60i on 24p timeline/deinterlace method

K

kmcgrath

Well-known member
If you select "none" for the deinterlace method, does Vegas combine both fields from a 60i frame for the 24p frame (giving full resolution)?

And if you select "interpolate" for the deinterlace method instead, does Vegas select only one 60i field for the 24p frame (giving half resolution)?

Thanks, Ken
 
Combining fields will never give you "full" resolution. You'll never get more than 70-75% resolution from an interlaced image.

Remember that if you combine (or "blend") fields, you're looking at two separate moments in time. If there's any motion, you're going to have artificating you need to get rid of, and that will require some kind of blur -- on top of the blurring that interlaced imaging already needs to do to reduce flicker.

"Blend" fields or "none" may help improve resolution in footage with little or no movement, as compared to "interpolate," but it won't ever be as high of resolution as native progressive footage.
 
Except for the motion issue and artifacts that you mention, did I get the handling of fields/frames correct by Vegas for the two scenarios that I posted (i.e., one 60i field used for a 24p frame with "interpolate" deinterlace method, and two 60i fields used for a 24p frame with "none" deinterlace method)? Thanks.
 
You'll never actually see the artifacting. What I meant was that the artificacting will be handled by Vegas -- you won't have to do anything yourself -- but it will reduce resolution. Like I said, if there's little or no motion, there's nothing or little to blur, so you get a resolution save.
 
kmcgrath said:
Except for the motion issue and artifacts that you mention, did I get the handling of fields/frames correct by Vegas for the two scenarios that I posted (i.e., one 60i field used for a 24p frame with "interpolate" deinterlace method, and two 60i fields used for a 24p frame with "none" deinterlace method)? Thanks.
Click to expand...

I'm not sure about Vegas's internals, but the answer to your question is in general yes. Interpolation synthesizes every other line of pixels as something similar to the two lines around it. "None" I would assume does no de-interlacing, so leaves both fields of each frame intact. Thus, you'll tend to have "jaggies" in fast motion elements when looking at complete frames.

Mind you, this is all separate from the issues of converting from 60i to 24p, where interpolation may be something entirely different - since the frames can't be used 1:1 without a speed change, there is often a process of interpolating entire frames - that is, generating new frames that are somewhere in-between two source frames. In the simplest cases, this is done the way the DVX does... the "interpolated" frames of 60i-encoded 24p or pA are just mix'n'match the different fields from the frames around them. But interpolating frames between two progressive frames can be done in much more sophisticated (and processor-intensive) ways... you can probably see some examples of this by playing around with slowmo and fastmo on a 24p timeline, or looking at examples of these made with expensive plugins.

My guess from your options above is that Vegas's interpolation option is referring to how to treat the interlaced-to-progressive conversion; and the 30p to 24p conversion is done by frame interpolation after that.

Now, here's the trick: intraframe interpolation ((60i->30p)->24p) has the potential of higher resolution, but less accurate motion, especially in fast motion - the interpolated frames must guess and synthesize more of the temporal information.

By using no intraframe interpolation on the source material (60i->24p), there is some loss of image resolution, but better temporal resolution - each final frame has closer-matched "real" information from one of the fields of the source material. So, with a good algorithm, the motion can be smoother.

Anyway, these are the basic tradeoffs, and, for any piece of software (Vegas native or plugins, etc.), there will be situations where each approach shows its benefits. The most sophisticated approaches probably blend both of these techniques, using one as a heuristic to guide he other...
 
But is only one field used if the "interpolate" deinterlace method is selected (for 60i on 24p timeline), and two fields used if the "none" deinterlace method is selected?
 
kmcgrath said:
But is only one field used if the "interpolate" deinterlace method is selected (for 60i on 24p timeline), and two fields used if the "none" deinterlace method is selected?
Click to expand...

This is absolutely true for conversion from 60i to 30p.

Without undisclosed knowledge of Vegas's internals, it's not so clear for 60i to 24p, since the framerates don't match. My GUESS above is that yes:

"Interpolate" option makes 30p frames as you've described (through intraframe interpolation), then interpolates these frames to get 24p (at higher image resolution)

and

"None" option uses all fields of the 60i frame to use in process of conversion to 24p, (at higher temporal resolution).


But this is a guess, since "only Vegas programmers know what's really going on."
 
Thanks much. I feel a lot better knowing that only the Vegas programmers really know exactly how it is done ( 'cause I was sort of thinking to myself that maybe I was the only guy out there who didn't quite understand Vegas' deinterlace methods). That being said, it would be nice if Vegas would put out a bit more info just so we could all up on the details.
 
ullanta said:
Interpolation synthesizes every other line of pixels as something similar to the two lines around it. QUOTE]

How specifically does "blending" differ from "interpolation"?
Click to expand...
 
Yes! I agree.

It seems from what people report here that Vegas does a good job of making decisions about how to mix'n'match footage, and that in many situations "everything just works" without one having to think about how... but for those of us who are interested in the details of exactly what it's doing, it could be a little more forthcoming...
 
kmcgrath said:
ullanta said:
Interpolation synthesizes every other line of pixels as something similar to the two lines around it. QUOTE]

How specifically does "blending" differ from "interpolation"?
Click to expand...

I think, basically, the word "interpolation" is a little more precise. Or at least fancier sounding.


In mathematics, interpolation is the term used for synthesizing points along a curve. If you have a line, and know two points along the line, generating any other point is trivial. But if you don't know the function (if, say, same two points are really on a sine curve), then guessing what lies between them is tricky. That's why there are so many heavy interpolation techniques (splines, etc.). "Blending" to me sounds kinda like simple averaging. But neither "blending" nor "interpolating" give you more than a real general idea of what's going on - using nearby pixels or frames to guess at what happens between 'em.
Click to expand...
 
It is my understanding that "interpolating" involves using one field only. It generates the data for field two by "interpolating" between lines in field one, for example.

Whereas "blend" uses both fields; perhaps by interpolating and then combining the newly-interpolated field with the not-being-used field to come up with a "blended" field. Not 100% sure, but I believe that's how it works.
 
Interesting... it would be nice to get a definitive answer. In general, it seems common to talk about interframe "blending" and intraframe "interpolation". Is there indeed (in Vegas, at least?) options for deinterlacing (intraframe interpolation) by "blending" vs. "interpolation"?

I'm not sure I see any benefit to intraframe "blending" with the distinction that that Green Barry makes... which reinforces the idea of blending as more inter-frame oriented. When dealing with combining frames over time, that kind of "blending" could indeed smooth out motion in some cases... this would correspond to my second ("none" or "(60i->24p)) option above...
 
So "interpolation" takes a single field, and generates the missing lines by "averaging" the line above and below the missing line to give the new line. Therefore, intra frame.a So far, so good.

"Blending" does the same (interpolation) as above using one field of a frame, then takes the other field and "blends" with the interpolated field to form a new frame. But it is still using only one frame, so how is this considered INTERframe?

Finally, what does deinterlace method "none" do? My guess is it just takes each field of a single frame and simply combines both to a single progressive frame? Anyone want to confirm (or deny) this?
 
Jubal28 said:
Yes, both deinterlace methods are available. Blending is interframe; interpolation is intraframe.
Click to expand...


Hmmm... I'm not sure what this means. Admitedly, we're in the realm of terminology that differs from system to system... but in my interpretation, "deinterlacing" is a single-frame, always intraframe operation. In converting frame rates, you may "blend" different fields from different frames, but I'm not sure I'd call that "deinterlacing". If I'm not mistaken, this was implied in the Green Barry's answer, too... So I'm not sure what "blending" would be as a deinterlacing option; I understand what it would be as a frame-rate conversion option.

Unfortunately, what we're discussing here is a single operation (putting 60i on a 24p timeline) that must include both of these things, so we don't know exactly what the terms apply to... anyway, I think we all understand (?) what we think is going on here; and are just semantically trying to pin down terminology, no?
 
Maybe I'm alone here, but I must admit that I'm not exactly sure what is going on, but I sort of kinda think I do (but, only after the above very helpful discussion!!). All of my comments have been from the standpoint of trying to understand specifically what the three deinterlacing methods do. I brought in the "60i to 24p" part only because that scenario necessarily requires deinterlacing (I hope I'm correct about that). So I really wasn't trying to figure out how Vegas goes from a 30fps timeline to a 24fps timeline in terms of reducing the number of frames, just how the deinterlacing part is accomplished.
 
Right. The point (and confusion) here is that deinterlacing and framerate conversion are somewhat interdependent. As i said before, deinterlacing a single frame (i.e., 60i to 30p or 50i to 25p) is clear and straightforward. But since the framerate conversion can take advantage of the individual fields, the deinterlacing method matters... for different footage, different deinterlacing technique and different frame-rate reduction techniques will be better.
 
I think we agree that Vegas' method of deinterlacing a single frame (from 60i to 30p for example) by the "interpolation" method is understood, but I am still unclear about the "blend" method of deinterlacing (even for the simpler 60i to 30p case). If I understood Barry Green's post above, his best guess for the "blend" mode is that one of the fields is interpolated, then blended with the other field of the same frame.

And then there is the "none" choice, which I think just takes each field and slaps them together into a single frame....but I could be wrong about all of this!! :-)
 
You must log in or register to reply here.
Back
Top