Camera tracking Background and Shadows?
-
Morning Dr.Sassi! I'm working with a basic camera tracking scene. Things have changed as I don't see the Background object.I'm new with Redshift! How do I see a test render of the 3D object with the footage background? Besides, how do I have the shadows in the alpha channel with some transparency?
I can't enclose the scene as it is too large. Is just a clip of 15 seconds walking on a road, taken with an iphone, and stopping in front of a clear space in the road, where a 3D cube will be. Very basic. Thanks for your help! -
Good Morning, and Happy Sunday, Capprim,
This seems like a simple question, but it is typically the content of a ten-part series. I try to keep it short...
I might share more than you need, as I'm not familiar with your background or that of anyone reading along.
Please keep in mind that the iPhone has a lot of things going on between lens and file. It is not always a "simple" case with iPhones.
Not always the best idea. Besides, that tiny lens might have more "options" to have an axial shift than a full frame lens, a logical thing based on the size.
Camera tracking runs along a lens distortion workflow. Here, the trouble begins when trying to do it completely in Cinema 4D. Tracking works with the original footage while using a lens distortion file. Which needs to be from the iPhone you have used, with the lens that you had set up. Any external data is useless.
The footage for the background needs to be rendered with that distortion first.Why is that? Because background footage is done in post, not in the rendering, for various reasons, light-wrap, for example, keeping the original footage untouched. Consider that a lens distortion "fix" is applied to the footage; each pixel is moved, most likely in a nonpixel distance, meaning it gets mixed with the neighbors. Which results in you degrading a 4K pristine footage to a simple SD footage. Hence, the renderings, which are typically available in any resolution, will be adapted to match the original footage, not the other way around.
To see the result of the background, the footage needs to be undistorted, like the tracker used with the lens distortion grid, but already back in for the background. Again, this destroys the quality. I mention that even many photographers I'm aware of apply lens distortion fixes as if that were without penalty. Therefore, the suggestion to use undistorted footage for the background is for testing purposes only, and any other use is highly questionable.
The footage for the RS-Camera needs to be an image sequence, not a QuickTime file (such as an MP4 or other container format).
In the Render Settings, you will find Redshift. In Advanced mode, the AOV (Arbitrary Output Variables) section contains an AOV named Shadow. This gives you the option to define the Shadow in a composition. There are numerous questionable ideas about this on the web, such as inverting and applying it via multiplication. This is not the best way to do it. The idea to have back for no shadows means zero for that area, and anything in the Shadow is above zero to one. This can be applied to a process, even to simple tasks like a Level or Curve adjustment. Here, you can match them comfortably in terms of the density and color of a shadow. Typically, a shadow is not black, except the blacks are crunched. We are in the era of HDR video, which means a huge number of values defining the blacks to achieve a larger dynamic range, not only with extremely bright content. For me, the real quality in HDR lies in the darker areas, specifically in color fidelity. Black shadows are super rare.
If the tracking was done in Standard render, the camera might not match the needs of Redshift. Place an RS camera below the Standard camera and zero out pos and rot while the Scale is 1. Check the focal length and the Sensor size.
All steps are done, as usual, with the current version (2025.2), as I do not answer from memory; things change, and outdated information from the past can cost you time, if not longer.
In any case, for comfortable compositing of all elements, explore the Red Giant Super Comp.
Enjoy
-
Hi DrSassi! I already did the tracking and is ok. I used to do in past versions. However I’m new to redshift, so I need to check some frames how the 3D object looks on the background first, and how to render the shadow. Redshift seems to have changed many things
-
Yes, Capprim,
As mentioned, I overshare here, as we are in a forum. You said you are working with tracking, but if I just imagine that you do all these steps, I might miss some, so I feel like sharing these details. Lens distortion workflow is something that comes to mind here. Without mentioning it, things might go wrong in many ways.
The "background" has changed and is now inside the Camera.
Please note that the problem with the background is that you need to have the Render settings exactly as the footage resolution. There is an option called "Fit", but that doesn't automatically produce a Pixel to Pixel match.
You render a shadow with an object that resembles the floor, and that object needs an RS Render Tag placed on it.
See, here I made the assumption that you know about setting shadow, and now I am uncertain about it. My mistake.
All the best
-
Hi Dr.Sassi! Thanks for your reply. I never compose the footage with 3D in C4D, that has to be made in Davinci, Motion, AE, Nuke, etc But I like to have a rough quick view in C4D I'm sorry to make you write. The problem with text is that the interface of C4D, and ways to do things, change a lot, so it's difficult to find the tool in the right menu. And C4D has infinite menus. It will be good if you point me to a tutorial. There are too many and with different opinions, so at the end you waste a lot of time in youtube until you find the right one. And may be outdated! Regards
-
Hi Capprim,
Yes, things change, hence why I recreate as much as possible when I get a question. Answering from memory has shown over the past two decades that it doesn't work.
I have searched for a video that fits your scope of questions, but nothing for Cinema 4D that I found is clearer than the documentation.
https://help.maxon.net/c4d/2025/en-us/Default.htm#html/ORSCAMERA-RSCAMERAOBJECT_GROUP_BACKGROUND.html#RSCAMERAOBJECT_BACKGROUND_TYPEI can't find a question about the background object in any "Ask The Trainer", nor is there an entry here:
https://cineversity.forums.maxon.net/category/3/tutorial-suggestions
Please feel free to express your needs.My best wishes.
-
Hi Dr.Sassi! What do you think of this tutorial? That's what I plan to do with a footage very similar
https://www.youtube.com/watch?v=K0uD1BpwVHE Is there something missing? -
Hi Capprim,
Thanks for asking. if that works for you… You asked: "Is there something missing?" Is he at all using a lens distortion, or is he addressing this point? That would be the answer.
If the lens distortion is ignored, typically, objects that need to be on the floor will swim at one point, i.e., lose contact with the floor or objects.
Besides, lens distortion of any kind (Barrel, Pillow, Mustache, etc.) will change the speed of details in relation to other parts in the "image", with that fooling the tracking algorithm. So, at any time you need to have contact with the elements in the footage, this is a must-have workflow. A different case is an airplane flying through the scene that can get away with lends distortion, but the tracking itself might be difficult.
How to get a lens distortion file (.lns)?
https://cineversity.maxon.net/en/tutorials/new_in_release_17_quickstart_lens_distortion_tool
Please note there are often complex distortions that need a few more lines to be evaluated. Anything below five lines that cover the edges, sides, and middle areas is not useful; also, assuming that anything is symmetrical from a lens is not advised.If you want to know about tracking inside Cinema 4D, use this series.
https://cineversity.maxon.net/en/series/motion_tracking_object_tracking_inside_cinema_4d?tutorial=motion_tracking_object_tracking_inside_cinema_4d_introductionIf you would like to know about compositing basic qualities those shots need to have, which partly can't be done with the quick preview, use this series here:
https://cineversity.maxon.net/en/series/integration1?tutorial=1_integration_introduction_01If you want to know what is in a typical workflow, here is another series.
https://www.youtube.com/watch?v=7za0ozspxIg&list=PLOLI4wu7NPsjGPK3BGKJpmAg9WqFodFHhAll of these share much more than you ask for–Just have the cube on the end of the street. I have not seen the footage nor the lens grid, so you ask me about something I can't just answer. I don't provide shortcuts that might let you end up with wasted time and no result.
I have been doing Motion Tracking since the late '90s, and I have gotten a lot of questions about it, hence my reluctance to get sloppy here and upset anyone. This is not my MO. Thanks for understanding.All the best
-
Thanks Dr.Sassi! I'm going to make several tests. Also will use a Fx3
-
You're very welcome, Capprim,
If you use your Sony Fx3, the lensgrid that you shoot should have (as far as possible) the same focus distance set. If you have a Zoom lens, the exact Focal length is critical for the lens grid and footage taken to match. Keep the camera straight (perpendicular) to the grid.
When you use it with the iPhone, take the video format you like for that, not a photo.
I have used this one here for over a decade or so, and if you have it print out large (Long side 36"), tell the one doing it to charge you as a text file, then it might be much cheaper. Think of a larger poster size.
https://cdn.borisfx.com/borisfx/syntheyes/lensAutoGrid.pdfAs usual, I tested it with the current version of Cinema 4D (2025.2), and all works fine.
All the best
-
I'll try Thanks!
