AVCHD 1080/60i 1080p bit rates

[iL1fe]

Learning more on video...

Now understand primary differences between the 60i vs 60p. Fields vs full frames.

Now trying to understand it with bit rate. The AVCHD standard on many consumer cameras has got me confused.

The PS for 1080/p60 uses 28 Mbps. So 60 Full frames. So for 60 frames in 1 sec it uses ~28 Mbps.
The FX for 1080/i60 uses 24 Mbps. So 60 fields 1920x540 fields. So for 60 fields in 1 sec it uses ~24Mbps.

I don't follow the math on why 1080p is better here. There is effectively double the amount of data being required in 1 sec yet the 1080p AVCHD standard is only 4Mbps higher than the AVCHD 1080i.

Yes there is 60 frames for smoother motion yet significantly less amount of data being stored.

How I'm interpreting it, which is probably incorrect...shouldn't the AVCHD PS 1080/p60 be close to double that of the 1080/60i. So close to ~48Mbps?

Thanks for feedback.

[smrpix]

Video compression works by interpreting temporal and spatial differences between and within images and minimizing the amount of information needed to express that.

There is no 1:1 relationship between frame size and compression bitrate. Sequences with a lot of motion require a higher bitrate, sequences with low motion -- such as a talking heads -- can be compressed very efficiently.

Since 60i and 60p are rarely 100% different, it does not require 100% additional information to express them.

Consumer gear is also limited by factors such as reasonable file size and speed/heat/cost of real-time compression electronics.

[iL1fe]

Hmm...I don't follow. Most likely as I have yet to explore the complexity of compression functions.

Not a 1:1 relationship on frame size. Ok. But Frames per second. I understand 60i is 1 1920x540 field per 1/60 sec. The interlacing is effectively acting like it's compressing the video, no? It requires only x540 pixels in 1/60 sec vs progressive which requires x1080 pixels in 1/60 sec. So it appears the bit rate would need to be significantly higher.

Perhaps I'm way off being unfamiliar with compression...

[smrpix]

Perhaps I'm way off being unfamiliar with compression...

You are.

Bitrate has NOTHING to do with frame size. You can encode 320x180 15fps at 60Mbs and waste a lot of disk space, or encode 3840×2160 72fps at 1.5Mbps and get fuzzy blobs (unless its a static title.)

Between that there's a range of generally acceptable options at a sufficient rate to convincingly express the content.

[iL1fe]

You example is obvious, but it doesn't help on clarifying the question.

Feel my question is about Interlaced vs Progressive "Frames/fields per sec" relating to bit rate. Not the resolution. I'm just using STD HD 1920x1080 as the example for common cameras.

[iL1fe]

Since 60i and 60p are rarely 100% different, it does not require 100% additional information to express them.

Ok so it doesn't require double due to the compression features. But correct me here that it does requires more data as 1/60th sec 60P has the full frame where 60i only has half a frame.

[jagabo]

Yes, in general more fps requires more bitrate. But...

Interlaced encoding is inherently less efficient than progressive encoding.

Less motion per frame may require less bitrate per frame.

Different encoder chips may have different efficiencies.

Different encoder settings can lead to different efficiencies. Longer GOPs give better compression but are harder to edit. More consecutive b-frames gives better compression but lower quality. Finer motion search precision gives better quality but requires more bitrate. Larger motions search areas require more horsepower to compress but lead to better compression. Etc.

The less time an image is on the screen, and the less it's enlarged (assuming you're viewing full screen), the less noticeable artifacts are.

Clean video compresses much better than noisy video. So other components of the camera have an effect on bitrate requirements.

The bitrates chosen by camera manufacturers have as much to do with storage requirements as they do with image quality. A manufacturer doesn't want to be in the position where their camera only stores 4 hours on-camera where their competitor stores 8 hours. It's too hard to argue image quality.

[iL1fe]

Yes, in general more fps requires more bitrate. But...

Interlaced encoding is inherently less efficient than progressive encoding.

Hmm. That's interesting, 60i is less efficient to encode?

Putting compression aside for moment. It I plug in my same example above into bit rate calc:

1920x1080/p60 gives = 2,99 Gbps uncompressed bit rate
1920x1080/i60 gives = 1,49 Gbps uncompressed bit rate.

Thus showing the increased data requirement on progressive (full 60 frames sec) vs interlaced (60 fields)

Back to compression:

Just curious on how that significant difference in data is being accounted for with only an additional 4Mbps on a AVCHD/H264 cameras.

Yeah, I'm going to put this confusion on hold for moment and read up on the H264 magic.

[jagabo]

Putting compression aside for moment. It I plug in my same example above into bit rate calc:

1920x1080/p60 gives = 2,99 Gbps uncompressed bit rate
1920x1080/i60 gives = 1,49 Gbps uncompressed bit rate.

That's the bitrate of uncompressed RGB frames. It's not linearly related to compressibility.

[Dougster]

Check this article, its a good place to start.

http://en.wikipedia.org/wiki/H.264/MPEG-4_AVC

[davexnet]

I think interlaced format requires more bitrate than progressive.
If the manufacturer determined 28 mbps is optimum for progressive (60 fps), then at first glance it seems
reasonable to infer that for 60i (30 interlaced frames), 14 mbps would be OK.

However, it's not the case - since interlaced requires more, 24 is used instead of 14.

[Cornucopia]

UNCOMPRESSED, 60p is 2x the bitrate of 60i. And if EVERY frame/field were totally different from the last, you would likely need close to 2x to give equivalent Quality-per-pixel, but those are extremely rare cases. Video is almost always compressed (file sizes are too huge otherwise), and each frame is USUALLY a slight modification from previous or next frame. Fields are more inefficient, because a single pixel address point, say 580 across & 231 down, is going to have a larger timeframe within which things in motion can move/change a greater amount. That point is only different by 1/60th of a second in 60p, but that point doesn't exist in the next/previous field (1/60th sec), so it is different by 1/30th sec (its nearest temporal neighbor in the same spatial address) in 60i.

Scott

[iL1fe]

That's the bitrate of uncompressed RGB frames. It's not linearly related to compressibility.

Yeah as already stated earlier in the form my confusion is result of getting a technical break down how 60i vs 60p frame/fields work per sec, but not having a grasp on the compression applied to said frames. Gonna read up on h264 / compression.

Thanks for your feedback.

[iL1fe]

UNCOMPRESSED, 60p is 2x the bitrate of 60i. And if EVERY frame/field were totally different from the last, you would likely need close to 2x to give equivalent Quality-per-pixel, but those are extremely rare cases. Video is almost always compressed (file sizes are too huge otherwise), and each frame is USUALLY a slight modification from previous or next frame. Fields are more inefficient, because a single pixel address point, say 580 across & 231 down, is going to have a larger timeframe within which things in motion can move/change a greater amount. That point is only different by 1/60th of a second in 60p, but that point doesn't exist in the next/previous field (1/60th sec), so it is different by 1/30th sec (its nearest temporal neighbor in the same spatial address) in 60i.

Scott

Well explained. Thank you Scott.

[iL1fe]

I think interlaced format requires more bitrate than progressive.
If the manufacturer determined 28 mbps is optimum for progressive (60 fps), then at first glance it seems
reasonable to infer that for 60i (30 interlaced frames), 14 mbps would be OK.

However, it's not the case - since interlaced requires more, 24 is used instead of 14.

Ok.
Uncompress = 60p double bit rate of 60i. Compressed = 60i less efficient bit rate compared to 60p yet 60p still higher bit rate as it's full 60 frames sec. I take it this is where the different AVCHD profiles come in as my camera does do 60i in 17, 9, 5 Mbps but only 60p at 28 Mbps.

Thank you

[Dougster]

I think you're starting to get it now!

Compression is scalable for the specific need.