Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

KOMODO....

Status
Not open for further replies.
jordan levy said:
it's Damn good, and listen, I made my living from being a focus puller for some of the biggest ASC cinematographers in the business, i was the first one to balk at using autofocus professionally, I even complain about AC's using wireless, still, i cant take over if im doing something artistic, they are like 15ft from me when they used to be next to me. (i yell more) we went through light rangers and cine tape and all sorts of tools in the day to get some almost impossible shots.

Not until I had just got my A7III for my new scout camera, got rid of the big 5D finally and i was doing some rehearsals for a dance shoot on my scout camera just to find out what camera angles should be as it was to be a one shot video. i would get back the tests and um, it was focusing for me and pretty well. I then dove into youtube tutorials to see what the best settings were and how accurate it was, especially facial tracking, starting doing tests, and it was so dead on i would up talking the band into buying me a ronin S and i shot the video on the A7. since then I've pretty much done every personal "friend" project on the A7 and i have never even thought about focus, it's just there. right before i sold my raven i had to do a video for a friend and i said, what hell, now that i know Davinici so well too, ill shoot this on the raven and see how it compares. it came down to both images been fine, i actually found the sony easier to grade, while the red footage definitely had more latitude, but what really made the difference was no matter how hard i tried to punch in, dial in focus for every shot, have the actors freeze, there was still a small percentage of shots that were slightly out. So, that's when i became and advocate, it's when it's became this good. and i dont use it all the time, depends on the shot obviously, i sometimes use it to set up the scene and sometimes ill use the quick button to disable AF and pull the rest in myself, mainly for racks.

I would still love to own something with a little better depth in resolution, so hoping for the 10 bit version, but if i was to shoot a doc again, i would be renting the new sony fx9. but i still have hopes for Komodo, i just couldnt deal with the BM pocket, i love my OG pocket and still have it a use it, but the new one has no excuse, its big, its weird, it need tons of things to be rigged off it to make it work, so i sold mine. its too bad because the image was nice and what i love about a personal camera is the ease of use, ease of shooting, ease of post, make it easy.
Click to expand...

Imagine this with a focus gearhead for cine-lenses. Imagine if Red bought a Sony license for their auto-focus technology and worked it into cine-lens support for Komodo. We could dream at least.
 
A perfect quality AF, is literally the first feature I spoke about on Jarred's facebook.
A good autofocus system paired with an Active Track on a Ronin-SC is the key to low to middle budget project with small crews.

I did my first mirrorless shoot on a Nikon Z6 with a ronin-SC, and it is amazing to be able to do that much with so few.
Sure it was not perfect, but I was alone to do it.

Komodo needs to be competitive in AF, an IBIS, in battery life, bootup time, playback/record switch and be compatible with a few DSMC²accessories.
These are the improvement needed on any Red cameras.

We already know battery life will be good and cheap. I just hope it has a pogo port on top to put a 4.7", and the weird on/off switch means fast boot up time.



On the camera side, I'm really hoping it does not have the highest specs of the world but the best image in its class, similair to at least Dragon on a DSMC² (which is still rocking hard)
Middle specs with new techs, less power consumption, less heat...

Not necessary cutting edge like DSMC² brains.
 
My first feature wish would be AF, and my second stabilizaiton.

Also its really time that we could have use of the built in gyro data for stabilization in post. We had gyro since epic mx but nothing happened yet....Some one please?
Was it the new sony that saves gyro data to meta data?
 
jordan levy said:
Not until I had just got my A7III for my new scout camera, got rid of the big 5D finally and i was doing some rehearsals for a dance shoot on my scout camera just to find out what camera angles should be as it was to be a one shot video. i would get back the tests and um, it was focusing for me and pretty well. I then dove into youtube tutorials to see what the best settings were and how accurate it was, especially facial tracking, starting doing tests, and it was so dead on
Click to expand...

This is one of the most interesting comments I've read in a long time. Back in the early '90s, I asked a friend of mine who was in film school, how come AF wasn't used in movie cameras when cheap 35mm compact cameras had it? I forgot his answer, but whatever it was, he was correct. ;-) If the technology was useful, at least some film crews would have used it.

We've had AF since the 1980s for goodness sakes, and only recently have we been given AF that actually works for motion (and even then, not in every situation). No matter how much you heard about how SLRs and DSLRs could focus really quickly (but not always accurately), that didn't change the fact that AF only worked for photos. Even predictive tracking wasn't good enough to follow someone in motion. Phase detection AF points above a mirror cannot see eyes or faces. When the mirror is up, it's even worse because they can't see anything at all. And when something interrupts their field of view, they don't always know what to do. But with active sensors which allow for image processing, you can have very, very effective subject tracking.

Almost 40 years after AF hit the mass market are we seeing something good enough for video. We would have never seen this coming from a 1990s perspective, but technology often surprises us - in what it can do, and in what it cannot do.
 
Christoffer Glans said:
Imagine this with a focus gearhead for cine-lenses. Imagine if Red bought a Sony license for their auto-focus technology and worked it into cine-lens support for Komodo. We could dream at least.
Click to expand...

Sony also said they are going to be releasing lenses Q1 2020 built for cinema but with their AF tracking capability, especially made for the fx9, and yes joel, mentioned in the sony announcement was gyro meta data recorded for easier post stabilization, which will soon be available in all editing programs.
 
Bérenger Brillante said:
A perfect quality AF, is literally the first feature I spoke about on Jarred's facebook.
A good autofocus system paired with an Active Track on a Ronin-SC is the key to low to middle budget project with small crews.
Click to expand...

If it's on par with the best autofocus systems on the DSLR market, then yes, if it's slightly working but not good enough it's better to just skip it and focus on getting other stuff better. I've worked with "good" auto-focus and it generally gets 50% of the shots correct and the rest is just unusable. So if it works just some of the time and isn't consistent and reliable I'd skip it as a feature.
 
Since we are on the features we'd want, AF is not my top priority (and if it had AF licensed from another company I would like Canon Dual Pixel AF).
Mine would be to have the highest pixel count possible in 2x anamorphic so IF the sensor is the same 6K resolution as Dragon 6K, then I'd want it to be 4203x3160 pixels in 2x anamorphic mode.
 
Also interested in the status of accessing the on board gyro data. Since it was first introduced, I've imagined a stabilization tool in RedCineX-Pro. Would love to be able to do a first light grade, stabilize to taste - I'm thinking sliders for pitch, roll and yaw - then push out ProRes, DPX, etc for editorial/finishing.

Cheers - #19
 
Blair S. Paulsen said:
Also interested in the status of accessing the on board gyro data. Since it was first introduced, I've imagined a stabilization tool in RedCineX-Pro. Would love to be able to do a first light grade, stabilize to taste - I'm thinking sliders for pitch, roll and yaw - then push out ProRes, DPX, etc for editorial/finishing.

Cheers - #19
Click to expand...

Definitely. I've advocated for something like this for a long time. There's a Kickstarter company who created SteadXP, which is mostly aimed at GoPro's, but can be used on Red's. It creates really great stabilization if you know how to use it, but it's complicated.

I don't really understand why we can't tap into that power with cameras, it's such an easy thing to do really; create a fast per frame gyro-data recording into metadata and let a stabilization algorithm be open in SDK of Redcode and options in RedcineX-Pro. The other truly powerful thing about this is that the gyro data would give you a direct camera tracking in 3D space for VFX. Instead of cumbersome camera tracking, masking out everything that interferes with the camera movement, the gyro data would give you a virtual camera directly so with a click of a button from the R3D metadata, you would have a virtual camera ready.

It's not really that hard of a feature to implement, just do it, and it will be truly revolutionary as an option to have both for stabilization and VFX.
 
Christoffer Glans said:
If it's on par with the best autofocus systems on the DSLR market, then yes, if it's slightly working but not good enough it's better to just skip it and focus on getting other stuff better. I've worked with "good" auto-focus and it generally gets 50% of the shots correct and the rest is just unusable. So if it works just some of the time and isn't consistent and reliable I'd skip it as a feature.
Click to expand...

It's not about R&D ressources, it's about features people in that price range want. Sony A7 series' AF is accurate 80% of the time, and when it's not, it's mostly time related and will get accurate in seconds. (i'll admit, or not at all)
Sure, that leaves room for improvement when the A7IV will be released in the years to come

You have to understand every market, and at that price range :
AF & IBIS are mandatory.
otherwise users will just be trading ease of use over image quality, and some won't.

Of course, feature films, or any film with repetitive actions, would need a 100% accurate AF.
Actually no, it needs a focus puller with a nice touch and feel of the scene. And for these project, there are several DSMC²s.

But 50% of commercials, documentaries, any music videos can be made with an AF and in 99% of them, the AC can be assisted with the AF.
Corporate work can be made 99% with an AF.

And research on these features can be added to other brains when the time comes.
 
Christoffer Glans said:
Definitely. I've advocated for something like this for a long time. There's a Kickstarter company who created SteadXP, which is mostly aimed at GoPro's, but can be used on Red's. It creates really great stabilization if you know how to use it, but it's complicated.

I don't really understand why we can't tap into that power with cameras, it's such an easy thing to do really; create a fast per frame gyro-data recording into metadata and let a stabilization algorithm be open in SDK of Redcode and options in RedcineX-Pro. The other truly powerful thing about this is that the gyro data would give you a direct camera tracking in 3D space for VFX. Instead of cumbersome camera tracking, masking out everything that interferes with the camera movement, the gyro data would give you a virtual camera directly so with a click of a button from the R3D metadata, you would have a virtual camera ready.

It's not really that hard of a feature to implement, just do it, and it will be truly revolutionary as an option to have both for stabilization and VFX.
Click to expand...

No the Gyro does not give you a 3D camera only rotation data an accelerations. For example constant speed it can not recognise. So if you make a smooth dolly move...

Butt yes with RED we already have all that is needed in the Metadata to do what Steady XP does.

As with EF mount the lens is known (Same even with data carrying PL lenses.) So thats even better then SteadyXP that need to learn the lens used.
We know the shutter speed and can get a chart for the redraw / rolling shutter time.
And the Red gyro data is written into each frame as I understand. even if you need to code to get to it. (or if its only in the THC, not sold any longer)


So I'm sure bigger VFX houses are already tapping in to the above, strange if they dont. Possibly the red gyro is not up to spec but strange if so. any mobile phone gyro would be plenty and those cost cents.


As for 3D tracking I have asked the guys making Syntheyes to tap into the r3d´s to help their pixel based camera 3D tracker to improves their solves and I think they are working at it. Because the thing with 3D tracking is that not only do you need to know if the camera pan tilt, pan or roll you also need to know if its moving and usually also you need to know where its moving relative to something in front of / around the camera.

Example shoot with free moving camera and replace somebody's head... then you need to know where the head is relative to camera. And even if tracking a fixed environment its kind of crucial to know where in 3D space everything is so image based tracking will still be needed even if the camera path is known.


We did a tricky one a year back. Actually won a VFX / Rokoko award for it. The man on the street will not understand the level of difficulty but basically we used a inertia sensor suite on a rollerblader and then we used a extra fixed witness camera to 3D track the skates and attached the suit tracking data to the skates all just to make a stupid rollerskating robot. :)

https://youtu.be/779U8MJtols
 
Björn Benckert said:
No the Gyro does not give you a 3D camera only rotation data an accelerations. For example constant speed it can not recognise.
Click to expand...

Yes, forgot about an accelerometer and then I think the gyro data as of now is only bound to 24/25 fps, not higher framerates. However, if an iPhone or Apple Watch both have a gyro and accelerometer able to track all movement of the phone/watch, it isn't expensive or complicated to include such a feature in a camera body. The application for such data to be included per frame, but also have it as a SDK feature so that RAW settings in for example Resolve can have a stabilization slider and a "copy tracking data" option would be extremely good to have.

The problem isn't the data itself, or how to grab that data, but that it's just as complicated to implement that into a workflow as it is to use SteadXP. I want features that take the data and make it integrated into the RAW settings of R3D files. So that you could change the stabilized values just as you can set ISO and white balance levels for a shot. This shouldn't be hard for Red to implement even as a basic feature for current gen cameras.
 
Christoffer Glans said:
Yes, forgot about an accelerometer and then I think the gyro data as of now is only bound to 24/25 fps, not higher framerates. However, if an iPhone or Apple Watch both have a gyro and accelerometer able to track all movement of the phone/watch, it isn't expensive or complicated to include such a feature in a camera body. The application for such data to be included per frame, but also have it as a SDK feature so that RAW settings in for example Resolve can have a stabilization slider and a "copy tracking data" option would be extremely good to have.

The problem isn't the data itself, or how to grab that data, but that it's just as complicated to implement that into a workflow as it is to use SteadXP. I want features that take the data and make it integrated into the RAW settings of R3D files. So that you could change the stabilized values just as you can set ISO and white balance levels for a shot. This shouldn't be hard for Red to implement even as a basic feature for current gen cameras.
Click to expand...

I think when I talked to Graeme about it the data is there so its more about people / third party VFX applications that does not care to implement it. The thing is you need more info than just the gyro data, sensor size, shutter speed and redraw time. You also need to know the focal length and what actual lens / the distortion and preferably also the vignetting for it to work properly... There is quite a few lenses out there so quite a bit of data to implement in the SDK... But yes, red is doing focus mapping for most common EF lenses into the firmware of the cameras so could start adding distortion as well I guess.
 
Björn Benckert said:
I think when I talked to Graeme about it the data is there so its more about people / third party VFX applications that does not care to implement it. The thing is you need more info than just the gyro data, sensor size, shutter speed and redraw time. You also need to know the focal length and what actual lens / the distortion and preferably also the vignetting for it to work properly... There is quite a few lenses out there so quite a bit of data to implement in the SDK... But yes, red is doing focus mapping for most common EF lenses into the firmware of the cameras so could start adding distortion as well I guess.
Click to expand...

But much of that data is already in there and it wouldn't be that much more work to implement the rest. We can also have a lens database of distortions and such, community-driven, third-party and from lens manufacturers.

I actually see no real problems with any of this and the benifit to everyone would be enormous. At least start with a stabilization feature in R3D RAW settings. Imagine the help that would give productions, especially those with fast turnarounds.
 
Björn Benckert said:
I think when I talked to Graeme about it the data is there so its more about people / third party VFX applications that does not care to implement it. The thing is you need more info than just the gyro data, sensor size, shutter speed and redraw time. You also need to know the focal length and what actual lens / the distortion and preferably also the vignetting for it to work properly... There is quite a few lenses out there so quite a bit of data to implement in the SDK... But yes, red is doing focus mapping for most common EF lenses into the firmware of the cameras so could start adding distortion as well I guess.
Click to expand...

I have looked for this data in the past but you can't get anything via Nuke for sure and i couldn't actually find any data in RedcineX either - so i have no idea where this might be. I do think it is a missed opportunity but then i also have to ask if it's not commonly exposed then is it just not accurate or even why this data is there in the first place...? What was it designed for?

cheers
Paul
 
paulcurtis said:
I have looked for this data in the past but you can't get anything via Nuke for sure and i couldn't actually find any data in RedcineX either - so i have no idea where this might be. I do think it is a missed opportunity but then i also have to ask if it's not commonly exposed then is it just not accurate or even why this data is there in the first place...? What was it designed for?

cheers
Paul
Click to expand...

I know people used it. But can be that they got it via the THC that also records focus data.

If its there it can not be to difficult to write a script that extract it.

But agree, roll data for example... Shit easy to make a script that let you adjust roll, set hold on a frame and then use the rotation data to give a fixed horizon which would be far more than what the completion got. So yes Christoffer is right. It really suck that red does not implement atleast roll neutraliser in the SDK. It would not take more than a few days or even hours for a half talented in the field programer to write such feature. I got guys here that can do it.
 
It would appear that RED saw the potential for gyro data, put basic hooks into the SDK and expected 3rd party interest. AFAIK, the uptake was weak which meant RED saw little reason to devote engineering resources to it. Can't argue with that, they had plenty of other things to tackle. Whatever the actual case, and whatever the inherent limitations of the gyro in current RED bodies might be, would be great to see RED come strong on this. As Björn suggests, the smart phone vertical has driven development of chips that seem ideal for the use case - so perhaps it's just a question of coding resources...

Would love to see DSMC3 include robust tools for exploiting positional data, and I don't think I'm the only one.

Cheers - #19
 
Status
Not open for further replies.
Back
Top