How noticable is different Bitrates?

[Statik]

Hey all, I'm just trying to grasp the concept of different bitrates for quality of movie watching experience. How would watching a movie in 1080p @ 2500mb/s compare to 4000? I personally can't see much of a difference on a 1080p laptop, but is it extremely noticeable on something like a 55" TV?

[Moonzy]

it depends on how many pixels are moving as well

as well as how many new elements are introduced

 

it's not as simple as drawing a line for the best bit rate

 

Spoiler

 

 

[Eigenvektor]

That can depend a lot on the content of the movie and how well it compresses. In most case you probably want to set the encoder to variable bitrate/constant quality, rather than a fixed bitrate. A fixed bitrate mostly makes sense if you want to achieve a certain size and quality isn't your primary goal.

[Statik]

7 minutes ago, Moonzy said:

it depends on how many pixels are moving as well

as well as how many new elements are introduced

 

it's not as simple as drawing a line for the best bit rate

 

  Reveal hidden contents

 

 

That video is awesome, but makes me realize my understanding on the topic is very minimal.

 

Would I be correct in assuming that a video with more movement/constant new elements (i.e. action, etc) would benefit from a higher bitrate than something of a more relaxed genre?

[Moonzy]

3 minutes ago, Statik said:

Would I be correct in assuming that a video with more movement/constant new elements (i.e. action, etc) would benefit from a higher bitrate than something of a more relaxed genre?

yeap

 

but it also depends on a lot of other things like CODEC used, like @Eigenvektor mentioned

 

ie, youtube used 2 different codec, depending on how big of a youtuber you are

one of them is horrible, one of them is good

here's an example

[Eigenvektor]

One thing I forgot: You can also do multi-pass encoding. For example if you use a fixed bitrate, you may be "wasting" a lot of bits on scenes that don't actually need it, while some other scenes that could use more are going to look horrible. Simply because each scene uses the same bitrate irregardless of how "busy" it is.

 

With multi-pass, the encoder can essentially get a look at the video, determine which parts could do with a lower bitrate and which parts need a higher bitrate and then attempt to achieve a usable quality on the second pass by increasing and decreasing bitrate for different parts of the video, while attempting to keep the average within your confines.

 

As I said above, this is mostly useful if you have a maximum size in mind that you need to achieve while still keeping the video of acceptable quality. It can take a lot of time to fiddle with settings and do the actual rendering. In most cases you may just want to set some acceptable quality and encoding speed and go.

[LAwLz]

Bit rate is only one out of several components which makes a video look good or bad.

A high bit rate file is not necessarily going to look better than a low bit rate file. Which format the video file is matters a lot too, as well as which encoding settings were used.

If you download something like HandBrake and play around with the settings you will see that there are A LOT of things to tweak. 

 

 

x264 is possibly the most popular encoder for H.264 (also known as AVC). What x264 does is take a video and encode it to an AVC file. x264 has over 100 different parameters you can tweak and they have an effect on how the final video will look and what size it will be. This is why we have "presets" which are carefully tuned and handpicked predefined groups of settings. Changing preset changes these 100+ settings in ways that tades speed for quality, or vice versa.

 

Here is an example for you. There are three different types of frames for AVC and an example of how different uses of those might result in different picture quality even with the same bit rate and file format:

 

Spoiler

 

• I-frame - A full image. Think of it like a JPEG file. They are big and contain all necessary info to fully render a frame in the video.
   
• P-frames - These frames are way smaller than I-frames but can look just as good. How? Because P frames says "okay, only these pixels changed color compared to the last frame, so I only contain info on these new pixels and you (the video player) have to fetch info about all the other pixels from the previous frames". So while I-frames are like JPEG files, P-frames basically partial images which reference other frames to paint a full picture. If you don't store all the info necessary to draw a full picture, you can be way smaller without any loss in quality!
   
• B-frames - B-frames are like P-frames but on steroids. While P-frames can reference unchanged pixels based on old frames, B-frames can reference information contained within FUTURE frames. So a B-frame can go "I am image number 5 in this sequence, info about 50% of my pixels can be found in frame 4, 25% of my pixels can be found in frame 6, so I only have to contain info about 25% of my pixels!". This is very, very efficient. The drawback is that this is REALLY difficult and computationally heavy during the encoding stages. Because in order for this to work the computer has to look ahead in the video file and constantly try and find the best way to optimize the video. It also means it might be wasting a bunch of resources if it figures out that a B frame can't reference something in a future frame. 

 

So two videos with the same bit rate and same codec might look different. A video without any B frames will have to waste more bits saving the same information as one video with B frames, but the video without will be way faster to encode. 

 

 

 

Then we have formats. HEVC is the successor to H.264 (AVC). It is generally considered to provide the same image quality as AVC but at ~60% of the bit rate. This means that a 600MB HEVC file will typically look as good as a 1GB H.264 file (assuming similar parameters and tradeoffs in terms of processing were made).

 

 

Then there are a whole bunch of other stuff that is relevant to final output quality, like which source you used (you can't re-encode something to look better than the input file), how you're playing the video back, what resolution the file is, etc.