PDA

View Full Version : Scenarist: Error with 23.976fps MPEG


BassPig
2nd August 2008, 02:04
I may be confused but I thought DVD supported 24fps film rate.

I have some true 23.976 fps footage encoded in MPEG2, which I tried to author into a DVD today, but Scenarist rejected all of it with the following:

"Error: Frame rate (1 = 24 000+1001(23.976)) is wrong."

I cannot figure out what to tweak in Scenarist so that it will accept 24fps footage. I know movie DVDs are encoded at 24fps from film, so the DVD spec MUST support it, but Scenerist doesn't?

Maybe someone can tell me what's going on.. maybe I'm just overworked this week...


EDIT
------------

Some new information.. after delving into my LaBarge "bible" on DVD authoring, I come to realize that there is no such thing as 24fps on DVD.. however, there is much information out there that leads one to believe otherwise.

So, I re-encoded entire streams to 29.97fps (why Adobe Media Encoder even offers a "DVD" compliant 23.976 frame rate for encoding DVD video, I may never know) and there are two disadvantages off the top:

"Ghost" frames in between real frames, irregular motion with pans and other smooth motions in the original 24P footage which gives it a kind of stuttery flwo to the motion

and

mosquito noise and compression artifacts visible in the 29.97 version for the same bitrate as the 23.976 version, which, examined frame by frame, looks just about perfect at 7mb/s avg.


Odd thing is, I can recall no commercial film on DVD where I see ghost frames in between the 24p frames. And none of the irregular motion flow, osrt of a 1-2-3-skip-12-skip-1-2-3-skip kind of cadence that is real annoying on slow pans.

From this, I deduced that Hollywood DVDs were encoded at try 23.976fps, but apparently not. But 3:2 pulldown doesn't seem to produce as good a result as native 24p MPEG2 footage. The motion flow is perfect on the 24P render and the frame quality is perfect for all intents and purposes.

Is this what I can expect from DVD output of true 24P footage?

mpucoder
2nd August 2008, 06:00
Don't re-encode the video, run it through pulldown. Either the original pulldown.exe, which can be found on the full software download page of Doom9, or DGPulldown (http://neuron2.net/dgpulldown/dgpulldown.html)
Pulldown changes the framerate in the headers and adds flags to cause the player to repeat fields. Your source needs to be progressive pictures (frames) in a non-progressive sequence (ie destined for interlaced display)

BassPig
2nd August 2008, 09:25
Thanks for the suggestion of pulldown.exe.

One concern with this is that repeating frames will cause a stutter to the motion. I'm already noticing this with MPEG rendered to 29.97 from 24P material. When the camera pans slowly, this disturbing smooth-stop-smooth-stop-smooth motion in rapid succession occurs. Maybe the DVD player will handle it better than a computer?

I will definately try the flagged version and see if that looks right on the burned DVD.


EDIT:
----------------

I played with both programs. My preference is to DGpulldown. Interface leaves no mystery about what's happening.
Tried it on a 23.976 elemental stream and Scenarist accepted like a charm.
Now remains to mux the DVD and burn a test disc. Playback on the workstation looks the same as before--no degradation (naturally, because no recoding).
If this works on the DVD player, I'll be VERY pleased.
Adobe should be setting these flags with their media encoder.

Eric69
2nd August 2008, 15:35
Here's a recent post that explains alot:

http://forums.creativecow.net/readpost/155/870715

BassPig
4th August 2008, 08:26
I tested that pulldown trick all the way to final DVD and watched it on two systems: a LCD with true 24f progressive and a S-video CRT on an older DVD player. Both handled it beautifully. I stepped a frame at at time and each frame was unique--no duplicate or ghost frames and no 3;2 judder cadence.
The benefit became clear when I paused during fast motion and examined the frames, sitting just arms' length from a 47" LCD--there were almost no detectable MPEG artifacts at all. And over on the CRT, no artifacts detectable whatsoever.
I pulled out a few commercial Hollywood megabuck movie DVDs and did some comparing, side by side with my equestrian video (which was shot on XDCam).. I compared live action films and CG animated films, which are generally ultra-clear, even on DVD. Our footage looked demonstrably cleaner, more detailed and less artifacted in all test examples. What a difference the extra bandwidth makes!

neuron2
4th August 2008, 14:24
One concern with this is that repeating frames will cause a stutter to the motion. Pulldown does not repeat frames! It repeats fields. It makes a big difference to the fluidity of the resulting video. I'm happy that you are pleased with DGIndex.

BassPig
4th August 2008, 15:15
Apparently, there's a huge difference between watching footage rendered from 24p to 29.97 progressive, and footage rendered 24p and pulled down (up?) to 59.94i by a DVD player. From what the articles I read were saying, it actually increases the duration of some 60i fields by 50% on playback. Whatever it's doing, it does it with no hint of an out of place field/frame and the quality... I still can't believe how GOOD DVD can look when you take away the grain and the oversharpening artifacts that Hollywood seems to put into their DVDs to make them look sharp on small TVs. If it looks good on a progressive 47" screen viewed from 24", it will look good under any normal viewing circumstance. Following a running horse with a fast-panning camera normally spells disaster for long-GOP CODECs, but the extra bandwidth made the difference between a ratty looking picture and a great picture, in this case.