Z1U PAL = better quality?

[Thomas Smet]

HDV for 1080i is locked at 25 mbits/s and there is no resolution difference between 60i or 50i for HDV so shouldn't shooting with the camera switched to 50i give slightly lower compression?

Pal HDV will have 20% less frames each second so that gives 20% more bits to be used each second.

60i = 0.83 mbits/frame
50i = 1.00 mbits/frame

[Eric Gorski]

sounds like it would...

[Filip Kovcin]

exellent point - but do you know where and who to ask?

is there any FAQ connected with HDV standards? and bitrates? and pal vs ntsc in HDV?

i was once on site where HDV standard and members are mentioned - but forgot where it was - maybe there is an answer to that question?


filip

[David Newman]

The analysis is correct, 50Hz fields will have more data per field than 60Hz fields. However, there is less correlation between fields in 50 vs 60Hz (so motion searches are less efficient) and the 50 Hz system has a shorter GOP length (12 vs 15 frames) which also reduces effeciency. So in the end all is pretty much the same.

[Zack Birlew]

Ummmmmm..... guys.... there's only one version of the Z1U. That's why it's NTSC/PAL switchable ^_^;;;;;.

[Brandon Greenlee]

<<<-- Originally posted by Jack Felis : Ummmmmm..... guys.... there's only one version of the Z1U. That's why it's NTSC/PAL switchable ^_^;;;;;. -->>>

I believe they were refferring to the merits of recording in PAL (50i) versus NTSC (60i) to get improved image quality.

[Zack Birlew]

Oh, okay then. My bad.

[Thomas Smet]

I forgot about 50i only using 12 frames in the GOP. I thought shorter GOP's had higher quality? With DVD's you can use a shorter GOP to get higher quality because there are not as many P and B frames which have the most compression between the I frames. Long GOP's are used to get more on a DVD. Thats why there are experimental GOP's for DVDs with over 30 frames in a GOP to fit a lot of video on a single DVD. Or you can use a GOP of 1 frame made up of only I frames. The compression would be about the same a jpeg so you wouldn't fit very much video on a DVD. If 50i already is higher quality because of the extra bits then it would be even higher yet because of the shorter GOP. Having only 12 frames in the GOP means less P and B frames. If anything the shorter GOP should have the same result as the 15 60i GOP since the same ratio of IPB frames occur every second. I think it would look better because the 11 or 12 P and B frames have a few more bits. With 14 or 15 P and B frames each frame has to share more of the bits.

60i = 25 Mbits/s = 15 GOP = each GOP is 12.5 Mbits/s
12.5 Mbits/s / 15 frames = 0.83 Mbits per frame

50i = 25 Mbits/s = 12 GOP = each GOP is 12.0 Mbits/s or 12.5 because I do not know where the left over frame comes from for the GOP.
12.0 Mbits/s / 12 frames = 1.00 Mbits per frame

each second of 50i has only 23 P and B frames compared to 60i having 28 P and B frames. That is 5 extra in between frames we see during each second of 60i

[Greg Boston]

Very interesting points, Thomas. I would be curious to see if the shorter GOP would help with reducing motion artifacts. That is one of the biggest issues with MPEG2 compression. It seems like it should. I don't own the Z1 but if someone could shoot the same motion (perhaps using 2 cams, 1 set NTSC, 1 set PAL) it might make for an interesting comparison.

Good discussion guys!

=gb=

[Filip Kovcin]

is it possible for someone to make following experiment:

-put camera on remote head - and prepare it to pan for certain ammount of seconds
- turn camera on in 60i mode - pan and shoot
- switch camera to 50i mode - pan and shoot.
- compare it somehow (do not know how yet... it's 50i and60i) - maybe using same point or position of the camera - to analize...

just a thought

filip

[Zack Birlew]

You guys could e-mail the DVXUser.com guys and ask them do a comparison shoot like they've done with the XL2 vs. the DVX100A.

[David Newman]

<<<-- Originally posted by Thomas Smet : I forgot about 50i only using 12 frames in the GOP. I thought shorter GOP's had higher quality? >>>

It doesn't work that way. You have a fixed bit-rate so longer GOPs are more efficient -- therefore can store more video for the same bit-rate. That is why a 15GOP DVD at 8Mb/s looks great, yet an 8Mb/s M-JPEG stream of the same video would be unwatchable. Their is nothing worth observing between and 50Hz and 60Hz mode it terms of MPEG quality difference. Your choice for a shooting mode should just be concerned with the frame rate needed.

[William Gardner]

Thomas wrote:
"I thought shorter GOP's had higher quality? With DVD's you can use a shorter GOP to get higher quality because there are not as many P and B frames which have the most compression between the I frames. Long GOP's are used to get more on a DVD."

For a fixed data rate, you get better quality with a long GOP since you're taking advantage of the correlation between frames of information. That's how HDV can get HD resolution with a DV data rate: DV compresses 1 frame at a time whereas HDV takes advantage of the interframe correlation.

So, if a DVD holds a fixed amount of data, you get the longest DVD play time using large GOPs. However, if there's a cut in the middle of a GOP then the correlation will be poor and you'll get a quality hit. So if you only have a short play time, you can use more I frames (i.e., shorter GOP) and still have comparable quality at a higher bits/sec rate.

Essentially, an I frame takes more data than a P or B frame to achieve the same quality, so your calculation on "average bits per frame" isn't accurate since the 50i rate has a larger percentage of I frames than a 60i rate.

Bill

[Joonas Kiviharju]

<<<-- Originally posted by David Newman : Their is nothing worth observing between and 50Hz and 60Hz mode it terms of MPEG quality difference. Your choice for a shooting mode should just be concerned with the frame rate needed. -->>>


People seem to be making this claim alot. But think about it. If you had say 1000 fields, that you'd have to fit on 25 Mb/s stream, you'd have your GOP set at 500, and there would be very little difference between your fields making it very efficient.

Are you seriously claiming that it would still look as good as 60i MPEG individual frame quality? Or if you had 2 frames to put into that 25 Mb/s. You suggest that, it wouldn't increase the quality of those individual frames enormously?

I can't say I understand MPEG compression issues, but this would just contradict common sense. There has to be a quality difference. If you pack info on to a known space, the less info you have, the better the individual frame quality?

[David Newman]

<<<-- Originally posted by Joonas Kiviharju :
I can't say I understand MPEG compression issues, but this would just contradict common sense. There has to be a quality difference. If you pack info on to a known space, the less info you have, the better the individual frame quality? -->>>

That is the way it is. You are assuming that the 'I' frame has more quality than P or B frames, and that simply isn't the case. P frames can have 'I' macro blocks when needed, like a scene change mid-GOP, but also anywhere in the image where the motion search fails to find a good match. B frames can use any macroblock type - B, P or I. So B frames have the highest flexibility, they certainly aren't inferior to 'I' frames. A 'B' can have the quality of an I at a fraction of the bit-rate. The longer the GOP the greater the efficency, and yes the greater the quality for a particular bit-rate.