CORRECTION: don’t flip the image for the reflection. This was an over site. In reality, you would need to film the other side of your body. Not sure how to do that practically. I still think the reflection helps sell it regardless.
If I were you, I would overexpose the backplate slightly (in post since you're shooting Red) in order to blend the two plates better together. In your test the dynamic range between inside and outside the car feels like too many stops of dynamic range. You'd be surprised how much these "mistakes" helps sell the illusion. Furthermore be aware of how the bokehs change if you're shooting out of focus plate, and make it even more out of frame in post.
idea for next time. Take a drive with your main camera positioned about where it's going to be and shoot some reference footage so you can better position your camera for your street footage to line up more naturally when you composite it in. That car passing looks like it's sitting way too high compared to your car. But this is a great method. I definitely want to give it a try.
This a fantastic breakdown of your testing process; it's SOOOO important to do these sort of practical experiments - thank you for sharing! Two things I've given a lot of thought to regarding Poor Man's Process (it always throws me off whenever I see it used), is 1) capturing plates, and 2) adding camara shake in post. For plates, I think it's really important to actually measure out where the camera is (or will be) when capturing your foreground footage in relation to the car, and then driving in a such a way so as to position the camera in the road where the camera would be if it were in the car. This, of course, can be super dangerous, as it'll almost always involve positioning the car between lanes and/or mounting the camera outside the car. But, the tiny perspective differences of x and y differences between, say, shooting from the passenger's lap vs. shooting a plate out the back driver's side window really stands out. And for camera shake, from a realistic perspective, if you're capturing everything for real, the camera is going to be either moving with the car, independent from the talent (if the cam is mounted), or moving more in sync with the talent and independent from the car (if the cam is being handheld). Adding to that, parallax will also take effect with how the shake affects talent, car, and background differently. For realistic camera shake for the foreground (regardless of if the cam is mounted or handheld), I recommend shaking the actual car - this'll give you the proper relationship between car and talent shake related to the camera. And for post shake for the plate, I think using the plate in 3D space, and adding the shake to the virtual camera (perhaps even tracked to the live shaking), with a more realistic 3D distance between foreground and background elements, should go a very long way to keeping that movement in sync and much more natural. Thanks for humoring my geeking out :) And CONGRATS on the feature!!!
This looks great! Just make sure to make the windows brighter than you (the subject) as the light source in real life will almost always be brighter (at least a portion of the windows) than the subject.
I kind of want to give this a try. I like the results of this, it did feel a little bit off but I'm not sure what about it that it is. Might be the interior was too well lit, or the background a little too blurry. Beats me. It's a fun test regardless and you're great at what you do.
Wouldn't it be best to have a smooth plate and a smooth recording of the driver and then add the jitter in post? Then the jitter would match both the outside and the inside as it would if you were to shoot it for real?
I think I'm going to try this out. I manage some studios and can use them when they aren't booked, plus I have a warehouse full of G&E including large blue screens. Sounds like a fun experiment to do. I've been thinking about doing a short that has a contained location and a car definitely fits this bill; Thanks for the idea.
From the start I saw the fake reflection on the window and I knew something was off. The reflection doesn't have to be mirrored up. It's not the way it works. actually you should have duplicated you arm, stretched up a little and put it slightly above on the mirror.
Love your channel. Feels like you shot 90% looking out of the car for the plate, but the angle on the subject isn’t the same angle. Is this an optical illusion because of the shape of the care window?
11:22 fascinating! I would have gone with a hard light, bouncing it off the windshield as my key light. Then a few lights to add fill inside the car, one pulsating and one consistent. That’s just speculation of course, but I think I’ll give it a try sometime. I’ve observed that in cars you’re dealing with a lot of indirect soft light inside the car, with direct, hard light coming from outside the car. In my own experience, as an actor, I was in a car scene where the crew let the sun do its thing, and added fill to the interior. But that was with a uhaul tow Dolly doubling as a process trailer. Still, there’s great things to apply from that. The hard part is thinking of how the sun works and reacts in a moving car, in the moment lol.
Gotta get the perspective right man.. doesn't really match. I'd suggest taking the car out with the camera mounted in position to get a reference without any green screen and then match as close as possible when shooting the backplate.
I tried that in my garage. The green screen was about 6’ away, unfortunately some green light bouncing from the green screen contaminated inside of the car. Lesson learned. 😀
Nothing gets you out of a movie trance, like an in car green screen and all the fake movements associated with it! Very retro considering the tech we now have to work with...
Your arm doesnt flip in the reflection like you comp'd it. Basically it would be exactly the same but it would show the opposite side of your arm, whic the main cam cant get, so just duplicate your arm the same orientation in the reflection.