A couple of technical observations on the solving that might help in understanding the underpinning logic. When you create the rectangle to focus the tracking, it included the person moving in the background, and there was at least track point on their head for a number of frames. The tracking of the person was probably why you had an error rate above "1", as the math is assuming that every tracked pixel is a stationary point, with only camera motion. That "Solve Error" above "1" means that based on the track points included in the math, there is a greater than "1" pixel averaged error, when trying to calculate the position and rotation of the camera in the shot. If you had added a second rectangle on the left side, the solve error might get below 1, or at least closer to it. While not necessary in your example shot, when camera tracking outdoor shots with moving clouds, or a plane flying by, it's a good practice to eliminate all those elements with a mask as well, as even a little motion in a tracked cloud will introduce the kind of track point shifts that the background character did above. One trick I've used in the past (with other 3D camera track software) is add contrast to the footage temporarily just so that the points are easier for the software to discern. This also works with 2D tracks, like the classic Position/Scale/Rotation tracking in After Effects. Although I haven't actually tried the following method of exclusion (because the tracker is pretty solid), it would be interesting to precomp a depth matte to exclude the person in the foreground, along with a simple rectangle to exclude the background person. Sorry about the nerding out. :)
What I love here is that you don't state that everything you say will be 100% of the worlds knowledge. Nobody knows everything, but your videos certainly are some of the best, showing real world examples (for example the person in the shot instead of some super steady and "perfect" shot. I love it
Magic mask in fusion is pretty easy but I have found it hiccup quite a bit, and can be really demanding on the system. but really all you needed was a magic mask out from the Video input, select the subject to isolate, then run it back into merge with the original video input, as long as you use multi merge or a few merge nodes and get the 3 things (video, plan, subject) ordered right, you will have it 'easy'
Thanks for the Video, was just what I needed. Also if you want to use the Magic-Mask directly in Fusion I would add the Video clip as a node after the VFX and then mask so ot is just like layering in the Edit page.
Doing whole thing in Fusion page would be definitely simpler for later adjustments, magic mask works exactly the same way in fusion as it does in color
really great tutorial, would of been nice if you could of made the clip and plane image available to download so we can follow you're tutorial step by step :)
hahaha nice tutorial. Like others here, I love your sincerity and although I love video editing, in my case I don't think I have the time and intelligence to learn all the technical aspects involved.
For the step where i have to add the camera tracker, i can't because it requires Davinci Resolve Studio version so i can use camera tracker. I don't meant that this not useful but i can't find any way how to make cgi without using camera tracker.
solve makes DaVinci Resolve do complex math using vector 3 to create a movement to the camera. if all the movement align perfectly to the camera then solve error will be 0. but that is vertually impossible so we look for a solve error of 0.3 - 1 for smooth. and 0.3 - 3 for normal tracking
magic mask is easier and far better in fusion, you can cache it to disk as well once you have a good mask so there are no issues and the mask is locked, use it like the rectangle node and you can remove him from the scene skipping the process of using the rectangles.
Nice tutorial!!, I do it all in fusion page, just duplicate the original clip in fusion, magic mask it and connect the magic mask node to the mask input of the airplane, then in matte settings in the magic mask node, check the box that says invert matte and you are done!. What bothers me the most about magic mask is that every 30 min I have to redo all my magic mask cuz they lost track :(.
my friend, i used magic mask to have a new background in my video. A person is dancing but since I put a new background the shadows are gone. So it looks more or less a fake background and it doesn't look like its a real background....any idea how to add a realistic shadow to make him blend in more with the background...thanx..
Mine gave this error when saving the track: Solve failed: not enough tracks Mathematically, there must be at least 8 tracks in common between consecutive frames to reconstruct the camera movement. ln practice, you want a lot more than 8 (typically 150ish) for a robust and accurate solution. You requested to solve the range [3229.0, 3279.0] and there were only 2 tracks in common between frames 3229.0 and 3230.0
5:27 does anyone know why images dragged into Fusion like this come out so tiny? Mine comes out so small that even Transform can't enlarge it enough. Yet the image is 1920 x 1080 like the timeline.
Paid unfortunately. Magic mask is only in the studio version! I don’t know about the camera tracking off the top of my head. There’s a lot of work arounds to be able to get a similar effect in the free version, but most of the time you’re going to be manually keyframing things for your masks and stuff. Could be a major pain depending on the kinda person you are.
Joris, i'm just thinking. Can it work if you magic mask out the guy in red first? So basically you "remove" him from the background. And then the camera tracker? And it would effect only the background? BUT! Thank you for this video. I'll try it to make some Sci-Fi thing. 😳
It's a small price to pay for the quality. We're lucky that it's as cheap as it is honestly. I've been involved in film and TV editing for years. The price of Davinci is amazing. It should cost 1,000 bucks for all that you can do on there. A few hundred is a great price.
Its way too complicated: Camera Tracking in Fusion, the node tree it produces, how to place objects in 3D space. That should be much more intuitive and handled by AI.
It's not too complicated. There is satisfaction - I could even say pride - in using tools to make something using your brain, vs the most lazy way possible by issuing language commands. It makes you no smarter nor challenges your natural curiosity.
A couple of technical observations on the solving that might help in understanding the underpinning logic.
When you create the rectangle to focus the tracking, it included the person moving in the background, and there was at least track point on their head for a number of frames.
The tracking of the person was probably why you had an error rate above "1", as the math is assuming that every tracked pixel is a stationary point, with only camera motion.
That "Solve Error" above "1" means that based on the track points included in the math, there is a greater than "1" pixel averaged error, when trying to calculate the position and rotation of the camera in the shot.
If you had added a second rectangle on the left side, the solve error might get below 1, or at least closer to it.
While not necessary in your example shot, when camera tracking outdoor shots with moving clouds, or a plane flying by, it's a good practice to eliminate all those elements with a mask as well, as even a little motion in a tracked cloud will introduce the kind of track point shifts that the background character did above.
One trick I've used in the past (with other 3D camera track software) is add contrast to the footage temporarily just so that the points are easier for the software to discern. This also works with 2D tracks, like the classic Position/Scale/Rotation tracking in After Effects.
Although I haven't actually tried the following method of exclusion (because the tracker is pretty solid), it would be interesting to precomp a depth matte to exclude the person in the foreground, along with a simple rectangle to exclude the background person.
Sorry about the nerding out. :)
Comment pinned! Thanks! 🙏🏻💯
nerding out ❌ helping out ✔
I love the fact that you don't know everything. We are all here to learn. Love your videos. Keep up the good work! :)
Always appreciate a cool new tutorial! Very easy to follow and great results, as usual! 🔥
FINALLY!!! didn't watch the video yet but I know what it is! I asked a lot of creators to make a VFX vid like this. appreciate you makker!
What I love here is that you don't state that everything you say will be 100% of the worlds knowledge.
Nobody knows everything, but your videos certainly are some of the best, showing real world examples (for example the person in the shot instead of some super steady and "perfect" shot.
I love it
Thank you so much Joris! Very good clip! Keep it up!
Amazing how much magic happens when you have a great imagination and an awesome editing program like DaVinci Resolve 👍🏻
That camera tracker does take some practice so start with something simple like a slow drone shot or something like that...💪🏻💥
your a legend when it comes to Davinci resolve, keep up the amazing work Joris
I always enjoy your videos. But this was extra helpful, because you took time to explain trial and error. Nicely done!
Wow this is really cool.😎 Thanks for the tutorial. I like your way of explaining things.
Great tutorial! Many thanks!
This is going to make cosplay skits so much better. Thanks for the video, mate.
Joris you became one of my fav RUclipsrs, your videos are super useful. Thank you and keep up the good work!
Thanks, will do!
Thanks for this tips & the french people !!! SALUT !!! good job !!!!
Its really nice to see that now you have over 200k subs. It seems like hard working is paying off! Gratz dude.
So cool!!!! Thanks!!
Wow, thats cool, can you pls make more tutorials for vfx in davinci ?
So good! I didn't realize magic mask was just in the color tab! Looks like there's going to be a lot of volcano's in all future videos from me :)
😅💯 Also, it works a lot smoother in the color page. Less lagging. 🤷🏻♂️
Super tutorial 🎉
Glad you liked it! 🙏🏻🔥
@@JorisHermans Do you know how to fix "failed not enough tracks" please :(
That's awesome
Magic mask in fusion is pretty easy but I have found it hiccup quite a bit, and can be really demanding on the system. but really all you needed was a magic mask out from the Video input, select the subject to isolate, then run it back into merge with the original video input, as long as you use multi merge or a few merge nodes and get the 3 things (video, plan, subject) ordered right, you will have it 'easy'
Thanks for the Video, was just what I needed. Also if you want to use the Magic-Mask directly in Fusion I would add the Video clip as a node after the VFX and then mask so ot is just like layering in the Edit page.
Loving your work man 8-). Thanks for sharing !!!! You have inspired me to start share as well. 8-)
This is exactly what I need for my next video! Awesome!
Doing whole thing in Fusion page would be definitely simpler for later adjustments, magic mask works exactly the same way in fusion as it does in color
Awesome as always - cheers, Joris 🙏🤘
really great tutorial, would of been nice if you could of made the clip and plane image available to download so we can follow you're tutorial step by step :)
hahaha nice tutorial. Like others here, I love your sincerity and although I love video editing, in my case I don't think I have the time and intelligence to learn all the technical aspects involved.
Nice tutorial - would it work with free version of DaVinci Resolve?
Nope he doesn't say it which blue balled me
interesting video
Good Video. But when you use the magic tracker in Fusion Page , you must no track everything again, when you change settings.
Very cool and simple, thanks man!
For the step where i have to add the camera tracker, i can't because it requires Davinci Resolve Studio version so i can use camera tracker. I don't meant that this not useful but i can't find any way how to make cgi without using camera tracker.
solve makes DaVinci Resolve do complex math using vector 3 to create a movement to the camera. if all the movement align perfectly to the camera then solve error will be 0. but that is vertually impossible so we look for a solve error of 0.3 - 1 for smooth. and 0.3 - 3 for normal tracking
magic mask is easier and far better in fusion, you can cache it to disk as well once you have a good mask so there are no issues and the mask is locked, use it like the rectangle node and you can remove him from the scene skipping the process of using the rectangles.
Doesn't work 🤷🏻♂️😭
You are good person ❤
Nice tutorial!!, I do it all in fusion page, just duplicate the original clip in fusion, magic mask it and connect the magic mask node to the mask input of the airplane, then in matte settings in the magic mask node, check the box that says invert matte and you are done!. What bothers me the most about magic mask is that every 30 min I have to redo all my magic mask cuz they lost track :(.
Insane free software, gonna Leave After effect from now 😂
my friend, i used magic mask to have a new background in my video. A person is dancing but since I put a new background the shadows are gone. So it looks more or less a fake background and it doesn't look like its a real background....any idea how to add a realistic shadow to make him blend in more with the background...thanx..
Mine gave this error when saving the track:
Solve failed: not enough tracks
Mathematically, there must be at least 8 tracks in common
between consecutive frames to reconstruct the camera
movement. ln practice, you want a lot more than 8 (typically
150ish) for a robust and accurate solution. You requested
to solve the range [3229.0, 3279.0] and there were only 2
tracks in common between frames 3229.0 and 3230.0
I think Depth Map instead of MagicMask will work too
You can do the magic mask direct in fusion. For me it‘s often easier to use…
solve failed ; not enough tracks...how to fix
is het mogelijk in de gratis versie
Great tutorial. I keep vetting an error message of "openvr: UNABLE TO INIT VR". This happens when I add second media and image plane. Any help ?
I wish we had the midia to follow along
Is this magic mask work on free version??
paid version only. need to purchase studio license
HY BRO CAN WE RECORD VOICE IN CAPCUT PC
THATS GOOD OR NOT PLEASE GUIDE US i am yors little subscrier
How to shoot in RAW with camera
5:27 does anyone know why images dragged into Fusion like this come out so tiny? Mine comes out so small that even Transform can't enlarge it enough. Yet the image is 1920 x 1080 like the timeline.
is this free or paid?
Paid unfortunately. Magic mask is only in the studio version! I don’t know about the camera tracking off the top of my head.
There’s a lot of work arounds to be able to get a similar effect in the free version, but most of the time you’re going to be manually keyframing things for your masks and stuff. Could be a major pain depending on the kinda person you are.
@@alukeofalltradescam tracker is a studio version feature as well
Legends know how to get that for free 💀
@@animeshorts4126💀💀
@@animeshorts4126? :)
are thesein free version
I was just wondering the same!
I still don’t get it 🤔..
Nice video! But its called VFX, Special Effects are practical
No
Joris, i'm just thinking. Can it work if you magic mask out the guy in red first? So basically you "remove" him from the background. And then the camera tracker? And it would effect only the background?
BUT! Thank you for this video. I'll try it to make some Sci-Fi thing. 😳
Yeah, was trying to do it like that all in Fusion but haven't gotten it to work yet 😰
I will try to do that too! :O :D@@JorisHermans
So you have to buy Davinvi to get this :(
It's a small price to pay for the quality. We're lucky that it's as cheap as it is honestly. I've been involved in film and TV editing for years. The price of Davinci is amazing. It should cost 1,000 bucks for all that you can do on there. A few hundred is a great price.
Wdym DaVinci resolve is free
@@nexfilms97 pro version is not
Its way too complicated: Camera Tracking in Fusion, the node tree it produces, how to place objects in 3D space. That should be much more intuitive and handled by AI.
It's not too complicated. There is satisfaction - I could even say pride - in using tools to make something using your brain, vs the most lazy way possible by issuing language commands. It makes you no smarter nor challenges your natural curiosity.
you’re just lazy and don’t wanna put in the work
awesome!!!