I am not sure I understand. I was under the impression that when we compressed files they became smaller. Now you are telling me to uncompress them?. I am ok to try but I don't know DnXHD nor Cineform where can I get this?
Ing.
Do some Googling on "mezzanine codecs".
Some quick background for you:
- h264 and h265 (aka, HEVC) are heavily compressed codecs, in that the image data for each frame is "compressed", something akin (in concept) to compressing a RAW image into a JPEG.
- A codec like Cineform or DNxHD are "mezzanine codecs", in that, while they still compress each frame of video, they do so in such a way that the loss/effect on the video is extremely minimal.
- Mezzanine codecs will produce files that are 5x-7x larger (or more!) than the original h264/h265 files, so they are NOT for archiving or "keeping" your footage, they are solely used as "intermediaries" during the editing and color grading process.
Why should you care out mezzanine codecs:
- When you're editing in Resolve for example... each time you scrub to a part of the video, your computer must do work to "decompress" the h264/h265. Then it must do work to apply whatever grades and effects you're working on, then it needs to render that to your preview. Mezzanine codecs like Cineform or DNxHD, while still "compressed", are significantly easier for your computer to "decompress" when working with them in Resolve (or any other grading/editing software). So basically... when you use a mezzanine file as your source, your computer can spend significantly less time decoding the source media, which allows it to spend much more time on things like applying your grades/effects to the preview.
- This also carries over to your workflow. If you're grading in Resolve, then want to move your graded content into Premier or wherever, the computer has to do this:
- - Source video in h264 -> Decompress -> Uncompressed Frame -> Apply Effects + Grades -> Compress -> graded video in h264 -> Premiere/NLE-of-your-choosing
- The "decompress" and "compress" steps above are EXTREMELY CPU intensive. Often significantly more so than the actual grading or effects aspects.
If you start w/ a video file transcoded to a mezzanine format, you save a lot of CPU cycles on the initial decompress, and if you render back to mezzanine codecs for importing into another program, you'll also save many CPU cycles there. Obviously, your final product will eventually need to be rendered back into h264/h265, but that only happens at the very end of your workflow.
- So basically, the mezzanine codecs are for carrying around a "less compressed" copy of your video, that's easier Resolve/Premier/etc to work with. The other upside, is that if you go from h264 -> resolve color grading -> h264 -> premier -> h264... you'll lose quality along the way each time you render to h264. DNxHD/Cineform allow for multiple rounds of import/export before you start to lose quality.
Reasons why you SHOULDN'T care about mezzanine codecs:
- While I started out w/ DNxHD, and eventually moved to Cineform as mezzanine codecs and used to transcode all my useful incoming footage w/ Adobe Media Encoder... I also recently realized that with my latest hardware upgrades, my desktop computer is just *barely* fast enough to handle working in Resolve directly from a h264 or h265 source. And since in many cases, I'm not doing much editing, and rendering my final deliverable from Resolve, I don't actually *need* mezzanine codecs at all. This definitely saves time, but when I'm not at home, and my desktop is unavailable, I do have to use mezzanine codecs for my laptop to even have a chance at working w/ 4K clips in Resolve. And even with that all said, it's still much smoother working in Resolve w/ a Cineform or DNxHD source file. Things like previewing grades/effects, scrubbing through footage, final render, etc... are significantly faster than when working with the original h264/h265 files directly.
- For reference, my desktop specs are quite high-end for a general user, and sort of medium-end for a professional video editor.
- - Specs are: 4790K "Haswell" @ 4.6 GHz, 32GB of RAM, Samsung 850 Pro SSD as scratch drive, and a GTX 1080 for GPU.
~~~
So I hope that all sheds some light on the reason for using mezzanine codecs!
And to more directly answer your question... yes, this use of mezzanine codecs makes the files "bigger" in terms of file size, but more "digestible" in terms of what your computer can handle. Your computer probably doesn't care so much about the "file size", it cares about how much CPU work goes into decoding that file. You, as a human, have conflated "file size" with "file complexity", which is NOT the way your computer sees it at all.
Also, the reason your GoPro footage is smaller in file size (and probably easier for your computer to handle) is that the bit rate is lower.
A video file is defined as having:
- A container format -> Typically MP4 for consumer equipment (like GoPro or Phantom 4)
- A video codec format -> Either h264 or h265/HEVC
- A frame rate -> 24 fps, 30 fps, or 60 fps generally
- A bit rate -> 100 mbit/s for
P4P 4K, 60 mbit/s for GoPro 4K or P4/P3P 4K
The primary property which determines the "file size" is the bit rate.
For example... I can make a 1 minute 4K resolution file (@ 30 fps, w/ h264 in MP4), with super low bit rate, than is only a few megabytes of file size.
I can also make a 1 minute 720p resolution file (@30 fps, w/ h264 in MP4), with a super high bit rate, that is gigabytes in file size.
The difference above is all about bit rate. Aka, how much data goes into each "frame" of the video. Don't confuse that with the number of "pixels", aka, resolution, in each "frame" of video.
Hopefully that makes some sense and gives you a bit of a crash course into video files!
- Sam