Here is a very short zombie scene I made following Film Sensei's animated Mixamo tutorial. Thanks Jay!
Très bon! Very good!
Very nice indeed ++
Good job marrying the live action and the animation into the same shot. The security camera overlay really puts the two together.
Thanks a lot! The security camera overlay is also a good mean to hide all the default in the original footage :-)
@SpaceEye one thought I had which was that maybe you should unsharpen (blur) the computer cgi a bit, it looks a bit crisp even under the security camera effect. Let the eye fill in the gaps and I think it will up the quality of believably, well at least until the final shot.
That's a good point and one of my issue: all my 3D models are in 3D unrolled mode, this I cannot apply any effect on them. I tried using a grade layer but of course it will also apply on the video footage below.
I also tried to work my 3D models in 3D 'non' unrolled but I'm struggling with the shadow projection and the camera placement (in this case sometimes a part of the 3d model is not rendered).
So all of that said, how do you guys work with 3D model? Unrolled? How do you apply effects on them?
You cannot apply effects to 3D unrolled. But you can affect the lighting of a 3D unrolled object in your master scene.
Or you can create a comp shot for your object, add lights there to get it the way you want then bring that into another comp as 3D unrolled, go to Properties and check Promote Lights.
Ah! (that's a French stupefaction mark)
I had never heard of the promoting light property in embedded composite shots. I guess that it was to high level for me during my last reading of the HF user manual.
Thanks for pointing that out, I made a couple of tests and it's pretty interesting. Even though I'm not sure of how it really works.
For instance I created an embedded CS with my 3D models & lights & shadow catchers. When I tick the "promote light" option in the main CS it looks great as if everything was into a single CS. But... I tried to add a grade layer with the blur effect in the embedded CS. It applies well on the 3D models and the shadows, but it seems also to be promoted to the main CS since the blur also apply on my video footage... Is it the expected behavior?Here is a sample project with the blur effect "promoting" to the main CS: https://1drv.ms/u/s!AkN-MiCixbTCl80FUhxSVOc2TeRJkg?e=0gykFt
Anyway, I'll try to learn more about it, thanks a lot for highlighting this to me.
@SpaceEye I will take a look at your project when I get back to my workstation, but the promote light thing does have one quirk in that it is cumulative. If you bring in multiple copies of the same comp shot and turn on promote lights for all of them, the lights add one on top of each other. So if you have 3 copies of something you only need 1 set of lights to light all 3.
I did a tutorial one time on reusing comps for objects, lighting, etc. Go in to about 9 Min. and that is where I talk about bringing in multiple comps and how the lighting is cumulative.
THanks for the tutorial, you made a massive work explaining all that. The more I learn, the more I feel like Jon Snow (I know nothing...). I also read another thread in the forum with great explanations from Triem23.
Still, I have no clue on how to apply effects on a 3D models. My main issue is that when I use an embedded CS "unrolled", any grade layer within this embedded CS is propagated to the main CS, and affect the original footage. Let's say I want to blur my 3D model, I have the blur effect applied on my video footage.
Let me detail my problem :
What is working : I embed my 3D model (3D unrolled) and lights in a CS. I apply a grade layer. I use this CS in another CS *as a 3D plane*. Everything is great, lights, etc.
What is not working : I embed my 3D model (3D unrolled) and lights in a CS. I create a shadow catcher (white plane, receive shadow, blend multiply) and set it up. I apply a grade layer.* If I use this CS in another CS *as a 3D plane*, the shadow catcher is black since I guess the blend applies to nothing.* If I use this CS in another CS *as a 3D unrolled 3D*, the shadow catcher plane is nice, *but* the grade layer applies to my original footage also.
I'm not sure if I'm clear enough but this sample project (https://1drv.ms/u/s!AkN-MiCixbTCl80FUhxSVOc2TeRJkg?e=0gykFt) tries to explain what I want to achieve. So far, I know that I don't understand something but I don't know what. I'll watch again your tutorial since I guess the answer might be there.
@SpaceEye Took me a while to get to your sample file. Looks like you've got your shadow to come over. Is this what you had in mind (less the Grade layer Glow)? http://lcprod.com/hf/shadow-test/shadow-test.png You threw me for a loop until I realize you had renamed "3D Model with Lights and Shadows" to "Composite Shot 20". Actually you can turn off layer 3. It is doing nothing here. It looks like you successfully brought the shadow plane over within the 3D unrolled comp.
Here's the deal on the 3D unrolled comp shots. You can only really unroll one 3D object in the original comp and bring it across. So if you have other 3D unrolled objects in the same comp then they won't occlude correctly.
In the case of the spaceship model - you Can bring lights over with the 3D unrolled comp. As long as I only have one ship in the original comp with all the the lights placed, then that will come over as a 3D unrolled comp and work correctly embedded in the next comp. But if I want to add a second 3D object in with the original comp then things go wonky. This technique is somewhat limited.
The only possibility I see is what Triem23 is talking about in this thread https://fxhome.com/forum/discussion/50703/compositing-3d-models/p1 where he talks about putting "Multiple Models in a Single Layer" - something I haven't really experimented with yet. I assume that this would require going back to your 3D program such as Blender and add separate groups there.
A good example is an aircraft where the landing gear, propellers, etc. are in different groups within the same saved object. Each can be animated individually after importing the parent object into HF. Basically you wind up with sub-objects under the same HF imported object. Then 3D unrolled might work for the whole smack. Just haven't tried it yet.
you're right, Layer3 is totally useless, it was an artefact of a previous test. So far the unrolled CS works as I expect (minus the grade layer issue). I guess I'll stick to that method, it seems that unrolled 3D objects are much more easy to handle and work with that 3D planes.
Thanks for your investigation and help, I'll go back again to your tutorial and to Triem23's discussion :-)