greer]

I've just been getting into CAD/RevIt for the pipe industry. Knowing that an iPhone Lidar scan can be put into Recap is pretty cool. Great video not only for the content but how it was edited and broken down.

[@RamiTamimi]

Thank you, glad you liked it!

[@geogreg1750]

Rami…… great videos and explanations on everything. I learned my surveying at Ft Belvoir, US Army, in the late ‘60s. Did sunshots for locations ‘in the wild’, surveyed construction projects, ie roads, bridges, landing zones… in Vietnam.. artillery firing in Texas… and later summer jobs while going to college. 30+ years federal geologist/ geographer and GIS specialist. Anyway this all boils down to being born 40 years too early and hoping to influence all you younger folks to seriously look at surveying and GIS as a field to go into. And hopefully getting an instructor as good as Rami here!! You are perfect for this field of instruction.

[@zicronix]

Great video I would love if you could add an overlay in the video for us that uses metric. And make an excel sheet in metric that you could just show real quick so we could pause the video to take a look.

[@RamiTamimi]

Noted! Man, it seems like the whole world uses this "metric" except us Americans.... (That's a joke btw)

[@andrewgunawan4502]

THANKS FOR SAYING THIS, although I realise that another way of interpreting the conclusion is that the iPhone’s LIDAR sensor has a percentage uncertainty of about 0.5%

[@revolverich]

@@RamiTamimi No I think its the truth! ;-) Really though in the UK we are stuck between both, mostly metric but some things still just stuck in the past! It makes so much more sense and easier to make calculations. Good vids btw.

[@abunchofnothing8603]

For just an average person wanting to get an idea of the measurements. That pretty good.

[@CivilSurvey]

It's a hard work to prepare & making this type of video bro 👍 keep up this type of video .Thanks a lot to spreading knowledge.

[@WTCHME]

Excellent video, I have minimal interest in this but it got recommended to me and I watched this one and your last. Very interesting take on things, good information, and I appreciate the time it took you to break down the way you’ve done things and why it matters.

[@RamiTamimi]

Thanks for watching. Welcome to the world of Geodetic Surveying!

[@WickedOnMars]

Could you let me know what app you used on the iPhone for capture?

[@curtisoliver1006]

A 0.5% error on the iPhone is actually very impressive. For most that is accurate enough for their needs. For those who like metric, that's 0.005m/m or 0.5m/100m

[@soarersan1]

Hi Rami , I've been studying your iphone Lidar videos and I've decided to trial the Iphone as a method of detail surveying on residential house lots. Typically , here in sub tropical Brisbane ( Queensland , Australia) , I am dealing with lots of vegetation , trees , scrubs etc and slopes of 2 to 4 metres across a 25m x 40 m site. My method will be create controlled ground targets in good locations for GNSS pick up with my GPS and then Lidar scan the yard up to the fence line ., including the house , any sheds , tree trunks etc. Then I'll use Poly Cam on the iphone & Cloud Compare to register the point cloud , then import it into Autocad Civil 3D . Do you have any recommendations on the processing software or method ? I'll be using an iphone 13 with a gimble for stability. Looking forward to more research on scanning technology , as it impacts on surveying . Have you looked at generating ground contours from the Lidar scan ?

[@Bigen-11]

The phone has to keep track of where it is, as it’s performing LIDAR. So as you are walking taking shots with it, it must be using GPS to track where the phone is, thus also the locations in the LIDAR surface are dependent on the accuracy of the GPS??

[@merseyviking]

It's almost certainly not using GPS, instead it'll use a combination of sensor fusion (combining the accelerometer, gyroscope, and compass data to perform inertial tracking) coupled with SLAM - Simultaneous Location and Mapping where it uses the data gathered in the previous few frames to align the data captured in the current frame. This is the reason for the fairly linear error in the northing (the long axis), because inertial tracking will always drift over time. GPS can only be relied on to within 10 metres or so, and can vary wildly over only a few readings.

[@ishazimi2770]

It doesn't use GPS AT ALL. I tested this.

[@RamiTamimi]

As John and Ish said, it's not using GPS, it is using internal gyroscopes to measure the movements of the phone. If I was using GPS, that would be considered Absolute Accuracy, not Relative Accuracy.

[@Apprehence]

This particular app isn't using GPS at all to create this model. It is however possible to extract EXIF coordinates from the photos it takes to texture the model. Those coordinates are really inaccurate though.

[@sergemarlon]

@@RamiTamimi How would using a phone stabilizer like the DJI Osmo effect the results?

[@tangodman]

Apple should sponsor you. They themselves never talk about LIDAR's capability. And here you are demonstrating it and helping the people. Top-notch content.

[@FedericoTesta1]

It's more accurate than i thought! Impressive

[@RamiTamimi]

I know, the more we know hahaha

[@Hamhandful]

Professional video, really well explained and kept my interest to the end. Bravo!

[@Valter19819]

love this content!

[@Bigen-11]

Thanks for the Info, now I’m thinking about upgrading phones

[@RamiTamimi]

Any time!

[@AhmedOta]

بارك الله فيك يا رامي

[@sjmoir]

Great video, thank you. I’m not a surveyor- but I’ve been experimenting with an iPad Pro and point clouds / scans for architectural visualisation etc. I understand why you skipped the scale step, but I wondered how the data would compare if you had scaled. The data sets could be useful if used together perhaps? Not that I have a total station, or need one! I have been using 3D Scanner App and Polycam, and also SiteScape, which only records point clouds but (without using your rigorous testing methods) seems more accurate somehow. Subscribed 👍

[@R0LLM0P]

Interesting Video. What was the app you used for the iPhone scan?

[@yuimilungleng]

This is so impressive👍🏻 really appreciate the time U put to make this video

[@RamiTamimi]

My pleasure 😊

[@andrewpndavies]

Superb video - I may have missed it, but which app did you use on the iPhone?

[@florianberger8839]

Hi Rami, great video! Btw. what's the iPhone app you're using for lidar scanning?

[@OnRansom]

Dido, I have the same question

[@michaelm2149]

It’s called 3D scanner app

[@Hooveyx]

used to be a surveyor but dude… that would be CRAZY for a Topo survey. To see the change of the land overtime if they are gonna do work with it such as a strip mall or something, that would be perfect for slopes and tiny tiny slopes ACCURACY IS KEY

[@RamiTamimi]

Yes, and I think this technology is very relevant in surveying, if we introduce some control, I'm pretty sure it would be very useful.

[@EmilioBaldi]

Why not use the fence posts?

[@jeremyhale8132]

Would love to see the error after scaling it to control.

[@RamiTamimi]

Great video idea!

[@yadeshiergete74]

Thanks for sharing

[@marknielsen9315]

How would the accuracy improve with a gimbal for that distance like you did in a more recent video.

[@supermanny73]

Great job! Thank you for putting this together

[@RamiTamimi]

You are very welcome!

[@randallthomas5207]

How am I going to see the bubble? IPhone in rear facing camera, held above the level so you can see it from below.

[@keithwhisman]

Wow the iPhone LiDAR just looks so much more convenient especially for combat engineers on the battlefield.

[@RamiTamimi]

Hahaha, well convenience sometimes comes at the cost of accuracy. But for $1000, its not bad.

[@woolfel]

excellent attention to detail. Pretty impressive how accurate the iPhone12 LiDar is even though it wasn't designed for this type of use case.

[@paulliddle6079]

Sorry if it’s a dumb question, but why do you measure from the top of the stake?

[@leicageosystemspublicsafet7253]

Well done.

[@aliday9968]

Sorry, I missed the name of iPhone app. What is it ?

[@prospero768]

See, just last year I was trying to use photogrammetry to get a topographic map of my property (because I'm cheap and it didn't need to be accurate). It didn't work well because I didn't have a good feature point detector and lidars were all kind of expensive. But this seems promising, and I'm pretty sure the drift can be fixed to an acceptable level if I sprinkle some registration marks all over the place for the phone to pick up and then locate them with a transit and tape.

[@liamcarper6517]

yup

[@johnthompson7448]

with the near consistent scale deviation could it be more of a calibration issue or offset for the lidar?

[@KB3M]

Looking to use the iPhone 12 Pro to do a property survey for a building permit. So 1/2' over 100' is not great. I assume the iPhone is using its internal accelerometer and gyro to determine the location of the phone while scanning. I'd be interested in whether the speed and smoothness of motion while scanning affects accuracy.

[@dobrinyanicitich7514]

good job, thank

[@guccioFTW]

very interesting video! which app did you used in the iphone for the lidar scanner? thanks in advance!

[@CarlosPerez-go8tw]

Good test..!!

[@cybernetix86]

what app do you use on the iPhone?

[@shableep]

Awesome content! Thanks for doing this. So, I’m no surveyor so please correct me if I’m wrong. But isn’t this margin of error reasonable compared to the ability to place the center of the stake at each 10 foot increment, and then the additional error of trying to find the center of the stake in the point cloud? It just seems like a more clean testing environment would account for the relatively low range of error here. Each stake should have to be leveled to account for error there as well. Like, it seems clear it’s losing some accuracy over distance still. But now that this is proving to be possibly useful in the field, it would be good to find a way to get extra accuracy in the testing.

[@RamiTamimi]

You could try to minimize the environmental errors, but honestly, must of the error we see is systematic, meaning even if we adjusted the environment, it wouldn't change the results too much.

[@dchall8]

I would not put too much confidence in these numbers. After he creates 100 strips of data on the same path and averages the data, then you can start to draw some conclusions. To take it a step further I would try to eliminate possible errors from hand-holding the iPhone while taking the measurements.

[@gelisob]

what about the accuracy after using a cheapo laser rangefinder-measurer and inserting distance correction between first and last point? You have the pointcloud, if u draw direct line between first and last and enter defined distance, would it scale everything to quite acceptable level?

[@liamcarper6517]

D>R>I>F>T>

[@RamiTamimi]

Yes indeed.

[@ElPinchiWero0]

Hola, Saludos desde mexico!!! what application did you use in the iphone? i have a iphone 12 pro, and I do surveys for gardening. thx for the videos!

[@fakhrulislamchoudhury3070]

Sir, is it true or grid distance i.e. 200.05 ft.

[@Elfin4]

Could you create something like this but indoors to measure ceiling soffit defections at the mid span, perhaps using an inverted staff? Relative measurements up to 6m spans ..

[@TheJoaolyraaraujo]

Thank you

[@dolphinpops]

Well done for the experiment. An easier way to measure is to scan along the track of the playground and see if they meet each other.

[@RamiTamimi]

The reason I did the high stakes was because I haven't been able to find the intensity output on the iPhone, so I just rely on elevation change to find the points.

[@vgoj]

Do you think It's possible to scan an entire bridge with the iPhone? Also, which app are you using ? Thanks in advance!

[@dalekkiller]

Could part of the errors be the result of how the iPhone is held?

[@rootmeansquared]

What “app” are you using it measure the LiDAR?

[@sakethkumar7458]

It's in built to iPhones

[@RamiTamimi]

3D Scanner App

[@roymwale1967]

Is this a specific LiDAR app?

[@sailorsailing9195]

What about full scan a building? Do you think its possible with the same method? Maybe 3d align with total station in the end?

[@liamcarper6517]

scary

[@LucasTabachnick]

What LiDAR app are you using?

[@c0rvu5albu55]

4:00 dont you have something like a "lattenrichter" (dont know the english term). In most cases a piece of angled metal with a bubble so you can hold it agianst your pole. Also why use the crappy imperial if you could use metric?

[@adamrak5245]

why you align point cloud by points from two different sides instead using one side ? You be able to compare how the cloud behaves as it moves away from the "solid" points

[@Arqbrunao]

Amazing video, I’ve been watching your videos but I got a doubt , how much storage do I need to use a iPhone 12 Pro as part of my workflow?

[@mariyoh5003]

as much as the bucket is a good laugh, you need a mini prism haha

[@danieljatnika4338]

Mini prism don’t have RC equipment for radio controller

[@frankaugustine9832]

Do you have any videos comparing gnss or total station/robotics to drones with lidar?

[@LiDAR3D]

Thank you Rami for this fantastic content! We're a company that has specialized in converting apple lidar scans of building interiors into CAD 3D and for that use case the apple hardware is really great. Keep up the good work!

[@RamiTamimi]

Thank you for your support!

[@liamcarper6517]

to what level/accuracy/precision are y'all relying on?

[@pippensculps]

Heh, the bucket method works but what I usually do is just hold my phone screen over the bubble (with my phone locked) and use the reflection. Love your videos!

[@amstelmax]

This was interesting!

[@michaelm2149]

What app is this?

[@carliones1]

Parabéns pelo vídeo. Muito interessante o assunto. Saudações do Brasil

[@RamiTamimi]

Obrigado!

[@diegoalonsovillarroelchave2851]

amazing video. what app do you use on the iphone?

[@RamiTamimi]

3d Scanner App

[@dchall8]

I love what you're trying to do here, but something is not right. It could the the iPhone sensor is what's wrong. What is the difference between the iPhone point cloud and a point cloud taken from an airplane at 3,000 feet AGL? Sorry I don't have the sensor details, but with the airplane data I can see the skeleton of the steel power poles along with the wires running between with very little error...or so it seems. With the iPhone the poles in the ground look like they were covered with a sheet and lack all levels of detail. With that much signal imprecision, how can you assume the highest point represents the top surface of the posts?

[@ronelgantu7002]

Puedo hacer un levantamiento topográfico. Con el scaner lidar de iphone 13 pro max. Y que tan presiso sale el levantamiento topográfico

[@rls151]

Can you do this with the iphone 13?

[@RamiTamimi]

Yes I just made a video about it. Check it out: kzbin.info/www/bejne/iGrPeJxqrLKXh7M

[@hiteshsinghx]

Which app are you using in iphone for scanning?

[@RamiTamimi]

3D App Scanner

[@danvaldivia6155]

What app are you using on the iPhone?

[@RamiTamimi]

3D Scanner App

[@Zeval31]

very enlightening. what application do you use on the iPhone 12 Pro to use the LiDAR scanner ?

[@RamiTamimi]

Glad you liked it. 3D Scanner App

[@martola9855]

Again, the problem that Ican observe is the distance of the Iphone to what we want to scan...... How far can I go from the object without damage the precision/accuracy ?

[@RamiTamimi]

Depends on your tolerance. If you are okay with 0.5 feet of error, you can go 100 feet. If you can only afford 0.1 feet of error, you can go about 20 feet.

[@mccfokl]

i was expecting that you were going to keep standing at point 0+00 while scanning each point until 2+00

[@jasperlawrence5361]

thx

[@quantuminfinity4260]

I’d be interested in a comparison to the default “Measure” app with the iPhone

[@EthanSeville]

I think that also use the LiDar sensor to assist it

[@quantuminfinity4260]

@@EthanSeville It does, but it would be interesting because its much much faster to measure simple distances or areas. Just tap to place two points and there is your distance in literally a second.

[@jakewillsman5129]

What’s the app called? The LiDAR one he used in the video

[@quantuminfinity4260]

@@jakewillsman5129 It's called "Measure", it comes with iOS since iOS 12, but can now use the lidar sensor to improve the measurement accuracy and speed

[@jakewillsman5129]

@@quantuminfinity4260 Hey nah sorry the mapping one. The app that you used to LiDAR map the objects.

[@ackerman8405]

Can you do a video on the north backsight process ..

[@RamiTamimi]

Can you elaborate more on what you're looking for?

[@TersusGNSS]

Rami, wow, that's what we are trying to do, great job.

[@RamiTamimi]

Thank you!

[@troelsnielsen2848]

you should plot the error on a graph. but i think it would be equally intereseting to learn the range and accuracy og the lidar sensor when the phones is in a fixed position scanning the distance to objects of different size and ranges

[@Apprehence]

Range is 5 meters max on this sensor, so stationary scans are accurate but very range limited. I've done a small research paper on it (in Danish), and it basically shows the same results as this video (likely due to scaling issues as a product of inaccuracies in accelerometer/IMU software).

[@tpolerex7282]

Looking forward to upgrading my 6 (I know haha) to a 12s/13Pro for my architecture work, looks like it will be plenty accurate for my uses except for maybe some very tight millwork tolerances.

[@RamiTamimi]

Believe me I have a Galaxy S8. This is my wife's phone haha. I too will be getting the 12S/13

[@Q50Diaries]

Could accuracy potentially be increased by utilizing something like a gimbal? Love your videos!

[@Ironbodybuildingxxl]

What software are u using in your allegro to remote control the sokkia srx? Im using survey pro

[@RamiTamimi]

I use Carlson, but I also use MagnetField.

[@ericromero3444]

Excellent video, congratulations, which application is used to perform the scan on the iPhone? It would be that you put subtitles in Spanish in more of your videos.

[@RamiTamimi]

Thank you, 3D App Scanner. I will work on adding more subtitles in other videos.

[@farley9255]

Great video. Would this work with getting correct measurements for a swimming pool liner?

[@scanningcars6500]

That seems really impressive, especially over grass which would seem challenging for the phone to do tracking. Would be interesting how the results changed it you did a zigzag pattern.

[@EreNat89]

Hello Rami! Great video, appreciate your effort in making this content! I'm just interested what was your scale factor during measurements with TS.

[@j0van87]

Great video 👍 I just have problems with imperial units. I'm used to metric system

[@RamiTamimi]

Thanks 👍, yea I know, the whole world uses metric except America.

[@j0van87]

@@RamiTamimi for future videos maybe you can put the results in meters as well

[@TheMitts]

Only recommendation would be adding conditional formatting colour to the results (red/amber/green) via the menu option.

[@RobertSmith-wj7zf]

The title suggested a “height” of 200 ft. I was surprised to see you do a path distance instead. Have you done a height/lateral distance accuracy test for the iPhone LiDAR? Like sweeping a 200 ft radius circle from about chest height. Position different objects of varying sizes, shapes and heights at different distances from the center of the sweep. That would be very interesting.

[@juliaswati5242]

Thanks for sharing. I'm not a surveyor, however I suspect that the results are highly dependent on the methodology. I suppose it's representative of how people would use it to generate point cloud data. The primary issue here is that of error accumulation as you walk (integration of accelerometer data, and mesh density which also depends on elevation angle). Can you try with a pano scan, while sitting near the middle point (at 100')? And, take the scan with the phone close to the ground so you get a side view of the stakes. Use optical zoom feature to enhance resolution. Am very curious to see if this improves the results dramatically.

[@c0rvu5albu55]

that error could be elimintaed if you use a gimble

[@liamcarper6517]

it's not a long range Lidar. that's why it's only $1000

[@Mr_Yeah]

Interesting video. I have no idea about surveying, but it's nice to know how precise the iPhone is for personal measurements. One video idea: How precise is the measurement tool integrated into the iPhone?

[@eliakimdelacruzrodriguez6099]

CON EL PHONE 12 NORMAL NO SE PUEDE???

[@LiDARs]

I would be interested to see how well it would do with a semicircle. Going in a straight line puts the iPhone at its best. Having it track while turning and using more of its IMU would be interesting. Great video tho. Presented well!

[@RamiTamimi]

Oh that's a fantastic idea, I should look into that, like a closed loop traverse!

[@Apprehence]

​@@RamiTamimi This particular app might have issues with closed loops too. At my last surveying exam i experimented with scanning different objects with photogrammetry and the iPhone LiDAR, and verifying those with standard measurements. I chose multiple 20ft and 40ft shipping containers, scanned them by walking clockwise around the containers, and sadly every attempt netted some issues at the 4th "overlapping" corner. The distances seem to steadily decrease over time, creating a "squished" model. For smaller details like the curb, however, it seems really accurate.

[@lorycalvia6171]

@@Apprehence what is the name of the app?

[@Apprehence]

@@lorycalvia6171 3D Scanner App. I think they've done some updates on it since, so it might behave better. but the LiDAR sensor still has a max range of 5m.

[@BloodHawk31]

Excellent vid. Now that we know the iPhone error we can use an error of 0.5ft/100ft with readings and get a fairly accurate reading for a quick check, but like you said, can't rely on a phone to get exact points, BUT for the time it takes towards the accuracy level, not bad, very surprised. I have a CAT S61, the laser measure is sometimes out by 0.3m/2m after calibration, and this is inconsistent, so it can't be reckoned in, so it's useless.

[@tadamb1]

This is really cool Rami. Thanks for doing this. I'm in the excavation industry and imagining a situation where we could scan dirt stockpiles to get an approximate volume. You think maybe you could do a demo like that in the future?

[@liamcarper6517]

how approx y'all want? you have GPS/GNSS? or a scanner?

[@tadamb1]

@@liamcarper6517 We do have GNSS stuff, but that is a lot of work. I'm picturing a scenario where one of our superintendents could whip out is phone, do a quick scan, send the point file to the office, and know how many trucks to order for haul-off within a few minutes. I actually accomplished this (since posting this question) on a very small pile near our office. I got it to work and return a very believable volume result, but the problem I had was that it produced WAY too many points. I'm talking 31,000 points for a 6 CY stockpile. I need a point every couple of feet. Not several per inch. I think I might try a different app.

[@Tydud3]

Would the measurement be even more detailed/accurate if you swept slower over the 200 feet, giving it more time to scan with the sensor? EDIT: Removed first bit of comment as i realized .500 off would be half a foot as stated in the vid, so six inches.

[@westcoastaerialimagery4690]

I hate to be the bearer of bad news but I just scanned ~30m of wall with the iphone 13 pro, total station pinged it at 30.275m, lidar data in cloudcompare has it as 29.241m. On hard stand surfaces it's around +/-50mm but on grass and dirt it was ~200mm too high More testing to come but still has a way to go Scanned using scaniverse app

[@liamcarper6517]

u gotta manage the drift. without a proper IMU, a mobile LiDAR needs the drift managed.... the iPhone is crazy accurate, but it's not setup to manage the D>R>I>F>T

[@westcoastaerialimagery4690]

@@liamcarper6517 I mean, the wall I measured was half the length of what OP measured so even less time for drift and yet here we are, terrible linear accuracy, surface accuracy is on par but overall pretty average I can’t even scan pelican cases without them looking whack, the imu in a phone is pretty decent but we’re just not there yet for the “live” processing of the data within the apps

[@monatamimi8125]

Great 👍

[@eblalzakzok1094]

Can your iPhone Lidar be used. On UAV? Can you survey a facade of a building from a tens of members and compare with traditional TS or laser scanner survey? You next experiment if possible.

[@trollenz]

With a UAV you're better off using photogrammetry. Take good measurements with a TS (both registration and controls) and you'll get much better results.

[@liamcarper6517]

good luck. that would really show how "off" the iphone is....