What is this Artifacting and How Do I Avoid It?

[Steven Reid]

Can someone please explain to this dunce the nature and cause of the blocky and color artifacts that are in my frame grab below? Obviously, it is a high motion scene. The boy's red shirt, his gray arm sleeve, and the man's exposed forearm in the background all show these artifacts.

This is how I generated the frame grab:

1. Source footage = 1440x1080 60i HDV (from Canon XH-A1).
2. Converted HDV source to Cineform intermediate avi ("high quality" setting; maintained all other properties of source footage).
3. In Vegas, I matched media to set the project properties.
4. Following a tutorial on Eugenia's site (my goal is to generate 60p for use in slow-motion), I changed the project properties from "interlaced" to "progressive", interlace method = interpolate, and I set the frame rate to 59.940 (double NTSC). Clip property = "smart resample."
5. Rendered to Cineform avi as progressive with frame rate = 59.940.
6. Placed rendered clip from (5) onto timeline with same project properties as summarized in (4). Exported frame grab. The artifacts are visible in Vegas as well as in Windows Media Player when I play the rendered clip and hit pause.

Other potentially informative notes:

A. I also deinterlaced the source HDV clip in Cineform Neoscene by checking the appropriate box. I have no idea how Neoscene deinterlaces.
B. I also deinterlaced the source HDV clip in VirtualDub.
C. Also tried Mike Crash's smart deinterlace filter.

(A) - (C) gave exactly the same visual result as my original attempt in Vegas. This leads me to believe that the artifacts are not caused by my XH-A1 or any of the software I mentioned, but rather relate (somehow) to the deinterlacing itself. The artifacts are noticeable in high motion scenes, as in the frame grab, but (hardly) noticeable in mostly static scenes, i.e., those with little motion.

So, am I doing something "wrong," or do I have to live with the artifacts because they are an inevitable consequence of what I'm trying to do (60i -> 60p)?

Thanks!

Steve

[Joe Parker]

I don't see anything beyond normal video compression artifacts. Get used to some of that stuff. But don't worry; clients really don't tend to see it if the content is good.

[Steven Reid]

Originally Posted by Joe Parker

I don't see anything beyond normal video compression artifacts. Get used to some of that stuff. But don't worry; clients really don't tend to see it if the content is good.

Really? I shot the same subjects in progressive (24f on the XHA1) and the footage was absolutely pristine: no blocky or chroma artifacts. But the average viewer, as you note, won't really notice these things (I do because I'm staring at it close up in the edit bay!).

Steve

[Perrone Ford]

Originally Posted by Steven Reid

So, am I doing something "wrong," or do I have to live with the artifacts because they are an inevitable consequence of what I'm trying to do (60i -> 60p)?

Thanks!

Steve

Issue #1 You are shooting HDV. This offers precious little information for the compressors to go on it post.

Issue #2, you are shooting interlaced. This means you are recording half-frames in the already stressed HDV format.

Issue #3, 60i translates to 30p. So as soon as you say you want 60p, the computer must guess at creating an extra 30 frames per second with relatively little information to go on (interlaced HDV).

Essentially, you are creating the worst of all scenarios to try to accomplish what you want.

[Ron Evans]

Issue #3, 60i translates to 30p. So as soon as you say you want 60p, the computer must guess at creating an extra 30 frames per second with relatively little information to go on (interlaced HDV).

Interlace records an odd or even field every 60th sec ( NTSC) so to go to 30P the computer still has to make up ALL the 30P images. It is unfortunate that the convention says 60i is 30 frames since this doesn't mean 30P, just that two fields are needed for a frame and 60 divided by two is 30 . The odd and even fields are not from the same time they are 60th sec apart so there is always half the vertical information missing whatever the final progressive frame rate will be. The computer always has to interpolate to create the progressive frame. This is why its better to start with progressive recording if that's what is needed.
Simplistically the 30P frames from a 60i source may be thought of as full frames halfway in time between the odd and even fields but held for 30th sec. To do this correctly the computer has to interpolate over several frames or strange judder may be induced.
Other than the amount of storage the same action is needed to go to 60P and several DVD players do this as well as some of the high frame rate TV's.
OPPO BDP-83 Blu-ray Disc Player
To the main question. Vdub seems to have the best conversion so I would use the source HDV video( not Cineform) first in Vdub to get a deinterlaced source file. Then see how that performs.

Ron Evans

[Perrone Ford]

Yes Ron, thanks for the clarification. I suppose what I meant was that if you take a 60i file into a deinterlacer, it's going to give you back a 30p file, not a 60p file.

[Steven Reid]

Ron, Perrone: many thanks for the helpful replies. I had a hunch at the outset that I was headed up a steep hill to begin with, so to speak.

The slo-mo tutorial that I was following instructed to shoot in the highest frame rate possible (e.g., 60i for "most" camcorders) and, perhaps unwisely, I did just that. The whole idea was to (1) generate progressive frames from interlaced source footage and (2) apply a velocity envelope with smart resampling then and render out the slo-mo clip in progressive.

In retrospect, especially since deinterlaced "60i" = 30p, I wonder if I could achieve better end results by simply shooting in 30p (available on the XHA1) to begin with and use that in my slo-mo workflow. That is, eliminate my step (1) above and simply apply velocity envelopes to 30p source footage. Is this the point where I smack my palm to my forehead?

Steve

[Perrone Ford]

Steven,

The advice you read is seen often. The idea behind it is that 60i gives more temporal information to work with. Essentially, you are getting information updated more frequently than if you simply used 30p, even if it is not complete information. There is some merit to that.

So, let me ask this. Do you need to deinterlace the footage? Why not take the 60i footage straight into the timeline, and add a velocity envelope to it? Why deinterlace it at all? I am unsure what the results would be, but I'd certainly try it.

You may also want to go look at the three newest videos on my Youtube page, as they are all slo-mo examples done from progressive footage.

YouTube - perroneford's Channel

[Steven Reid]

Originally Posted by Perrone Ford

Steven,

The advice you read is seen often. The idea behind it is that 60i gives more temporal information to work with. Essentially, you are getting information updated more frequently than if you simply used 30p, even if it is not complete information. There is some merit to that.

OK, yes, that makes sense.

Originally Posted by Perrone Ford

So, let me ask this. Do you need to deinterlace the footage? Why not take the 60i footage straight into the timeline, and add a velocity envelope to it? Why deinterlace it at all? I am unsure what the results would be, but I'd certainly try it.

My main project and delivery format is in 24p because that is what I shoot 95% of the time. Slowing 24p just looks bad, and so that's why I looked around for Vegas-friendly slo-mo techniques, such as the one I described above. The slo-mo clips will be sprinkled throughout my 24p project, and so I thought that I needed to deinterlace the 60i-source to first get to progressive, then apply a velocity envelope, render out the slo-mo clip into the source frame rate (here, 24p), and drop it into my main 24p project (yes, the mismatch in frame rates gives some juddering, but in high action scenes it's not so noticeable).

I have not tried placing onto my 24p timeline, and adding a velocity envelope to, my 60i source. I think the steps I followed were designed to get as close as possible to sharp 60p before adding a velocity envelope. If you're interested, I followed Method 2 here.

Originally Posted by Perrone Ford

You may also want to go look at the three newest videos on my Youtube page, as they are all slo-mo examples done from progressive footage.

YouTube - perroneford's Channel

Those look very nice (subject ain't hard on the eyes, either). Would you mind sharing your workflow that resulted in those clips? 25% is about the slowest I go, and your 25% example looked quite nice.

Steve

[Perrone Ford]

I belive I just dropped the clip onto Vegas, stretched the event, disabled resample, and rendered out. Nothing tricky. I think I did one in Virtualdub also, but don't remember the workflow.

[Steven Reid]

Originally Posted by Perrone Ford

I belive I just dropped the clip onto Vegas, stretched the event, disabled resample, and rendered out. Nothing tricky. I think I did one in Virtualdub also, but don't remember the workflow.

From 30p source footage? (I couldn't tell from your YouTube video descriptions.)

Steve

[Perrone Ford]

I believe it was 30p. I'd have to check the source to know for sure.