In the FWIW dept.
I'm not all that surprised that full screen speeds up and enhances performance. Remember that the kernel now has to decide (via the OS obviously) which GPU to use. In full screen the "screen" and graphics underneath don't have to be refreshed, so the built in GPU (or gpu cause it's meant for really light drawing and refreshing) is for all intent and purpose "idling" while the GPU handles the foreground screen (i.e. full screen) It would be an interesting experiment to take one of the smaller or older Mac's and see if there's any enhancement in full s screen, I'd bet that full screen will show down the machines with only a single "lightweight" onboard "gpu".
If we really wanted to see how the onboard "gpu" and the big GPU world with each other, a movie wouldn't tell is too much. If we put in an app that does some REALLY intensive "Ray Tracing" or shading, that's where you will the the GPU really shine. The i5, i7, M series anyone's  architecture will offload the massive and entire job of calculating shading, ray traces, lighting etc. It was built exactly for those kinds of tasks, the CPU is "almost" relieved of duty, now it just has to make calls to the GPU all the nVidea and ATI are BEASTS for number crunching, their heritage began a few decades ago when Intel was still using 80X86 and 80X87 (the ...87 was the "math part". Smart companies like Sun, Apollo, Silicon Graphics kinda said, hey wait, that's cool for the math, do you think we could also make a sibling chip for graphics, YES they can, the microcode inside the modern day GPU chips is daunting.
Just my $.02

We also retested with VLC and (altough the power consumption is generally higher), we also see a reduced power consumption when running VLC in full screen.

We used METRAWin 10 together with the MetraHit Energy power meter that we use for our tests.

Awesome post and very interesting finding. I'm just curious, what application did you use to measure/chart the power while playing the video in quicktime?

You should probably test a few other popular third party video players like VLC. It would be interesting to see if this just is a product of some deferred rendering of background windows (finder, etc) or some special mojo that QT does specifically.

thanks for your feedback. I think we found out some really interesting stuff and we just have to make sure that this information finds its readers. no big headlines without good content!

Just came here to comment that please don't move to clickbait titles. Your site is amazing as is.

I assume they're likely pulling the same trick they do when showing a video fullscreen on iOS: they reduce the screen brightness (saving a bunch of power) and apply something like a gamma correction so that the brightness the user perceives is about the same.

This was mentioned in an iOS 'optimise your app's battery usage' session a year or two ago.

Perhaps the hardware decoder only kicks in when the video is played in full screen mode? That's highly unlikely, but still a possibility.

App nap, perhaps? It can slow down the timers on anything not shown to the user, so maybe fullscreen video allows more system processes to go in nap state.

Perhaps its because full screen can modify and limit the FPS and windowed mode works at top fps possible?

Seen similar difference on gaming laptops, game full screen is limited to 60fps, same game windowed runs at 110fps, and obviously takes more power.

Very interesting...

I bought one last year and it lasted 2 hours on every test, just running Youtube.. I think the screen brightness was in the middle. It was so bad I had to return it.

Just wondering, the test shown above, it lasts only 1 hour when the brightness is set to maximum, am I reading correctly? If so, what screen brightness setting was the test set on? Would appreciate this clarification, as I would really like to buy a new MBP Touch Bar.

Thank you!

Do you know if this behavior is exclusive to quicktime, or do you also see the same results in full screen mode in third-party apps like VLC?

Do you think that the reason for the lower power usage when in full screen is showing a difference in the way that MacOS handles background tasks during full screen applications?