An Analysis of the Microstutters

[DarrellP]

GeneWildersHair, if you're out there, you'll love this:

I've been playing around with my new Panny E85 DVD recorder and curiosity got the best of me. While playing back a Tom Petty DVD that I'd created, I saw a microstutter and thought "Why don't I frame by frame through the stutter and see what happens?".

So I did exactly that and here's what I found: The stutter involves 2 frames and it is NOT a dropped frame, rather it is a frame replayed from 2 frames previous with a still frame inserted after that, then it goes back to the frame where it first dropped, so it kind of looks like a quick double-take, really weird.

The theory that was kicking around was that due to the frequency timing and all that, there was a frame being dropped, but this is not the case. I repeated this process over 2 different microstutters and the same thing happened both times, so I am going to make a huge assumption that this is what always happens when a microstutter appears.

Just thought you might want to know this little tidbit, if someone already posted in the other huge microstutter post, my apologies, I thought I was having a revelation here.

 

[Ilya]

Great find! Was the STB set to the progressive scan mode (720p) or to the interlaced mode (1080i)?

 

[oddwunn]

Darrell, so how many frames total are "incorrect" during a single stutter? Based on your description, it sounds like 3 incorrect frames, as you mentioned "it goes back to the frame where it first dropped". And what exactly do you mean by "still frame"? Do you mean it is a duplicate of the previous frame?

I'm just curious, as I had a hard time believing that anyone could see a single dropped frame (as in the prevailing theory). I find it difficult to spot anything less than 3 dropped frames in a row, and thus thought that the stutter occupied at least 3 frames total.

 

[Ilya]

The theory that was kicking around was that due to the frequency timing and all that, there was a frame being dropped, but this is not the case.

Yes, it's not dropped it is repeated. Thanks for confirming this with your experiment! The way I understand it, the satellite sends 2997 frames in 100 seconds to the STB. The STB outputs 3000 frames per 100 seconds. Three frames are missing every 100 seconds, or 1 frame every 33 1/3 seconds. The STB makes up for the missing frame by repeating one of the previous frames. So if we watch some constant motion (e.g. CNN ticker) it visually pauses for a fraction of a second (for 1/30th of a second to be more precise) and then jumps ahead. And this occurs every 33 1/3 seconds.

In reality, it gets a bit more tricky than what I explained above because we are actually dealing with interlaced signals (480i or 1080i). Instead of one frame, we might be seeing two half-frames, further deinterlaced either in the STB itself or in the TV set...

 

[DarrellP]

I have my stb set to Auto, but it's really insignifcant since the recording is 480i all the time, regardless of the HD output.

There are a total of 3 incorrect frames:

Frame 1: Is inserted from 2 frames previous (like a rewind).
Frame 2: By a still frame I mean the image does not progress forward or backward.
Frame 3: Picks up from where the stutter began and continues on, so there never is any picture info lost, just 3 frames appear to be repeating (sort of).

My Panny records the signal in a progressive format (or at least with all the signals split out so it can be played back progressively in 480p).

I find it pretty incredible that our eyes can catch this miniscule "burp", but there it is.

 

[squicken]

Great committment Darrell! I always wondered how it was possible to see 1 dropped frame. If you're tests are correct, I can understand how 3 dropped frames are noticed.

 

[oddwunn]

I find it pretty incredible that our eyes can catch this miniscule "burp", but there it is.

Yes, it is quite incredible! But not nearly as incredible as people who believe that they can see a single dropped frame. Three frames makes sense, as I mentioned earlier that I have been able to spot as few as 3, and I consider my eyes to be at least somewhat trained. I would never say that it is impossible to spot a single dropped frame, but my eyes are not good enough. My conclusion, based on my own ability to observe, was that there were 3 or more dropped frames (or more accurately, repeated frames), though I never discount the possibility that some people have better eyes than others and therefore could see a single misplaced frame.

Three bad frames also explains why some people don't see the stutter at all - Even some of my video enthusiast friends have been unable to detect 3 dropped frames which I could see, so I figure that eye/brain synapses occur at different speeds from person to person. You know, maybe the same reason that some people see horrendous rainbows with DLP projecters, while others see them mildly or not at all. Rainbows are a time related phenomena due to the strobing effect of color wheels, and my guess would be that the same people who see the worst rainbows would tend to be the same people who saw the stutter the worst. Just a theory, though....

Thanks, Darrell, your frame by frame analysis was excellent, and at least in my mind explains the stutter more accurately than before.

 

[DarrellP]

and my guess would be that the same people who see the worst rainbows would tend to be the same people who saw the stutter the worst.

And I am one of the lucky few that rainbows affect severely. I used a DLP from work for about a year before buying my LCD PJ and that was the deciding factor for me to go LCD.

 

[rang1995]

Post this over on Yahoo ASAP so wilt can see it or e-mail him..great work!!

 

[Ilya]

I figure that eye/brain synapses occur at different speeds from person to person.

It's like with CRT computer monitors. Some people are more sensitive to flicker than others. I can't work on a monitor that is set to anything less than 72 Hz, but some people don't see much difference between 60 Hz, 72 Hz or higher refresh rates.

 

[Ilya]

Post this over on Yahoo ASAP so wilt can see it or e-mail him..great work!!

Since this problem is already fixed in the next version, I guess it's more for our own curiosity...

 

[GadgetRick]

Has it been fixed?

So this is one of the things which is supposed to be fixed in the next version of software? I sure hope so. I don't notice it all of the time but I do notice them from time to time.

Great catch. Let's hope this annoying but interesting problem is fixed!

The Rickster

 

[GeneWildershair]

great find Darrell! This needs to be brought to the attention of someone at V* imediatly, as a lot of us saw this problem occuring all the time, and found it very annoying. Very interesting about the 3 frames being repeated. And it is funny how some notice this stutter, and others don't. The whole time we had our V* set up, my dad, and I would notice when he was over, but the wife never saw it? hehe... just goes to show you just how crazy some of us are to be able to see it.

 

[Sean Mota]

Jay,

microstutters are fixed in next software version...

 

[FunkyBoss]

So any guesses on what they did to "fix" it? Somehow only display the 2997 frames in 100 seconds?