Great advice, and I totally second the importance of having survey data. I can’t tell you how many times this kind of data would have been so helpful. I only got survey data twice in my 21 years of VFX work and both times I was supervising. :) Trying to get this data from other productions was impossible.
I am consuming your tutorials via brute force repetition. I enjoy and value your content. I hope I can get the hang of SynthEyes soon. It's so confusing, and I went to law school lol.
Awesome. The next two will be critical as well, mostly for the artists downstream. This is a video I really made to share with my production company partners, and I've been telling my filmmaker friends to share this far and wide.
Fantastic as always Matt. This really cements the value of a good survey shot, and has really simplified my life for a shot I'm working on out in woodland. The location has zero orthographic features, so to orient the scene, in my survey I hung a piece of string with two white beads I can track. They're exactly 1m apart, with a lead weight at the bottom. This, along with gravity gives me a precise 'up' vector and scale all in one. I've never seen this method used anywhere else, but surely I'm not the first to use it... Is it something you've come across before?
Sounds like what you are doing is just some good, creative tracking marker placement/rigging. I've seen it all, including things like what you describe, which sounds good. Just wait until you see the rest of the tutorials in this series! 😂
@@MatthewMerkovich Good to know. :) I'm really looking forward to see what your next tutorial delves into. By the way, do you still have a Discord account? If so, any chance of an link/invite?
Great video Matt! do you find that rolling shutter has any kind of tangible effect on the results or are you going so slowly that even on phone cameras this isn't an issue?
Also it's a pity that the iphone 12 lidar features don't have more range! I find as you pan over walls or flat surfaces, it kind of jumps a bit and adds in steps / levels onto what should be a single plane!
I'm never moving fast enough for rolling shutter to become an issue, and because the surveys are NOT being used in a comp directly, one doesn't need to worry about any rolling shutter artifacts in the footage. In SynthEyes, you may need to do some rolling shutter calculations at solve time to get the error down, but really, it isn't an issue. Great question though.
Hello Matthew, was just wondering, have you worked before with a witness camera rigged on the hero camera, but aimed down towards the street pavement, as a way to provide the tracking data for the hero camera's otherwise shallow depth of field and cropped frame (longer lens, which is giving me less points for trackers). I am using a Insta360 Leica 4k module that is sharp and deep. I plan on using its footage to solve the camera tracking. I am attaching it rigidly on the hero camera at an angle, downwards towards the area of the location that has the most markers, but always on the same axis as the hero camera's sensor. What do you think of this technique?
Hi, Rafael. Your method may work, but then you end up with the same issues one has with motion control: no point cloud, or in this case, no **meaningful** point cloud. That's my first thought. Second, if you are working on a big budget studio production, there are better ways to acquire more useful data sets to help with 3D tracking, LIDAR for one. On your own productions, where **you** are the camera department, I'd definitely shoot a set survey. The latest version of SynthEyes makes using the data even easier, which I will cover in part 3 of this series ... when I can make the time to get to it.
@@MatthewMerkovich Hi there Matthew, thanks for the quick response. Ok, so capture set survey with iphone or Insta360 One R for sharpness and depthness and use point cloud, check. But to acquire the camera motion, wouldn't the "witness camera" give me sharp data that a super shallow and blurred, tightly framed cine camera would not be able to?
@@___x__x_r___xa__x_____f______ Again, how do you align that witness camera data to your photography? Seed path mode? It really wouldn't help me if **I** got that witness camera footage. (And what do you do about gen lock, by the way?) I deal with whole sequences of shots in movies or TV shows and this approach is a non-starter on the production side, and it would be useless to me on the post side. But if you figure out how it would be helpful, please do share. I'd love to be proven wrong! ;-)
@@MatthewMerkovich so no solution then short of lidar or camera tracker like Ncam or MoSys. No hacks for something very simple? Like a very basic short movement?
@@___x__x_r___xa__x_____f______ Shoot a set survey and when doing your supervised trackers, aim for the middle of the blur. No need to reinvent the wheel on this.
Great question, and no, frame rate does not effect the set survey. I'll be discussing speeding up the footage in part two as a time saving method. You can even see that the footage shot by Image Video has been sped up in this tutorial.
Awesome Waiting for 2nd and 3rd Part
Every single video of yours is a GEM thank you very very very much for sharing Matthew, I just can't thank you enough.
Very Cool! Thank you for creating and sharing.
Thanks Matt!
Thanks Matthew. Now on to part 2!
Great stuff! Looking forward to the next parts!!
What an amazing deep content about topic. Man, thank you a lot that you Exist ! U make this CGI world better 😎
nice
Man, this is gold, thanks a lot!
Informative as usual!
Thanks! operating a camera and doing vfx for an independent feature with lots of action... these are great tips!!!
hopefully, You will have time to make other parts! This is what I want for a long time.
Extremely helpfull series, thanks!
Great advice, and I totally second the importance of having survey data. I can’t tell you how many times this kind of data would have been so helpful. I only got survey data twice in my 21 years of VFX work and both times I was supervising. :) Trying to get this data from other productions was impossible.
These are awesome. Thank you so much for doing them.
I am consuming your tutorials via brute force repetition. I enjoy and value your content. I hope I can get the hang of SynthEyes soon. It's so confusing, and I went to law school lol.
Fantastic! I will definitely request at Track VFX that we use this video where possible to aid with shoots.
Awesome. The next two will be critical as well, mostly for the artists downstream. This is a video I really made to share with my production company partners, and I've been telling my filmmaker friends to share this far and wide.
Fantastic as always Matt. This really cements the value of a good survey shot, and has really simplified my life for a shot I'm working on out in woodland. The location has zero orthographic features, so to orient the scene, in my survey I hung a piece of string with two white beads I can track. They're exactly 1m apart, with a lead weight at the bottom. This, along with gravity gives me a precise 'up' vector and scale all in one. I've never seen this method used anywhere else, but surely I'm not the first to use it... Is it something you've come across before?
Sounds like what you are doing is just some good, creative tracking marker placement/rigging. I've seen it all, including things like what you describe, which sounds good. Just wait until you see the rest of the tutorials in this series! 😂
@@MatthewMerkovich Good to know. :) I'm really looking forward to see what your next tutorial delves into. By the way, do you still have a Discord account? If so, any chance of an link/invite?
@@DerekGoveDesign Here's that Discord. It's mostly other users contributing as I have been too, too busy to really remain involved. discord.gg/pAtThY7
Who knows what will happen next indeed...
Great video Matt! do you find that rolling shutter has any kind of tangible effect on the results or are you going so slowly that even on phone cameras this isn't an issue?
Also it's a pity that the iphone 12 lidar features don't have more range! I find as you pan over walls or flat surfaces, it kind of jumps a bit and adds in steps / levels onto what should be a single plane!
I'm never moving fast enough for rolling shutter to become an issue, and because the surveys are NOT being used in a comp directly, one doesn't need to worry about any rolling shutter artifacts in the footage. In SynthEyes, you may need to do some rolling shutter calculations at solve time to get the error down, but really, it isn't an issue. Great question though.
Hello Matthew, was just wondering, have you worked before with a witness camera rigged on the hero camera, but aimed down towards the street pavement, as a way to provide the tracking data for the hero camera's otherwise shallow depth of field and cropped frame (longer lens, which is giving me less points for trackers). I am using a Insta360 Leica 4k module that is sharp and deep. I plan on using its footage to solve the camera tracking. I am attaching it rigidly on the hero camera at an angle, downwards towards the area of the location that has the most markers, but always on the same axis as the hero camera's sensor. What do you think of this technique?
Hi, Rafael. Your method may work, but then you end up with the same issues one has with motion control: no point cloud, or in this case, no **meaningful** point cloud. That's my first thought. Second, if you are working on a big budget studio production, there are better ways to acquire more useful data sets to help with 3D tracking, LIDAR for one. On your own productions, where **you** are the camera department, I'd definitely shoot a set survey. The latest version of SynthEyes makes using the data even easier, which I will cover in part 3 of this series ... when I can make the time to get to it.
@@MatthewMerkovich Hi there Matthew, thanks for the quick response. Ok, so capture set survey with iphone or Insta360 One R for sharpness and depthness and use point cloud, check. But to acquire the camera motion, wouldn't the "witness camera" give me sharp data that a super shallow and blurred, tightly framed cine camera would not be able to?
@@___x__x_r___xa__x_____f______ Again, how do you align that witness camera data to your photography? Seed path mode? It really wouldn't help me if **I** got that witness camera footage. (And what do you do about gen lock, by the way?)
I deal with whole sequences of shots in movies or TV shows and this approach is a non-starter on the production side, and it would be useless to me on the post side. But if you figure out how it would be helpful, please do share. I'd love to be proven wrong! ;-)
@@MatthewMerkovich so no solution then short of lidar or camera tracker like Ncam or MoSys. No hacks for something very simple? Like a very basic short movement?
@@___x__x_r___xa__x_____f______ Shoot a set survey and when doing your supervised trackers, aim for the middle of the blur. No need to reinvent the wheel on this.
Does frame rate matter here? Do you try match it?
Great question, and no, frame rate does not effect the set survey. I'll be discussing speeding up the footage in part two as a time saving method. You can even see that the footage shot by Image Video has been sped up in this tutorial.