InfinityMR Universal high performance gimbal for GH4/A7S size cameras

[jakuban]

@tombrown1 I would appreciate if you could post all those thousand videos. Could be here so everyone could see them. Of course tests from flying in such wind + at least GH4 size camera.
I would say this: I think krall result is really impressing but probably it is possible to achieve with one camera and lens on 2500$ dollars zenmuse.
When it comes to horizon - as far as I know all gimbal electronics manufacturers have the same problem (more or less). Probably we need to wait still. I've been talking with Aleksey few days ago, I know he is working on that too.

 

[tombrown1]

Jakub, I think you misunderstand me. I've seen a thousand "gimbal testing" videos over the years. And the only ones that I've seen perfect horizon on are Zen. It was your quote of "horizon is perfect" that led me to post. The horizon isn't perfect. In fact it's way off at 2:52, and a little bit off in several other places.

I'm not saying this is your gimbal's fault - in fact I think your gimbals are excellent, that's why I just purchased an MR gimbal. The fault, as you well know lies with the gimbal controllers.

My point is that saying "horizon is perfect" when it clearly isn't, is misleading. I guess I'm being pedantic, but if you said "horizon is very good", or "horizon is almost perfect", then I would certainly agree.

Not trying to start a fight here - I'm one of your biggest fans and I love your passion for gimbals. But let's be honest about these things.


Best,

Tom

 

[crayfellow]

@tombrown1 I can't imagine DJI has some sort of magic proprietary secret sauce in the algorithms used by whoever developed the controller they are using. As an engineer I'd certainly not minimize the complexity of these control systems but the essentials of all these controllers and their associated firmware is going to be practically equivalent. Certainly "features" differ, but I strongly suspect any difference will come down to tuning.

Lots of excellent $xxxx+ gimbals used by pros every day use alexmos or some derivative, they just don't announce it or publicly associate. Sort of like how Linux may well be running on your washing machine. It doesn't really matter if it does the job, and that's one of the points of open source.

I didn't take your comments as adversarial, and I don't think Jakub did either. I agree it comes down to the controller, but my contribution is that any differences are better attributed to tuning than the capabilities of a given piece of electronic hardware with a microcontroller and some motor steppers.

TL/DR: as InfinityMR owners let's demonstrate this gimbal can do through careful tuning, it's certainly got the balance and stiffness covered.

I feel like most of the noise and hype surrounding gimbal-of-the-week is just clever marketing and a smokescreen to distract you from the fact that there can't possibly be any significant difference in the electronics (unless you're talking 32-bit vs. 8-bit or something, which we're not).

 

[Motopreserve]

@tombrown1 I can't imagine DJI has some sort of magic proprietary secret sauce in the algorithms used by whoever developed the controller they are using. As an engineer I'd certainly not minimize the complexity of these control systems but the essentials of all these controllers and their associated firmware is going to be practically equivalent. Certainly "features" differ, but I strongly suspect any difference will come down to tuning.

Lots of excellent $xxxx+ gimbals used by pros every day use alexmos or some derivative, they just don't announce it or publicly associate. Sort of like how Linux may well be running on your washing machine. It doesn't really matter if it does the job, and that's one of the points of open source.

I didn't take your comments as adversarial, and I don't think Jakub did either. I agree it comes down to the controller, but my contribution is that any differences are better attributed to tuning than the capabilities of a given piece of electronic hardware with a microcontroller and some motor steppers.

TL/DR: as InfinityMR owners let's demonstrate this gimbal can do through careful tuning, it's certainly got the balance and stiffness covered.

I feel like most of the noise and hype surrounding gimbal-of-the-week is just clever marketing and a smokescreen to distract you from the fact that there can't possibly be any significant difference in the electronics (unless you're talking 32-bit vs. 8-bit or something, which we're not).

I think the major difference is the inclusion of GPS in the Zen and Movi gimbals. This apparently helps to correct for horizon issues, which are inherent in these machines.

 

[crayfellow]

I think the major difference is the inclusion of GPS in the Zen and Movi gimbals. This apparently helps to correct for horizon issues, which are inherent in these machines.

I have to admit I have not yet worked on camera stabilizer hardware/firmware, but I have been around commercial avionics and the GPS for quite awhile. It's not clear to me how having a lat/lon/alt from GPS would help level horizon. I can see it potentially playing into the formula to a moderate degree, since you obviously can use GPS in fusion with accelerometers and magnetometers to determine a good guess at absolute position, but GPS is no magic bullet.

The 'obvious' key to level horizon is a persistent handle on the vector representing acceleration of gravity, but algorithms that also account for movement on other axes and blend input from multiple sensors will have inherent design compromises. In some ways including GPS only complicates things further, since it has drift and multipath challenges of its own.

Bottom line, nothing is perfect even if you spend 6 figures on camera stabilization gear. Meanwhile the pros are out working and just deal with it the best they can

Seems like folks get awfully carried away on technicalities when what matters most is a thoughtful/quality build, solid design, and straightforward integration. There are a handful of gimbals that offer that, and I believe InfinityMR is one of them. For whatever reason, I see units getting hyped far more with nowhere near the demonstrated capabilities even based on the handful of examples we've seen thus far.

 

[econfly]

The 'obvious' key to level horizon is a persistent handle on the vector representing acceleration of gravity

It's a tricky problem. That's for sure.

we [...] assume the complete physical equivalence of a gravitational field and a corresponding acceleration of the reference system.

 

[crayfellow]

indeed! hence I think we ought to focus on the meaningful bits and divide by the common denominator. Any modern controller can give excellent results, build quality and design are the differentiating factors. Especially for this one since you can use your choice of controller.

I'm still not ready unfortunately due to work schedule, but does anyone else have some fun sample material to share?

 

[Motopreserve]

It seems to me, without fully diving into the principals, that the GPS could be just one more piece of the puzzle that can lend a hand. True, it may complicate things. But we obviously use it to our advantage for the flight controller - so if the algorithm is written well, tapping into the reading the FC is receiving, and adding that to what a gimbal controller is sensing - may just be an advantage.

I agree with you in terms of the mechanics of the gimbal. Bottom line, controllers like the CP have, if nothing else, pointed out flaws in gimbals that were previously thought sound. The need for solid mechanics is a given.

While the pros that you speak of may be using the AM boards, countless stories show that they have A) spent MANY hours tuning or B) broken down and bought a very high dollar gimbal.

The most encouraging new development (to me) on this Infinity gimbal (aside from the apparent/reported rock solid construction) is the addition of encoders (which interestingly, the high dollar gimbal have had for quite some time). What these may do (or not) for the gimbal will be interesting - and hopefully help push the technology forward.

 

[SleepyC]

It seems to me, without fully diving into the principals, that the GPS could be just one more piece of the puzzle that can lend a hand. True, it may complicate things. But we obviously use it to our advantage for the flight controller - so if the algorithm is written well, tapping into the reading the FC is receiving, and adding that to what a gimbal controller is sensing - may just be an advantage.

I agree with you in terms of the mechanics of the gimbal. Bottom line, controllers like the CP have, if nothing else, pointed out flaws in gimbals that were previously thought sound. The need for solid mechanics is a given.

While the pros that you speak of may be using the AM boards, countless stories show that they have A) spent MANY hours tuning or B) broken down and bought a very high dollar gimbal.

The most encouraging new development (to me) on this Infinity gimbal (aside from the apparent/reported rock solid construction) is the addition of encoders (which interestingly, the high dollar gimbal have had for quite some time). What these may do (or not) for the gimbal will be interesting - and hopefully help push the technology forward.

The test for BMG's is the ability to perform on the set with lens changes and even adverse weather.

So far it's..

1) Movi (M5,M10,15)
2) Gremsy H6, H14, H16 H3?
3) AM - any gimbal properly tuned...
4) CP - if you have perfect conditions.

 

[SamurAchzar]

Not sure what's it about "perfect conditions" with the CP. Care to explain?

 

[SamurAchzar]

@tombrown1 I can't imagine DJI has some sort of magic proprietary secret sauce in the algorithms used by whoever developed the controller they are using. As an engineer I'd certainly not minimize the complexity of these control systems but the essentials of all these controllers and their associated firmware is going to be practically equivalent. Certainly "features" differ, but I strongly suspect any difference will come down to tuning.

Lots of excellent $xxxx+ gimbals used by pros every day use alexmos or some derivative, they just don't announce it or publicly associate. Sort of like how Linux may well be running on your washing machine. It doesn't really matter if it does the job, and that's one of the points of open source.

First, Alexmos isn't open source and has no derivatives. Second, assuming all gimbal controllers are equivalent is very far from reality. Gimbal controllers are potentially more complicated than flight controllers (they certainly require much higher bandwidth, precision and consistency). There are huge variations between the controllers, and that even without considering autotuning.
The sophistication and level of math that goes into the development in Phobotic (both past and future) is something I've never predicted when I started this company, and there's lots more to do in this field.

 

[SamaraMedia]

IMO, the Infinity MR is one of the better gimbals on the market, especially results posted by krall and jakuban using the AM with encoders, both in windy conditions. Hopefully more videos will appear with similar results.

After watching SleepyC's back lot auto video and all the out of level horizon footage from the Gremsy (no disrespect Sleepy, I know that was shot on a un-tuned mr since you had that radio mishap) I would put my money on the Infinity MR but since I'm cash strapped right now I will continue to try and tune my QuadFrame DSLR 3 axis gimbal with CP once I get a new yaw motor ( shorted out the windings when I inadvertently tightened the mounting screws down too far ). And since QuadFrame is out of stock on the gimbal motor and I will need to order directly from RC Timer or purchase a different model from those in the states that have a similar motor (different winding 150 vs 200). I did purchase a T-motor gimbal motor I thought would work according to motor dimensions of both but upon closer inspection found that the RC Timer motor is 23.5 mm vs 24 mm claimed in their spec sheet (should have known the * meant something...)

I'm hoping to find something that will work well under my X6 and when I get the funds it will likely be the Infinity MR for its ability to accept a variety of camera/lens.

@crayfellow - have you had a chance to get yours airborne?

 

[crayfellow]

The sophistication and level of math that goes into the development in Phobotic (both past and future) is something I've never predicted when I started this company, and there's lots more to do in this field.

As I said, I've not contributed to gimbal controllers specifically, but have worked on many similar technologies including commercial air data computers and GPS/navigation systems which are spectacularly complex. But certainly, yes, these things are not simple.

Cheers.

 

[crayfellow]

IMO, the Infinity MR is one of the better gimbals on the market, especially results posted by krall and jakuban using the AM with encoders, both in windy conditions. Hopefully more videos will appear with similar results.

...

@crayfellow - have you had a chance to get yours airborne?

Not yet, work has been crazy. And now that I lost my 250 last night I will be a bit more skittish than usual putting several orders of magnitude invested in the sky (mostly kidding, except for the loss; I'm getting close on the build so maybe over the next week or so)

 

[jakuban]

Wow, what a discussion we have here
@tombrown1 I'm not taking your words as fight trial, no worries. I just think that what @krall achieved was really good. Of course you're right, in the last part of that video horizon went sideways and I shouldn't say it is perfect.
You can watch one of my tests when I am flying on 35mm lens - in that test horizon is perfect, I think...
And thank you for those kind words about MR and my job
To be honest I don't know why but sometimes I can't see any drift on horizon and sometimes I can. It happens for example, when I land just to change main battery. First flight is ok, but second not. Maybe it just shows how complex the problem is. As Roee said, maybe gimbal electronics is much more complex than autopilot systems. For sure, every electronics manufacturer has smaller or bigger problems with horizon, even DJI, and we know they are working on that.

I have some feedback form MR users and hopefully soon we will see lots of tests. Anyone who has already InfnityMR with AM is asked to send me his good-working profile. I am collecting those data to make a nice list of cameras, lenses and multirotors that work with InfinityMR. This list will be available for everyone and everyone will be able to use profile that we know works good with proper camera-lens-multirotor setup.

If there is anyone who's thinking about InfinityMR purchase, now it is good time to order one. I have last few units on stock ready to ship within 24H. Production of the next batch is starting next week and it could happen that for a while (1-2 weeks) I won't have any gimbal on stock.

IMO, the Infinity MR is one of the better gimbals on the market, especially results posted by krall and jakuban using the AM with encoders, both in windy conditions. Hopefully more videos will appear with similar results.

Thanks a lot for those words!

 

[saenzm]

Hi guys, very interesting discussion.
What I've seen in the infinity tests is:
In last Jakubs video when yawing horizon is perfect, even in fast movement .
On the last video ,(Krall's) I can see the horizon drift in a sideway motion, rolling.
The other issue is when going forward or backwards the gimbal tilts with the kopter. Jakub told me that Alexey is working on that.
So yawing is working, rolling not.

What I noticed in the video when Jakub shows how the gimbal returns slowly to position after pushing it gently with the hand.
I have a zenmuse H32D for gopro, and when its powered you can't move it if pushing it with your finger. It is really stiff. Perhaps there, is something that makes the difference. This little gimbal works awesome. And of couurse it is connected to the FC.

 

[saenzm]

That is the video from the settings I did send you Can't wait to try out the changes you told me about. The wind was horrible, so in normal conditions, and with your changes I guess the result will be awsome

Hi Knut,
I was wondering how did you get on with the Infinity. Any other video to show?
Thx

 

[Mediakopter.no]

Hi Knut,
I was wondering how did you get on with the Infinity. Any other video to show?
Thx

Hi.

I have been really busy, so I have not had much time flying. I still have some tuning to do, but I am getting better and better video. Jakub has been really helpful in the tuning prosess. (Remember that I am a total idiot when it comes to AM, but I am learning) I will post PIDs and video vhen I get near perfect video. Right now I'm at good enough for TV. I want cinema quality

I have a meating and demo with a producer tomorrow, and a photo job friday, so probably will not have any time tuning until next week.

 

[saenzm]

Nice to hear its getting better! Perfection doesn't exists, but if you are looking for perfection you will surely arrive to excellence.
Good luck in your meating

 

[Pantos]

Hello Jakub,

I'm planning to buy the new sony rx100 iv what seems incredible good for the proportions, and I wanna ask you:

I saw the extremely good work you've done with the InfinityMR, do you have something similar (with encoders also) compatible for the rx100 iv? like an miniInfinityMR?
Or will you recomend the InfinityMR for the rx100? It seems an overkill.

Thank you and keep the good work!!