How does one select an appropriate bitrate?

[GoodEnough]

I don't even know what it is.

I do know that it affects file size.

I tested different bitrates with sample video of me fixing a laptop.

129,000 kbps = good
10,000 kbps = good
4,000 kbps = good
2,000 kbps = just starting to pixelize
300 kbps = full pixelized

Note:
10,000 kbps in Properties = 10,000,000 in Vegas
2,000 kbps in Properties = 2,000,000 in Vegas

Stick with 4,000,000 setting bitrate.

Does this depend on the machine viewing the video?

What situations require different bitrates? When would you ever choose 300? Does it depend on what is in the video? Static interview vs. panning movement?

I just looked at the bitrate of my source video file from my phone. It has 16,000 kbps. So, my rendered video probably should not be higher than that! But, it looks like I can go down to 4,000 and potentially make it 25% of the size.

[davexnet]

As long as you don't need a specific file size, don't use a specific bit rate, use CRF (control rate factor) encoding.
CRF 18 good place to start

[GoodEnough]

Can you clarify "complexity" ? What is simple? What is complex? Movement mainly? (Zoom whiteboard has no movement. Sports cam has large movement). What is movement ? Panning of the camera?

[smrpix]

The more stuff that changes from frame to frame, the more complex the image, therefore the more data required to describe it correctly. In a nutshell.

[KarMa]

I'm not sure that Vegas supports x264 CRF or even x264 for that matter. 10-15Mbit for 1080p content is usually good enough for video content. Usually higher end professionals will go to much higher bitrates or use a non-GOP format. But since you are asking, 10-15Mbit for 1080p content is probably just fine. As mentioned, higher motion requires more bitrate to retain a given level of quality. 2-pass would also help a lot in your situation if Vegas supports it.

[_Al_]

GoodEnough - Yes, Vegas export plugins do not offer 1pass Quality encoding to H264, so bottom line, all their users are participants in a guessing contest for their proper average bitrate. Spending evenings, like you did to come up with something that is good enough.

[babygdav]

Lots of testing.

Otherwise, easy rule of thumb is to use 1x to 2x the rate common camcorders use.

https://helpguide.sony.net/cam/1440/v1/en/contents/TP0000557915.html

E.g. If you have a 1920x1080 video and want nice quality, Sony uses 28Mbps. 28, 50~Mbps and you've got a great encode.

Assuming you use a fixed bitrate.

...

Variable bitrate encodes need more passes to optimally use fewer bits (only a concern if you're trying to save space) to get the same quality.

Encoder have slider bars for this.
Look up the documentation to see what the high end is (e.g. For handbrake.fr encoder, it's 10-15, if not higher like 5-10), then adjust up and test. Usually, such good programs have optimized settings for great encodes without much work (the presets in handbrake).

Vegas video does have the ability to export to h.264/265 mp4 1 pass cbr or vbr (customize the preset to access this on the encode window popup listing codecs).

...

If you're going for just enough bits to look decent streaming/online, look at the dvd/bluray rip scene.

The bitrate and settings used are a good start.

E.g. https://www.thewebernets.com/2019/01/29/easiest-best-optimal-settings-for-handbrake-10...-january-2019/




Optimizing - picking the settings that get you the smallest possible encode that looks great takes time to test. Programs like the Intel Media analyzer, now https://vicuesoft.com/ let's you dig into the frame by frame encodings to do this. Expect to spend loss of time, like Hollywood does to make movie dvd/bluray discs.

[_Al_]

Variable bitrate encodes need more passes to optimally use fewer bits (only a concern if you're trying to save space) to get the same quality.

For constant quality--qp (using x264) or --crf (if using x264 "fake" constant quality, good enough for a human) 1pass is enough. It is also VBR.

About analyzing, problem is we can analyze 1 video but things change with other video again. Luckily if using camcorder footage, there can be a similar pattern if always shooting the same. So, so. Not movies.

Then just frame serving from Vegas and encoding , or using lossless export or almost lossless (prores, in Vegas using Intermediate template) export first, then encode outside of Vegas makes more sense if trying to be efficient. It is a mandatory if trying to be efficient basically.

[GoodEnough]

Had to look this up again.
2,000,000 bits started to pixelize.

4,000,000 is my number.

Actually, in this last video, I could still see some pixelating on that static background at 4,000,000
So, I upper it to 10,000,000.

[jagabo]

Had to look this up again.
2,000,000 bits started to pixelize.

4,000,000 is my number.

For a particular video maybe. But different videos will require different bitrates. With bitrate based encoding you get whatever quality the encoder can deliver at that bitrate. 4000 kbps may be way more than some videos need (slideshows for example) and way less than others need (grainy film, high action movies). Which is why quality based encoding was suggested earlier. With quality based encoding the encoder encoder analyzes the video and uses whatever bitrate is required to deliver the requested quality.

[abolibibelot]

Can you clarify "complexity" ? What is simple? What is complex? Movement mainly? (Zoom whiteboard has no movement. Sports cam has large movement). What is movement ? Panning of the camera?

Even a fixed shot can be "complex" from an encoding point of view : rainy background, foliage with even a slight wind, and especially digital noise, which can be very strong when the footage was shot indoor with insufficient lighting using a camera with a small sensor.

GoodEnough - Yes, Vegas export plugins do not offer 1pass Quality encoding to H264, so bottom line, all their users are participants in a guessing contest for their proper average bitrate. Spending evenings, like you did to come up with something that is good enough.

Other solution is to export as a lossless intermediate (Lagarith, MagicYUV, UT Video), then encode this with x264. (As already mentioned above by "_Al_".)

Variable bitrate encodes need more passes to optimally use fewer bits (only a concern if you're trying to save space) to get the same quality.

No, they do not.

If you're going for just enough bits to look decent streaming/online, look at the dvd/bluray rip scene.

Among the "scene" releases, there are very low bitrate encodes and there are very high bitrate encodes of the same movie / episode, so barely helping.

About analyzing, problem is we can analyze 1 video but things change with other video again. Luckily if using camcorder footage, there can be a similar pattern if always shooting the same. So, so. Not movies.

One possibility if a certain approximate file size is a requirement (for instance to fit a video on a DVD), is to extract a few seconds at regular intervals with Avisynth and SelectRangeEvery(), so as to get a short sample which is still representative enough of the whole, then run a few tests with various CRF values (and presets).
For instance SelectRangeEvery(3600,60) ouputs 2 seconds every 2 minutes for a 30 FPS video, which amounts to 1/60 of the total duration.

[Brainiac]

GoodEnough please check your inbox.

[GoodEnough]

Instead of 4mm, I've just been using 10mm, to be safe.