Blender screen replacement & marker removal

Lately I’ve been doing more and more for 3DArtist magazine. I got to meet editor Steve Holmes at Siggraph last year, and since then, he keeps inviting me back to do more fun compositing tutorials in Blender, which of course is a pleasure.

This is one I did for them in Issue 88. I wanted to introduce people to the idea that Blender isn’t only a 3D program, but could be used in place of Nuke or After Effects for most compositing tasks. One of the simplest and most common jobs is replacing a screen on a device. Monitors, phones, tablets, televisions, this technique works for any of those. It also works for anything flat in general, like billboards, walls, floors, replacing book covers, photographs, etc.

It’s pretty basic stuff, but I like to think there’s some good techniques on display here, especially when it comes to removing the tracking markers, putting the reflection back over the image, and rotoscoping the finger back over. Basic compositing stuff for professionals, but not something you see done in Blender very much. I touch on a lot of subjects here, and don’t really go into detail on any of them. It moves pretty fast, and assumes you know your way around Blender a bit.

Since 3DArtist Online has posted the written tutorial, I figure it’s okay for me to post the video versions here. You can find the written tutorial here at 3DArtist’s site.

Assets required:

Here’s the main tutorial video. Click HERE to download this video.


And if you want to learn how to remove the markers, here’s that video. And you can download this one by clicking HERE.


    1. Sure, bolaji, let me check out how best to do that this weekend. I’ll try and have them posted by next week.

  1. There is share link, you can copy that link and paste it in your downloading software. I am on MAC and i use 4K video downloader. I just download videos with out any problem.

  2. Hi Sean,

    Your tutorials have been FANTASTIC! Total noob question for ya…I noticed when tracking, the footage is always an image sequence (rather than a movie… .mov .mp4 etc.).

    Are there technical reasons for that? (I assume it’s standard industry practice, though not sure why).

    Do clients normally provide the image sequence? Or is that on your to take the client provided footage, turn it into what works best, and then create a deliverable output file?

    Thanks for all your hard work and sharing!

    1. Hi Dan,

      Thanks for liking everything. I’m still just getting started, and haven’t really started getting into the stuff I really created the blog for, so lots more coming up.

      Yes, industry standard is working with frame sequences. Usually DPX, CIN, or EXR. We only render movies for reviews and approvals. The client delivers frame sequences, and that’s what we deliver back to them. So rarely will I post anything as a movie clip. I don’t like posting JPG sequences much, but they are MUCH smaller than 10+ bit logarithmic frame sequences, so I do that when the camera I used to shoot the footage is nothing great (my Samsung or GoPro). Eventually I’m going to shoot on real cameras for this footage, and when I do that I will offer the much better footage for download.

      Working with compressed footage is a pain, anyway. Scrubbing around a timeline doesn’t work well, because the program is constantly having to decode the codec on the fly. Frame sequences, especially small ones like JPG, are really fast to work with.

  3. Great tutorial!

    Instead of overlaying the magazine with “alpha over” and then blending the reflections back in with “screen” you could just start with the cleaned video and add the magazine with “mix->add” node. Exactly like the real IPads screen would emit light in addition to the reflections on the glass. To limit this addition to the Screen area just connect the “Plane”-output of the deform-node to the factor-input of the mix-add node.

    1. Hi Gunther,

      Thanks for the tips! Absolutely those would work. Myself, I prefer the way I showed, but your ways are certainly good techniques, and a bit faster. 🙂 If you use the Add technique, though, keep an eye out for anything going brighter than it should be. Add mode tends to lead to that rather easily. And when plugging in the alpha from the Plane Deform node into the factor, watch the edges to make sure there’s not a thin, bright outline. I’d bet you still have to dilate that plane matte 1 pixel.

  4. Hi Sean,

    Thanks so much for your training, much appreciated, have followed your tutorials for years.
    Do you know if there is anyway to get the plane track into After Effects?
    Is there away to connect and parent emptys to each corner of the plain so that the After Effects exporter can send them? Anyway thanks for your time, Hope that make sense.

Leave a Reply