Komodo + DJI RS 2

[Greg M]

I'm a bit underwhelmed by the RS 2. It feels like it's trying to be a "pro-tool" but no d-tap- which means no easy way to mount a pro focus motor, high nit monitor, or teradek shows me that they came up short. So freaking close, but no cigar.

it cost $800...maybe expectations out of line with reality?

 

[Christoffer Glans]

With the addition of this little thing, the RS2 will be a killer using cine lenses and Komodo. I talked a while back about making the AF system on the camera interact with focus gears for using AF with cine lenses instead of internal motors. Back then everyone seem to say that it would be too slow and that it wouldn't work. But DJI seem to have done it exactly like I imagined the system to work. If DJI implements a face tracking system into this, we won't need any AF on the Komodo at all but instead could get a much more reliable AF through the DJI system, but be able to use it with lenses like the Master Primes.

That will be pure gold and I feel like not many really seem to grasp the potential of such a system. Being able to use an AF system for a majority of shots in a film while doing more specific focus pulls only when needed. And working with super shallow DOF like the Master Primes provide at wide open but being able to essentially lock the focus point on a moving object in a low light scenario would be massive.

https://store.dji.com/se/product/ronin-3d-focus-system?from=search-result-v2&position=0

 

[Björn Benckert]

With the addition of this little thing, the RS2 will be a killer using cine lenses and Komodo. I talked a while back about making the AF system on the camera interact with focus gears for using AF with cine lenses instead of internal motors. Back then everyone seem to say that it would be too slow and that it wouldn't work. But DJI seem to have done it exactly like I imagined the system to work. If DJI implements a face tracking system into this, we won't need any AF on the Komodo at all but instead could get a much more reliable AF through the DJI system, but be able to use it with lenses like the Master Primes.

That will be pure gold and I feel like not many really seem to grasp the potential of such a system. Being able to use an AF system for a majority of shots in a film while doing more specific focus pulls only when needed. And working with super shallow DOF like the Master Primes provide at wide open but being able to essentially lock the focus point on a moving object in a low light scenario would be massive.

https://store.dji.com/se/product/ronin-3d-focus-system?from=search-result-v2&position=0

As I understand there is quite a few of these systems on the market or brought to market as we speak. They work differently then Pdaf systems, which work off the image of the actual lens so yes this kind can work with lenses that pulls focus more slowly. But these are running into other inaccuracies then the Pdaf way of doing things. A system like this will struggle more with inaccuracies between what the system measures the focus to and how that is applied to the optics, basically it does not see if the focus is slightly of, if there is play in the focus motor or if the scale of the lens is inaccurate or is not following a 3 point curve, etc. So while Pdaf is searching for a sharp image, this only provides a distance to target, when working with short focus the tolerances are very narrow so everything will need to be set up correctly, from back focus, motors without play, alignment with camera backplane etc. So with accurate lenses like masterprimes, good motors and alignment etc it will likely work quite well but could easily be slightly of when changing lens or such, where as Pdaf try to reach and find tack sharp no matter what plasticy little lens you trough at it. So both have their benefits.

 

[Christoffer Glans]

As I understand there is quite a few of these systems on the market or brought to market as we speak. They work differently then Pdaf systems, which work off the image of the actual lens so yes this kind can work with lenses that pulls focus more slowly. But these are running into other inaccuracies then the Pdaf way of doing things. A system like this will struggle more with inaccuracies between what the system measures the focus to and how that is applied to the optics, basically it does not see if the focus is slightly of, if there is play in the focus motor or if the scale of the lens is inaccurate or is not following a 3 point curve, etc. So while Pdaf is searching for a sharp image, this only provides a distance to target, when working with short focus the tolerances are very narrow so everything will need to be set up correctly, from back focus, motors without play, alignment with camera backplane etc. So with accurate lenses like masterprimes, good motors and alignment etc it will likely work quite well but could easily be slightly of when changing lens or such, where as Pdaf try to reach and find tack sharp no matter what plasticy little lens you trough at it. So both have their benefits.

Of course, but if the mount is locked into place and you use Master Primes with correct back focus, it should be the same between lenses since that's the whole point of using expensive cine glass; that they are exactly the same throughout the lineup. But I agree that auto-focus in-camera is superior and that was what I proposed before about a system that could pull focus data out of the camera to an external focus motor. But this idea went to deaf ears which I don't understand. A system that utilizes a good in-camera AF system with face and eye-tracking that can pull that data out to a focus motor would be perfect. But no one seems to care about this. The lack of innovation leads to solutions like this from DJI, which is by far better than the lack of interest that so many other manufacturers have about these kinds of things.

I still don't understand why no one puts effort into making a really good AF system that works with focus motors and in-camera data. The tech is so cheap and out there in inexpensive consumer gear that someone somewhere would have just DIY themselves something like this.

But, until that happens, DJI focus system is by far more impressive than anything else I've seen and it's not really that worse than a regular focus puller when it comes to wide open fast lenses. It's only going to get better when face tracking is implemented.

 

[Thad K Donovan]

The Lidar solution DJI is using has a lot pf potential and can certainly be used well in many cases (especially as the aperture gets higher). However, there's a lot of soft focus issues in the Youtube video example, especially in the eyes, which is the most important part IMO. A face that is in focus with blurry eyes is not a shot I can use.

Also, the inability to have a focus range outside of the middle of the image is unfortunate and limiting.

I think this is a great first step, but we have a ways to go to reach that holy grail of trustworthy autofocus. I have full faith that these current limitations will be resolved eventually. I also believe that the current tech is quite usable in many cases. Somebody needs to combine the best of all the technologies to ensure eyes being in focus is the priority and that we have full range of the image to focus on.

I look forward to the day when all of this stuff is built in to cameras/lenses so we can avoid clunky work arounds. Thankfully there is huge demand for good autofocus and a lot of market pressure to innovate. We're getting closer!

 

[Christoffer Glans]

The Lidar solution DJI is using has a lot pf potential and can certainly be used well in many cases (especially as the aperture gets higher). However, there's a lot of soft focus issues in the Youtube video example, especially in the eyes, which is the most important part IMO. A face that is in focus with blurry eyes is not a shot I can use.

Also, the inability to have a focus range outside of the middle of the image is unfortunate and limiting.

I think this is a great first step, but we have a ways to go to reach that holy grail of trustworthy autofocus. I have full faith that these current limitations will be resolved eventually. I also believe that the current tech is quite usable in many cases. Somebody needs to combine the best of all the technologies to ensure eyes being in focus is the priority and that we have full range of the image to focus on.

I look forward to the day when all of this stuff is built in to cameras/lenses so we can avoid clunky work arounds. Thankfully there is huge demand for good autofocus and a lot of market pressure to innovate. We're getting closer!

As I understand it, tracking subjects may be coming in future firmware. The key thing here is that it's doing what I've asked for, for years now; the ability to use AF for cine lenses. I really don't understand why it would be hard to engineer a focus motor that gets AF data from the camera itself. If you have something like dual pixel eye-tracking and you simply have a cable from the camera to the focus motor pulling that data and transferring it to the external motor instead of the internal motor of a regular still lens, that would open up any cine lens that's ever been around to be able to function with AF, without the stutters and bad quality of still lens focus pulls. It boggles my mind that this thing isn't even on most people's radar. Here comes DJI with a somewhat rough first step that sidesteps camera AF and just... does it. That gets a gold star from me, even if it's not fully functioning right now. Many have talked about cine lenses getting internal motors, but even if one company does it, it doesn't really matter. There are so many cine lenses out there that people choose for their looks, so why not just make AF systems that work with all of them instead of waiting for every cine lens manufacturer to make cine lenses with internal motors.

I really wish there was some way for Red to pull the AF data out of the camera to a focus motor. It's like the same with how the internal gyro in the DSMC2 has the ability to post-stabilize in an instant if only we got such an option in the R3D RAW settings. It's all there, but people only seem to understand the power of them when it's only theoretical.

 

[Björn Benckert]

As I understand it, tracking subjects may be coming in future firmware. The key thing here is that it's doing what I've asked for, for years now; the ability to use AF for cine lenses. I really don't understand why it would be hard to engineer a focus motor that gets AF data from the camera itself. If you have something like dual pixel eye-tracking and you simply have a cable from the camera to the focus motor pulling that data and transferring it to the external motor instead of the internal motor of a regular still lens, that would open up any cine lens that's ever been around to be able to function with AF, without the stutters and bad quality of still lens focus pulls. It boggles my mind that this thing isn't even on most people's radar. Here comes DJI with a somewhat rough first step that sidesteps camera AF and just... does it. That gets a gold star from me, even if it's not fully functioning right now. Many have talked about cine lenses getting internal motors, but even if one company does it, it doesn't really matter. There are so many cine lenses out there that people choose for their looks, so why not just make AF systems that work with all of them instead of waiting for every cine lens manufacturer to make cine lenses with internal motors.

I really wish there was some way for Red to pull the AF data out of the camera to a focus motor. It's like the same with how the internal gyro in the DSMC2 has the ability to post-stabilize in an instant if only we got such an option in the R3D RAW settings. It's all there, but people only seem to understand the power of them when it's only theoretical.

The gyros in DSMC2 does not work, they are there but their accuracy is not anything you could use to stabilize an image with. The jitter in them is worse than a camera operator that suffer from severe Parkinson.

No cinema cameras had Pdaf sensors before. This will for sure change things. But likely it will take some time, and also there is likely a huge minefield of patents between where cameras are today and a place where you connect a focus motor directly to the camera and viola, autofocus with any lens. As I understand Arri has patent for connecting a focus motor directly to the camera body... Arri does not have a Pdaf camera... Arri likely does not want to license out such tech to other manufacturers...

 

[Sergio Perez]

I use a Ronin S all the time with my Monstro for some unscripted shots that I love to use in my promotional videos work. Here's a recent example (there are some steady cam and dolly shots in there too- all sot on RED Cameras except the drone footage)

https://www.youtube.com/watch?v=IYBrObVdkt4

I have a Ronin S2 coming my way and can't wait to try it on a DSMC2. If I can use it properly without having to use it "reversed" , as I currently use to support the weight, and with the amazing looking Tilta ecosystem of support gear, this will truly be a game changer for my work.

Anyone tried it with a Bare bones, self contained DSMC2?

Can't wait for it to arrive!

 

[Christoffer Glans]

The gyros in DSMC2 does not work, they are there but their accuracy is not anything you could use to stabilize an image with. The jitter in them is worse than a camera operator that suffer from severe Parkinson.

Many others have mentioned how they have used the data for just this purpose, but if not, then why not have improvements to the gyro? If a SteadXP gadget can do it, why not just put a sufficient frame-synced gyro in the cameras? It's a no brainer as to how powerful this feature would be, just ask Fincher.

No cinema cameras had Pdaf sensors before.

No, but Canon has dual pixel which is superior, so AF systems for cinema cameras exist.

This will for sure change things. But likely it will take some time, and also there is likely a huge minefield of patents between where cameras are today and a place where you connect a focus motor directly to the camera and viola, autofocus with any lens. As I understand Arri has patent for connecting a focus motor directly to the camera body... Arri does not have a Pdaf camera... Arri likely does not want to license out such tech to other manufacturers...

And this is why it never happens, why things are at a standstill. But there are workarounds. If Arri has a patent of connecting it directly to the camera, then connect to something else first. Come to think of it, hasn't Red already done this hardware of a direct connection to the camera? With that said, patents can be worked around based on how they are formulated. I don't see how a camera can't have a better internal gyro for 3D camera track and stabilization, and lidar sensors that work in tandem with PDAF AF. Patents are often very specific to single purposes so how you think about the tech defines if it breaks a patent or not.

But that's not really what I'm talking about, I'm talking about the unwillingness of the industry to really innovate, to really push tech forward. It's the small guys that innovate, companies like DJI who tries to change things up. While Canon's cinema cameras aren't their largest focus, they still aim in the right direction with stuff.

It's really only a matter of time before one of the less industry-standard companies release something that truly shakes things up. Imagine if Canon blasts together a compressed RAW global shutter high DR camera with a full-frame 8K, quad pixel AF sensor going into an infrastructure relation with DJI and unreal engine for cheap motion control, stabilization and camera tracking application that works in realtime with high-resolution lidar depth maps for VFX. We are certainly not near that today, but it won't be Arri or Red doing this when it happens.

 

[David J. Buchanan]

I use a Ronin S all the time with my Monstro for some unscripted shots that I love to use in my promotional videos work. Here's a recent example (there are some steady cam and dolly shots in there too- all sot on RED Cameras except the drone footage)

https://www.youtube.com/watch?v=IYBrObVdkt4

I have a Ronin S2 coming my way and can't wait to try it on a DSMC2. If I can use it properly without having to use it "reversed" , as I currently use to support the weight, and with the amazing looking Tilta ecosystem of support gear, this will truly be a game changer for my work.

Anyone tried it with a Bare bones, self contained DSMC2?

Can't wait for it to arrive!

+1 on this question

 

[Christoffer Glans]

I believe these would help any setup that gets out of balance
https://store.dji.com/se/product/r-roll-axis-counterweight-set

 

[Johannes Helms]

Made a little video on that topic

 

[Christoffer Glans]

Made a little video on that topic

Great video as usual. I agree with the cable conversion, it would have been perfect with an SDI instead. I think they should do a Raveneye update with SDI.
Is it possible to feed an SDI signal to a monitor that has an HDMI output going into the Raveneye? Could that be a solution?

My personal combo with this is going to be the Canon speedbooster and CN-E lenses so I'm worried that it's gonna be front heavy. I ordered the 955 batteries which I think are smaller than the ones you used in the video? Are those 975?
Do we know when the Lidar focus is coming?

 

[Christopher A. Bell]

Has anyone else noticed that the 'Sport Mode' is a lot less aggressive than on the original Ronin S? The original is very responsive and quick, allowing you to throw and whip it around with pretty darn good precision and control, but the RS2 is very "laggy" and "soft" in 'Sport Mode'(feels like typical telltale gimbal delay). Don't get me wrong, it is a great gimbal that, so far, feels like a worthwhile upgrade, except for the 'Sport Mode' taking a big step backwards(not even being at least 'as good' as the original, but instead worse). Pretty sure this could be fixed in a firmware update. Just curious, though if anyone else has noticed or feels the same.

 

[Johannes Helms]

Great video as usual. I agree with the cable conversion, it would have been perfect with an SDI instead. I think they should do a Raveneye update with SDI.
Is it possible to feed an SDI signal to a monitor that has an HDMI output going into the Raveneye? Could that be a solution?

My personal combo with this is going to be the Canon speedbooster and CN-E lenses so I'm worried that it's gonna be front heavy. I ordered the 955 batteries which I think are smaller than the ones you used in the video? Are those 975?
Do we know when the Lidar focus is coming?

The SDI to monitor with HDMI out and back into raven eye should work as all the raveneye takes to do it's magic is the video signal. I could try it with my Shogun Inferno if I find time.

Actually those are 955s

And not sure, one guy on YouTube said in November. I have a stock reminder set and will get it asap.

 

[Tyrone Braithwaite]

What plate are you using. The arca Swiss or the manfrotto? Surely the arca Swiss will only allow one bolt to the body of the komodo?

 

[Daniel Kelly Brown]

Any word on if Red will be working to get Komodo integrated into the RS2? It would be awesome to have the ability to control focus on AF lenses with the wheel on the RS2 as well as have the ability to start/stop record from the RS2 record button.

Would this even be possible?

Cant seem to find a good way to control focus on Komodo and RS2 even with the DJI focus motor and manual focus lens. I could be wrong but there does not seem to be a way to calibrate hard stops unless you have the separate little focus wheel DJI sells (out of stock).

Really wish RT latitude would work on Komodo, wondering what the advantages are to messing with the ext port on komodo so this does not work. Thought I could work around it with the RS2 focus tech, but even if I can, the fact that it does not work to control AF motors is a downer.

If I don't get some clarity soon on these issues from either Red or Teradek I might be swapping the Komodo for the A7sIII so its all just works and I can mess around with the 3D focus active track tech without hdmi to sdi adapters. I really don't want to do that, but I do need to control focus on the Komodo via AF motors and a thumbwheel on a gimbal.

 

[Christoffer Glans]

Any word on if Red will be working to get Komodo integrated into the RS2? It would be awesome to have the ability to control focus on AF lenses with the wheel on the RS2 as well as have the ability to start/stop record from the RS2 record button.

Would this even be possible?

Cant seem to find a good way to control focus on Komodo and RS2 even with the DJI focus motor and manual focus lens. I could be wrong but there does not seem to be a way to calibrate hard stops unless you have the separate little focus wheel DJI sells (out of stock).

Really wish RT latitude would work on Komodo, wondering what the advantages are to messing with the ext port on komodo so this does not work. Thought I could work around it with the RS2 focus tech, but even if I can, the fact that it does not work to control AF motors is a downer.

If I don't get some clarity soon on these issues from either Red or Teradek I might be swapping the Komodo for the A7sIII so its all just works and I can mess around with the 3D focus active track tech without hdmi to sdi adapters. I really don't want to do that, but I do need to control focus on the Komodo via AF motors and a thumbwheel on a gimbal.

Why not get cine lenses? Focus motor and wheel is primarily for that. Komodo autofocus is not yet out of Beta so we won't know what it will be able to do.
But if you want the functionality of an a7s using still lenses, then go with that instead. Komodo will never be a DSLR type camera.

 

[Russ Fill]

Made a little video on that topic

Wonderful video and loads of info. Thanks.
I think Im going to keep my R2 but might want a new RS2 for some jobs.

 

[Christoffer Glans]

I wonder if Red could make the Komodo Link work with the RS2? If not using the HDMI, the USB-C works with cameras using that instead (like the A7SIII), so why not make the Komodo Link compatible with the RS2?