Dvd Authoring and projection

angevin1

I use Adobe- everything. CS5.5

I prefer Blu-ray but many of my clients and performers have only Dvd.

I thought I had worked my way thru the Dvd jaggies by doing some test encodes from Encore using Media Encoder. And In fact I have improved my results.BUT the other night in a local facility I saw one of my Dvd's being played and it was awful. Of course this was a full size screen with projector AND my old recipe for transcoding.That was what got me back in the encoding mode to apply my new-er recipes to the old disc they showed.

What part does the display/projection play in playback of media? For instance on my Monitor inside Encore, after transcoding I can view the files, playing and pausing as needed. I can also Zoom it to 200%, which I do to judge the final encode.

Then I go to play the resulting Dvd on my Tv and I see all sorts of things (I use a Zoom funx on the TV too, but it only goes to 136%).


Such as:
1. a ghost image-outline of bodies depending on light in the scene.

2. Jaggies or pixelation that simply was not there on my Monitor in Encore.


So what role does projection/Display play?

Any other ideas?

My sample test footage is 3 copies: Mpeg-2Dvd, and H 264-720p, H 264-1080p When I bring it into Encore. the latter two requiring transcode. I see no difference when I apply my recipe to the H 264's. They look as good as the Mpeg-2 Dvd transcode. I Author them all at once as different Chapters to make it easier to compare.

Dan7312

Just some thoughts the might help...

Have you tried out one of your Blue-Ray disks on your tv and does it show any artifacts?

I've seen strange artifacts sometimes when youtube transcodes the simple wmv files I make for local computer playback. From the looks of what I see it seems like some of the frames are played out of order because the pieces are there but they jump around a little and it loses sync with the audio. If I instead generate the wmv file directly from the sequence in Premiere using the youtube preset it plays much better, though still not as good as the original.

My wmv's are 1024x768 and of course youtube is some HD variant aspect ratio. I've kind of assumed that transcoding the aspect ratio is hard to do. I get better results by going back to Premiere and making an HD sequenc, then copy my tracks from the 1024x768 sequence into the HD sequence, with pillars. Then use the Premiere youtube preset to encode the sequence.

So, in a nutshell, is what I have found is that transcoding my 1024x768 wmv files does not produce as good a result as making a new sequence in Premiere, copying the tracks over, then encoding the new sequence.

angevin1 wrote: »

I use Adobe- everything. CS5.5

I prefer Blu-ray but many of my clients and performers have only Dvd.

angevin1

Dan, No. My Br disc's all show just my limited Video Skills:D

No strange Frame sizing, just 720p, 1080p and MEPG-2-Dvd encodes into Encore. My movie files are just fine. In fact a week before at the film festival, same projector, same screen and venue, my files I delivered @ 1080p were super. They were even handbraked way down and looked fine.

Dvd encode is my issue. or rather the playback of dvd is the issue.

ziggy53

There are several issues that need to be handled first.

Getting 1080p or 1080i to DVD resolutions is a type of "downconversion" and it may be causing the jaggy-ness. If the video projector is then converting to 1080 for projection it may be scan-doubling or upconverting, either of which can also cause jaggy-ness.

It's best to start with one variable at a time and work from there.

1) What do the downconverted 720x480 (DVD SD resolution) files look like on your computer display? If they have the jaggies, it will not improve with projection.

2) What do commercial DVDs look like coming from the projector? If they also have jaggies then the projector, or possibly the projector setup, is at least past of the problem.

(Each of the above is introducing a single variable per scenario.)

angevin1

ziggy53 wrote: »

There are several issues that need to be handled first.

Getting 1080p or 1080i to DVD resolutions is a type of "downconversion" and it may be causing the jaggy-ness. If the video projector is then converting to 1080 for projection it may be scan-doubling or upconverting, either of which can also cause jaggy-ness.

It's best to start with one variable at a time and work from there.

1) What do the downconverted 720x480 (DVD SD resolution) files look like on your computer display? If they have the jaggies, it will not improve with projection.

2) What do commercial DVDs look like coming from the projector? If they also have jaggies then the projector, or possibly the projector setup, is at least past of the problem.

(Each of the above is introducing a single variable per scenario.)

Exactly. One variable at a time! I had done the work regarding the down-sampling and therefore on my Jaggy-Test file, I made a three copy version. Each 15 second segment had five snippets of a movie and then I rendered them: MPEG-2_dvd, 780p, and 1080p, and strung them all together in the timeline for Authoring. And honestly I couldn't and Cannot see a difference post transcode between files.

And I thought I had the HD thing figured out! Below are two requested screen shots straight from Windows media player playing the MPEG-2-Dvd file (first).




And this Screen Shot Below off of the Disc playing on Windows Media player on my Computer.



I see the Ghosting I am talking about on the gal with Red leggings. I do see ghosting sometimes on Hollywood Dvd's. But most of the time I am watching Br.

They both look fine as can be expected I think. I wasn't sure about the TV settings but did what you suggested regarding changing it to 480p to see if I thought there was a difference. There wasn't really. I suspect that I may have gotten as good as can be got based on the equipment I am showing these on.

I suspect you are onto something regarding the Projector. I am gong to set up some time with the projectionist to take a Disc in to see what these dvd files project like. But even on my Tv I don't particularly like them. A better term might be: They seem more pixelated on the TV.

More ideas? Things I am missing?

Thanks Ziggy53!

ziggy53

The exact equipment you have for your TV may not be optimized for DVD playback. Scaling error may be part of the problem. Please do look at some commercial DVDs, like from a major movie release. Try to find somewhat similar content to compare.

Note: I see now that you have noticed some similar artifacts from commercially produced DVD disks. That would seem to confirm that your TV system (DVD/BD player and TV screen) is producing "some" of what you are seeing as artifacts, from all DVDs. I would canvass a few additional people (viewers) to see if it is generally noticed.

For the projector it may be similar. The player that they use for DVD playback may be optimized for HD, for instance, and the projector settings itself may need to be tweaked too.

angevin1

ziggy53 wrote: »

The exact equipment you have for your TV may not be optimized for DVD playback. Scaling error may be part of the problem. Please do look at some commercial DVDs, like from a major movie release. Try to find somewhat similar content to compare.

Note: I see now that you have noticed some similar artifacts from commercially produced DVD disks. That would seem to confirm that your TV system (DVD/BD player and TV screen) is producing "some" of what you are seeing as artifacts, from all DVDs. I would canvass a few additional people (viewers) to see if it is generally noticed.

For the projector it may be similar. The player that they use for DVD playback may be optimized for HD, for instance, and the projector settings itself may need to be tweaked too.

Right, will do.

So, what about the frames above? Do you find them acceptable, or do you see deficiencies/problems?

Many thanks again~

ziggy53

The exposure for these seems awfully good, which is critical for a theatrical production.

What camcorder is used for acquisition?

Dan7312

I'm just curious what, if anything, you might expect to see different in these two frames. To my non-critical eye except for the fact that they are different frames in time I don't see anything really different between them.

I do see reflections of the actors in the stage. In paticular the reflection of the left leg of the dancer in coveralls behind red-leggings seems to line up pretty well with red-leggings right calf. Maybe that reflection ends up looking like ghosting in the dvd playback?

angevin1 wrote: »

Right, will do.

So, what about the frames above? Do you find them acceptable, or do you see deficiencies/problems?

Many thanks again~

angevin1

Dan7312 wrote: »

I'm just curious what, if anything, you might expect to see different in these two frames. To my non-critical eye except for the fact that they are different frames in time I don't see anything really different between them.

I do see reflections of the actors in the stage. In paticular the reflection of the left leg of the dancer in coveralls behind red-leggings seems to line up pretty well with red-leggings right calf. Maybe that reflection ends up looking like ghosting in the dvd playback?

Hey Thanks Dan. No what I see if I realy scroll in is what I think is visible below in the crop. I assume it is a component of the transcoding and perhaps it is necessary via the recipe to get rid of the jaggies.





Below is a good example of what I used to have regarding jaggies. Easily visible in her chin shadow and Shoulders too.



I viewed the Dvd again this a.m. with fresh eyes and realized I am making much-ado about nothing. Perhaps even guilty of pixel peeping. Considering this is a Standard Definition format, I ought to be happy with the result.

As a side note I very often pause Hollywood Dvd's just to see how well they handled different scenes. And about 80% of the time I can find jaggies and/or Halo's around subjects, but only in Pause mode. When in action can't see a thing. In my research over the course of these several Months I found that Hollywood uses proprietary-software that is lent to them by Sony and others. They cannot buy it. However they have engineers that have recipe's even for individual scenes or lighting scenario's. So considering all that, I'm happy with the result compared to what I had had as a result.

Should I upload a file clip to get opinions, or are we good enough with the screen shots?

Dan7312

I think pause is a pretty difficult feature to implement and I'm not sure it really is a good measure of the underlying video quality. You probably know this already but most of the frames in a video are just a compressed list of differences of the frame that preceded them. They really depend on the eye to smooth out the differences.

Years ago I worked on a project to cut and slice together random clips of mpg. The typical way to do that at the time was to completely render each clip and, then splice together complete frame results, then re-encode that. Our deal was we used math to do the splicing so we didn't have to render and re-encode and as a result could do it all in real time on a commodity PC (a 386 was a big deal for us). We literally had an phd applied mathematician on staff to figure out all the math.

The trick wasn't to come up with a perfect frame before and after the splice, is was coming up with the differences that made the eye, actually a Hollywood film editor eye, happy that the splice was done correctly.

So coming up with a frame from a vid that's the equivalent of full single frame is really difficult. But that's ok, I'm a pixel peeper too, I just can't resist.:D

BTW (off thread alert) the 5D3 can shoot in I-frame only mode, which means every frame is a complete image. In theory, I think, that should mean that rendering different encodings and resolutions should come out better because the software won't be trying make i-frames out of intermediate difference frames.

angevin1 wrote: »

As a side note I very often pause Hollywood ?

angevin1

Dan7312 wrote: »

I think pause is a pretty difficult feature to implement and I'm not sure it really is a good measure of the underlying video quality. You probably know this already but most of the frames in a video are just a compressed list of differences of the frame that preceded them. They really depend on the eye to smooth out the differences.

Years ago I worked on a project to cut and slice together random clips of mpg. The typical way to do that at the time was to completely render each clip and, then splice together complete frame results, then re-encode that. Our deal was we used math to do the splicing so we didn't have to render and re-encode and as a result could do it all in real time on a commodity PC (a 386 was a big deal for us). We literally had an phd applied mathematician on staff to figure out all the math.

The trick wasn't to come up with a perfect frame before and after the splice, is was coming up with the differences that made the eye, actually a Hollywood film editor eye, happy that the splice was done correctly.

So coming up with a frame from a vid that's the equivalent of full single frame is really difficult. But that's ok, I'm a pixel peeper too, I just can't resist.:D

BTW (off thread alert) the 5D3 can shoot in I-frame only mode, which means every frame is a complete image. In theory, I think, that should mean that rendering different encodings and resolutions should come out better because the software won't be trying make i-frames out of intermediate difference frames.

Okay. That was entirely over my head. Almost all of it anyway.

Yeah. Plan on renting the 5DIII to see the difference in resultant video before purchase. If I did more photo work, I'd just go ahead and get it.

Thanks, Dan~

Dan7312

Didn't mean to go so techno:D All I meant is that most frames are highly compressed, but about once every, maybe, 15 seconds the is an full, uncompressed frame. The uncompressed frames are called i-frames or key frames. In the 5D3 you have the option of making every frame an uncompressed one, at the cost of a lot more space in the cf card.

Converting a highly compressed frame to an uncompressed one is, of course, lossy so you end up seeing jaggies that you wouldn't see in video playback because your eye smooths them out.

angevin1 wrote: »

Okay. That was entirely over my head. Almost all of it anyway.

Yeah. Plan on renting the 5DIII to see the difference in resultant video before purchase. If I did more photo work, I'd just go ahead and get it.

Thanks, Dan~

angevin1

Dan7312 wrote: »

Didn't mean to go so techno:D All I meant is that most frames are highly compressed, but about once every, maybe, 15 seconds the is an full, uncompressed frame. The uncompressed frames are called i-frames or key frames. In the 5D3 you have the option of making every frame an uncompressed one, at the cost of a lot more space in the cf card.

Converting a highly compressed frame to an uncompressed one is, of course, lossy so you end up seeing jaggies that you wouldn't see in video playback because your eye smooths them out.

I'm familiar with Keyframes from Media Encoder and the idea of changing those for given movies. Sounds like something to investigate on Canon's new camera for me.

Dan7312

When you re-encode what you are doing is un-compressing those compressed frames then re-compressing them in the new encoding format. Needless to say each time you do that you lose a little bit of quality.

There actually is a spec on how much you might lose, but it's been a while and I don't remember the details. But it was something like... you can consider an mpeg file as archival if you can uncompress and then recompress that 12 (maybe 15? don't remember) times and have a Hollywood film editor say the final result was essentially the same quality as the original.

I just forget the data rates involved but I think on the project I worked on some time ago we could meet that spec' with MPEG that was compressed (I think) to about 7.5MBytes per second, but that wasn't for HD, it was standard definition.

Anyhow that's why I think the i-frame only option on the 5d3 might produce results that have a better chance of being re-encoded to other formats without loss of quality.

We'll who get's their 5d3 first to try it out:D

angevin1 wrote: »

I'm familiar with Keyframes from Media Encoder and the idea of changing those for given movies. Sounds like something to investigate on Canon's new camera for me.