[FFmpeg-user] What is 'yuv420p(tv, smpte170m, progressive)'?

Mark Filipak markfilipak.windows+ffmpeg at gmail.com
Tue May 19 22:22:08 EEST 2020

On 05/19/2020 12:30 PM, Carl Eugen Hoyos wrote:
> Am Di., 19. Mai 2020 um 17:51 Uhr schrieb Mark Filipak
> <markfilipak.windows+ffmpeg at gmail.com>:
>> Regarding why "smpte170m" & "progressive" & "top field first" and other
>> notations appear to be arguments to a supposed yuv420p() function,
>> well, you'll have to ask the developers.
> There is no such function, the encoding properties that are not necessarily
> related to each other are not arguments to anything. These properties are
> in parentheses because they are - from FFmpeg's point-of-view - less
> important for decoding than resolution, aspect ratio and other properties.

> "tff" and "bff" do not necessarily imply interlaced (or telecined) content,

Then why are they reported by ffprobe & ffmpeg, eh?

> while "progressive" can theoretically be interlaced (or hard telecined) it
> is very unlikely.

That statement seems contrary to the MPEG standard.

How well does ffmpeg conform to the definitions found in the MPEG specification? In particular:

"Pictures to be coded can be either interlaced or progressive."

Regarding interlace,
"Intro. 4.1.2  Coding interlaced video
"Each frame of interlaced video consists of two fields which are separated by one field-period."
That would appear to preclude hard telecine because for the combed frames, the two fields are 
separated by one frame-period, not one field-period. Thus, the hard telecine frames that exhibit 
combing are combed, not interlaced. I know that most people say "interlace", but that's just a point 
of confusion.

"3.64 frame: A frame contains lines of spatial information of a video signal. For progressive video, 
these lines contain samples starting from one time instant and continuing through successive lines 
to the bottom of the frame. For interlaced video, a frame consists of two fields, a top field and a 
bottom field. One of these fields will commence one field period later than the other."

Regarding progressive,
"3.103 progressive: The property of film frames where all the samples of the frame represent the 
same instances in time."

Thus, I conclude that the combed frames in a hard telecine are not interlaced and, by the preceding 
definition, are not progressive, either. Anyone who knows this stuff knows that without thinking too 
hard about it, but people who are new to this technology are greatly confused by references to 
interlace when the situation is not an interlace situation, or to progressive when the situation is 
not a progressive situation. It impacts the interpretation of deinterlace, what deinterlace does, 
when to apply it, and what to expect.

The stuff above is important because if the structure of a transcode is wrong, then all the 
smoothing filters or whatnot filters will not achieve what people want to achieve.

More information about the ffmpeg-user mailing list