Capturing legacy 60i SD footage with correct cadence

[Bryan Worsley]

Originally Posted by Christopher Young

The 50i file can be downloaded here:
https://www.sendspace.com/pro/dl/nzi8sf

Yes, that's standard interlaced (50i) footage with continuous motion between the separated fields (Top Field First).

But the DV sample you posted was not. What can I say ? Again, I'd like to examine a short sample of the original XDCAM footage.

[Paul R Johnson]

May I ask a side question? Us Brits have a list of US terms that initially confuse, but they're quite old and mostly understood. Pants, hood, trunk etc, but this topic introduces me to a new one - cadence. Here it's exclusively a sound term, so I'm trying to find the correct meaning. Is it a current US term for that strange shudders, jittery we get from mangling interlaced to progressive? Or is it just a general term used for something specific. I genuinely expected it to be something to do with mismatched audio sampling rates. Finding it to be about image quality has confused me greatly. I've never ever heard the term used outside audio before, so want to be able to use it corr city in conversation.

[Christopher Young]

Paul the term cadence is used in the old country with regards to RPM in cycling. As well as in music and film. Having been a keen cyclist over there it was a term used quite frequently with regard to "pedal power." In these Covid enchanted times, I still try to maintain a cadence of 90-95 (RPM) on the stationary cycle. The older I get I find the phrase "Use it or lose it" is developing greater significance.

https://blog.wahoofitness.com/cyclin...improve-yours/

I think what you were referring to is the cadence in music which has the famous four of 1. The Authentic Cadence, 2. The Plagal Cadence, 3. The Half Cadence and 4. The Deceptive Cadence. A very different meaning to cadence in the cycling term.

There again we now have the term cadence as used with reference to film and video. Have a look at segment 2.3 of the Alchemist manual for an understanding of cadence in the video/film domain. Quite different again to the previous two examples.

https://www.grassvalley.com/docs//Ap...ng_Cadence.pdf

Sorry Bryan I can't supply any of the original XDCam material as I don't have the right to do so but I shall check it's origin now that you have pointed out your findings. I didn't give it a thought as the material was supplied in the XDCam interlace format. The latter footy example is definitely 100% interlace I know as I shot that game in 1080i.

Chris Young
Sydney

[Bryan Worsley]

Originally Posted by Christopher Young

Sorry Bryan I can't supply any of the original XDCam material as I don't have the right to do so...

That's a shame. Could you maybe upload a sample of the XDCAM HD422 1080i footage 'double rate deinterlaced' to 59.94p and encoded to x264 - as you did with the DV and 'Footy' AVCHD clips ?

Originally Posted by Christopher Young

...Basically the same quality I can get from QTGMC can be attained by using a different workflow in a much quicker simpler workflow.

What deinterlace workflow do you use, by the way ?

Edit:

Incidentally, here's the 'Footy' AVCHD clip deinterlaced with QTGMC (default settings) to 50p and encoded to x264 (CRF=18)

https://drive.google.com/file/d/1rGV...ew?usp=sharing

Your deinterlace workflow does a good job at eliminating residual interlace artifacts, but QTGMC recovers more detail:

https://imgur.com/qZ5ZB2e

After opening link, click on (+) cursor to enlarge image

[Andrew Smith]

Both do good, but I definitely see more detail in the grass in the QTGMC version.

Andrew

[Bryan Worsley]

It was the players name (Petersham) on the back the white shirt that stood out for me in that frame - the letters are better defined in the QTGMC version.

[Ron Evans]

Looking at the Mediainfo the QTGMC data rate is 25% more than Chris' data rate that should make quite a difference at these low rates. 20.8 for the QTGMC and 16.8 for Chris. I will make versions with TMPGENC at these two rates and post latter.

For added info the set up in TMPGenc MW7 is just a matter of selection on three sequential pages and then my Threadripper took 33sec to encode. So just over 1 min total time.

[Ron Evans]

Got done quicker than I thought.But I think I messed up settings. Have to go out for an hour and will reset.

[Bryan Worsley]

Originally Posted by Ron Evans

Looking at the Mediainfo the QTGMC data rate is 25% more than Chris' data rate that should make quite a difference at these low rates. 20.8 for the QTGMC and 16.8 for Chris.

Fair enough. I encoded the QTGMC x264 mp4 file in 'Constant Quality' mode at CRF=18. I see Chris encoded at CRF=20. I'll try replicating his encode settings for fair comparison.

[Ron Evans]

I do not often post links so hope this works. The first time I messed up. https://drive.google.com/drive/folde...AJ?usp=sharing

[Paul R Johnson]

Originally Posted by Christopher Young

Paul the term cadence is used in the old country with regards to RPM in cycling. As well as in music and film. Having been a keen cyclist over there it was a term used quite frequently with regard to "pedal power." In these Covid enchanted times, I still try to maintain a cadence of 90-95 (RPM) on the stationary cycle. The older I get I find the phrase "Use it or lose it" is developing greater significance.

https://blog.wahoofitness.com/cyclin...improve-yours/

I think what you were referring to is the cadence in music which has the famous four of 1. The Authentic Cadence, 2. The Plagal Cadence, 3. The Half Cadence and 4. The Deceptive Cadence. A very different meaning to cadence in the cycling term.

There again we now have the term cadence as used with reference to film and video. Have a look at segment 2.3 of the Alchemist manual for an understanding of cadence in the video/film domain. Quite different again to the previous two examples.


Chris Young
Sydney

Interesting stuff - it does appear that the common theme in all these cadence examples is material with gaps - the cycling, the video the audio and British Telecom refer to the different dialling and ringing tones in different countries as cadence too. Just a new one on me - I suspect here we're lumping judder, jitter, and all those interlacing issues into the category of video cadence, which for me, is a totally new term.

[Ron Evans]

To me the film cadence in an NTSC interlaced feed means that some frames are repeated in the 2:3 pull down sequence. It is a way of placing 24 progressive frames into a 59.94 field sequence. So to my eyes it judders because of those repeating frames. A bit like pedalling a bike and the right legs goes down twice before the left leg comes around !!

[Bryan Worsley]

Originally Posted by Bryan Worsley

Fair enough. I encoded the QTGMC x264 mp4 file in 'Constant Quality' mode at CRF=18. I see Chris encoded at CRF=20. I'll try replicating his encode settings for fair comparison.

I'm not sure what x264 front-end/profile Chris used - I use MeGUI and encoding the AVISynth QTGMC deinterlace output to x264 at CRF=20, I get much lower bitrates - 14.3Mbps at CRF=20 with default settings (i.e. Medium preset).

Anyhow, here's a side by side comparison of Chris's, Ron's and my results:

https://imgur.com/BYjh3Tp

As before, after opening the link, click on the (+) cursor to enlarge the image.

[Christopher Young]

Hi Bryan

I use a variety of front ends if you mean encoder packages. I am in total agreement with you that QTGMC is I think without a doubt the best deinterlacer I have used especially for detail and that includes many quite pricy software deinterlacers from some of the bigger software names.

For the sports files the encodes have to turn around very quickly. Common example, maybe half a dozen 80-minute games back to back during the course of day where at the end of each game the team coaches and officials are hanging around like a bad smell waiting for their MP4 files on USB sticks. As soon as the first half of a game is finished we are encoding that first half of the game during the ten-minute break. As soon as the game is over the encoding of the second half of the game starts and needs to be finished or close to being finished by the time the coaches, officials, etc get up to the camera deck.

Therein lies the problem with QTGMC. It is slow even on a fast machine compared to some of the other deinterlace scripts around. Like in some cases four times as slow with motion intensive footage like sports.

Agreed for my own encodes I will use QTGMC. The quickest neatest way I have found to use QTGMC is in the German software from Selur. They make a free encoder app that comes standard with QTGMC/AVISynth as part of its internal scripts. Much easier to teach people to use that than trying to educate novice encoders through the ffmeg, virtual dub, MeGUI, AviSynth type routes. Nothing wrong with those applications methods but far too involved involved for this type of work where a simpler quicker Yadif encode will suffice.

Below is a similar frame from the footy using QTGMC in 'very fast' mode using "Hybrid" which is the software from Selur.
Their website is:

Welcome to the home of Hybrid | Hybrid

The Hybrid encode config with "x264 preset veryfast"

"x264 --preset veryfast --pass 1 --bitrate 20000 --profile high --level 4.2 --direct auto --b-adapt 0 --sync-lookahead 18 --qcomp 0.5 --rc-lookahead 40 --qpmax 81 --aq-mode 0 --sar 1:1 --qpfile GENERATED_QP_FILE --non-deterministic --range tv --stats "C:\Users\cyvideo\AppData\Local\Temp\Hybrid QTGMC Interlace 01.stats" --demuxer raw --input-res 1920x1080 --input-csp i420 --input-range tv --input-depth 8 --fps 50/1 --output-depth 8 --output NUL -

x264 --preset veryfast --pass 2 --bitrate 20000 --profile high --level 4.2 --ref 3 --direct auto --b-adapt 0 --sync-lookahead 18 --qcomp 0.5 --rc-lookahead 40 --qpmax 81 --partitions i4x4,p8x8,b8x8 --no-fast-pskip --subme 5 --aq-mode 0 --vbv-maxrate 62500 --vbv-bufsize 78125 --sar 1:1 --qpfile GENERATED_QP_FILE --non-deterministic --range tv --colormatrix bt709 --stats "C:\Users\cyvideo\AppData\Local\Temp\Hybrid QTGMC Interlace 01.stats" --demuxer raw --input-res 1920x1080 --input-csp i420 --input-range tv --input-depth 8 --fps 50/1 --output-depth 8 --output "C:\Users\cyvideo\AppData\Local\Temp\Hybrid QTGMC Interlace 01.264" -"

The QTGMC de-interlaced clip out of Hybrid is here:

https://www.sendspace.com/pro/dl/4l1xhs

Chris Young

[Ron Evans]

Hi Bryan good comparison. Not being too picky but both yours and Chris frame appears to be a frame after those of mine. Interesting that Chris's and mine do not look too different while the QTGMC appears to have more contrast, something I can alter in the TmPGenc encode too but have not really played with optimizing. I have uploaded another encode with just a little sharpening filter on the encode, same drive folder. This is not something I do so have not played with all the available filters so have no idea how much I could improve the encode. Encode time was now 40 seconds.