Marvel movies are extremely compressible

[YaBoyShredderson]

Has anyone else noticed this? Why might this be? I have a few assumptions, which ill get to in a moment. I just encoded ant man and the wasp, the 4k bluray. Crf 18 hevc, slow, main10 high profile level 5.1. It came out to just under 5gbs, just the hevc stream. When i was doing my testing to see what settings i want with 4k blurays, i noticed that endgame compressed really well too, i think to like 9% of its original size iirc. There isnt anything that stands out to me as to why these movies are so compressible, but if i had to guess, its probably the amount of cgi. There essentially just extremely advancded, photo realistic animation at this point, meaning theres basically no noise. Am i correct in that?

[takla]

Because they have next to no sharpness to them. Everything is very flat. Also, ALL Marvel movies (with Avengers) on UHD Bluray were simply upscaled with bilinear filtering. They literally do not exist in native 4K (not even as master studio files).

You will get a worse image experience (ignoring HDR) using the fake 4K blurays then with the normal bluray if you use a media player that lets you configure the scale filter. MPV for example. Or MPC+madVR.

[cogman]

Lots of reasons for this.

A lot of the movies are filled with dark/pitch black scenes. That helps a lot. (Think of how many space scenes are in GotG, for example).

Many of the scenes are pure CGI, which are usually really compressible.

Very low noise (as you've noticed).

[SquallMX]

x265 is very good with upsampled content, all Marvel movies are 2K, additionally, they are filmed using top-notch digital cameras in perfect lighting conditions which equals little to no noise/grain.

[RanmaCanada]

I've said this many times, HDR is highly compressable. As we move to more and more digital only "filming" there will be no grain or noise and thus more and more movies will be able to be shrunk down to save space on your NAS while the originals are safely tucked away in your viewing room, or in a rubbermaid under your bed. The most difficult part is the time it takes to encode these.

[Boulder]

One reason in addition to the non-grainy video is that the non-graded HDR image is so flat that it confuses the encoder. It means that you should test how it looks, the numbersmay deceive you. With HDR sources, I use CRF 14 while with SDR it's 18.

[FranceBB]

Originally Posted by takla

You will get a worse image experience (ignoring HDR) using the fake 4K blurays then with the normal bluray

Yes and no, but I'm more prone to "no".
The original master delivered to theaters of the movie is a Motion JPEG 2000 250 Mbit/s 23.976p 4:4:4 12bit in XYZ, so, assuming that it's how they checked it and how it was supposed to look like, it makes a lot of sense to go to H.265 rather than H.264. You're right in the sense that it might have not been shot in 4K and that it has been upscaled, but it was shot in Log-C with an Arri Alexa, which offers several stops/nits and high bit depth; hence, assuming that it has been edited as such ("assuming" 'cause not even us working in TV know what they do as we only get the final product), targeting HEVC H.265 BD 1000 nits offers an homomorphism and preserves all the stops recorded by the camera, while 10bit planar preserves way more frequencies than going to 8bit (H.264 BD). Unfortunately, there's no public release that is done in 4:4:4 nor in XYZ as the best the public can get is 4:2:0 Type 2 YUV in an H.265 HEVC BD, but still, it's way better than the transformation they do when they go down to BT709 with a function similar to a Sigmoid with a knee on high frequencies to match the 100 nits offered by SDR + Manual adjustments + Colorist's view. Last but not least, Marvel's movies are full of GFX and, as such, they have plenty of "flat" surfaces, which is a good thing for the encoder as it can recognize the blocks and macroblocks that make up a surface and almost perfectly recognize them with motion-compensation, hence saving bits.

Originally Posted by Boulder

One reason in addition to the non-grainy video is that the non-graded HDR image is so flat that it confuses the encoder. It means that you should test how it looks, the numbersmay deceive you. With HDR sources, I use CRF 14 while with SDR it's 18.

This is also true.
Flat images which have purely logarithmic curves are much "easier" to encode for an encoder.

[Blue_MiSfit]

Actually, blockbusters are almost always shot in 4K or higher (5.6k or 8k etc). It's the VFX that are rendered at 2k due to compute requirements, thus the movie is generally mastered in 2k.

[benwaggoner]

Originally Posted by RanmaCanada

I've said this many times, HDR is highly compressable. As we move to more and more digital only "filming" there will be no grain or noise and thus more and more movies will be able to be shrunk down to save space on your NAS while the originals are safely tucked away in your viewing room, or in a rubbermaid under your bed. The most difficult part is the time it takes to encode these.

If only creators would stop adding grain to digital productions! Some like the film look. Some like that adding grain makes VFX blend in better. And a surprising number of filmmakers have an intense atavistic emotional attachment to shooting with actual made-with-chemistry film.