This IS the Avid L2: Re: [Avid-L2] How does 29.97P 1080 get converted to 720P?

Thursday, December 17, 2015

Re: [Avid-L2] How does 29.97P 1080 get converted to 720P?

You mentioned linking, aka ama linking, instead of importing. I had curious behavior on my home Mac MC 8.3.1 NitrisDX system OS 10.8.5. In dealing with the network no longer allowing 2:3 pull down I open up a recent show I did for the network and it clearly had 2:3 pull down and I had no QC rejection. Then I exported a small section and ama linked back to the .mov SAS QT and the frames were getting blended even though I had exported proper 2:3 pulldown. I check the exported file in media info and the scan type showed interlace but when I stepped through the ama'd clip in a 59.94 project I saw blended fields on the split frames. I check source settings and made sure it was set to upper field first. I then noticed what I believe to be a newer feature the said split or merge . It was set to merge so I figured setting it to split would allow for split frames but I saw no difference when I applied that. Perhaps I then needed to refresh the motion adapter on the ama clip but I did redrag the clip into the source monitor after applying the split choice to the frame option. Nothing allowed me to see the clean 2:3 pull down. Then I did a traditional import of the file and on that I could see clean 2:3 pull down. I haven't had a chance to try this on my 6.5.4 work system yet but for some reason there seems to be a bug with the ama linking where it blends the frames even when the choice is set to split. I'm assuming the choice of merge and split is to handle split frame clips differently than true progressive ama linked clips.

---In avid-l2@yahoogroups.com, <cutandcover@...> wrote :

"What happens when I import an upper field first file into a progressive project or if it's a lower field first file are there flags in the metadata or video essence that tell the software what the files video structure is?"

It will do one of two things: blend the fields together to form frames, or throw away one field and use the other to make a frame. But really, you should not import if you can link. Link allows you to preserve everything about the source, and you decide how to commit them to progressive frames (via motion adapter).

"in reality is the video essence stored in a progressive sequential manner in the file?"

Yes.

"Wouldn't data be written in the same manner when capturing an HDSDI signal?"

No - because it's inefficient for the software subsequently to handle it this way. If SDI coming in is PsF, the data is read correctly on capture and written to form P frames.

I should say that this is mostly what happens. YMMV especially when you notice that capturing in a P project still somehow retains the ability to break the P frames apart into their interlaced origins. I'm sure there's a good and easy reason why this works, but I don't ever really get that deep into trying to figure it out, since it doesn't impair my workflow.

On Thu, Dec 17, 2015 at 12:38 AM, bigfish@... [Avid-L2] <Avid-L2@yahoogroups.com> wrote:

Okay so when I export from Avid or other software in HD the file is Upper Field First when dealing with 59.94. To me that implies that under the hood there is still some structural field nature. Obviously with interlace material things are different. When I export from Avid and reimport back I always check upper field first. When I'm in an 59.94I project. Now in a 29.97P project I see there is no longer upper field and lower field or progressive boxes in the import options. What happens when I import an upper field first file into a progressive project or if it's a lower field first file are there flags in the metadata or video essence that tell the software what the files video structure is? I have always superimposed the tape machines and HDSDI fields/Psf structure to how I thought the file essence was stored based on the way I can see both fields in a still in photo shop but in reality is the video essence stored in a progressive sequential manner in the file? It would seem easier to store the stream of video data as it comes in as a field structure in the same manner in the file to avoid the need for buffering but perhaps circuitry is so fast that isn't much of a design issue. Just like one head in a deck swipes out the information of one field then the next head swipes the next field. Wouldn't data be written in the same manner when capturing an HDSDI signal?

---In Avid-L2@yahoogroups.com, <cutandcover@...> wrote :

I agree with you - I'm saying the default mode for unattended machinery and personnel is to do it the worst way because it's the easiest. So 29.97p gets doubled into 59.94p. Simple. And bad.

If an actual quality-conscious person is doing a non-automated conversion, then yes, fields become frames and life is good. This is rare and the exception.

I understand what you mean (now) when you say PsF but you should not refer to any file as PsF, as that's incorrect. As I said before, PsF is a transfer vehicle, so a file which is P can transfer over SDI via PsF and/or be stored on an interlaced tape mechanism as PsF. But the file is P.

On Wed, Dec 16, 2015 at 5:53 PM, bigfish@... [Avid-L2] <Avid-L2@yahoogroups.com> wrote:

If my source file show master is 29.97i aka 59.94i then it would be a disservice to double the frames and would that not look like 30 pairs of blended interlace mess in 720P 60? To keep the temporal resolution of interlaced material, sports especially, wouldn't each field of the 59.94i media get mapped into a single 720P frame? Or am I not reading enough sarcasm in your post? I'm using the term field for 1080i 59.94 material and Psf when it's 29.97P or 23.976P. Is that the proper engineering speak? I hear the two interchanged all the time and I usually understand it's meaning based on the context of the conversation but I'd like to get a proper terminology nerd badge here.

---In avid-l2@yahoogroups.com, <cutandcover@...> wrote :

If you have a 29.97 interlaced OR progressive file and you hand it off to some other person/machine to convert to 720/59.94p, the frames will be doubled. Simple. And if your video contained 2:3 pulldown, it will be a nicely blended mess.

The solution is if you know the delivery is 720/59.94p, and your source is 1080/23.976p, you add the pulldown in frames (not fields) and it's nicely preserved. I do this all the time for ESPN delivery.

"For an interlace 59.94 conversion I'd have to bet that each Psf becomes a unique 720P frame thereby maintaining the smoother motion so I just figure basically the same things happens when the source is 29.97P but that's an assumption that seems to be incorrect according to my QC chat."

If YOU were doing this conversion, then yes. Otherwise, assume that whatever the WORST way a conversion can happen, it'll happen that way.

On Wed, Dec 16, 2015 at 4:44 PM, John Moore bigfish@... [Avid-L2] <Avid-L2@yahoogroups.com> wrote:

Just spent some quality QC talk with a network we are delivering to and much to my disappointment they no longer accept 2:3 pulldown. Even though they can broadcast it just fine the ancillary interwebs and international concerns are the driving force for not accepting 2:3 pulldown. I completely understand that a 2:3 pull down doesn't technically fit in a 29.97P but still this makes me throw up in my mouth just a little bit.

So I asked the question to the Network QC folks. So if you broadcast in 720P how do my 29.97/sec full frames that are packed in a Psf structure of 2 Psfs per frame get distributed across the 59.94/sec 720P frames. I was told that the 29.97 full frames were doubled up in the process. They didn't seem completely certain about that so I'm looking for some clarification.

In my mind the simplest way the 29.97P to 720P conversion would happen is for each Psf to become one of the 720P frames. Now I'm curious how the conversion of 540 lines per Psf to 720 lines in 720P happens. But according to the chat I just had with Network QC what they say happens is the two Psf fields are combined to a full frame and that is the info that is packed into the 720P frame and they repeat the same frame so the 30 original frames are doubled into 60 frames.

For an interlace 59.94 conversion I'd have to bet that each Psf becomes a unique 720P frame thereby maintaining the smoother motion so I just figure basically the same things happens when the source is 29.97P but that's an assumption that seems to be incorrect according to my QC chat. Can someone beat me over the head with split frames until I can let this go? ;-)

John Moore Barking Trout Productions Studio City, CA bigfish@...