While I rarely go out to the movies (because of my awesome home theatre setup), I did see Cloverfield in the theatre. Like many others, I wasn’t a fan of the extreme shaky-cam which literally left me nauseous. I need to start with my opinion on theÂ camera shakeÂ choice – because it’s relevant to what’s happening with the bonus features: I feel that they could have toned down theÂ amateur-home-video-shake and still conveyed theÂ cinÃ©ma-vÃ©ritÃ© “real found footage” feeling. I loved the idea of the movie and was impressed with the execution, but hated the over the top shaky-camera work. It made the movie very difficult to watch and therefore less fun. However, it was an artistic choice on the part of the filmmakers. Â Whether or not I agree with it’sÂ necessityÂ and/or effectiveness – it was technically correct. Â It was their choice. Â And even though I found that choice very annoying,Â I still looked forward to getting the movie on Blu-ray to get more info about the cloverfield mythology and see all the bonus features.
Well, the bonus features are harder to watch than the movie – and this time it was NOT an artistic choice…
As a professional editor, my eye is trained to see flash frames and field issues. Â When I watch something that has consistent field issues I see every glitch and it grates on me like cat claws on a chalkboard, typically resulting in aÂ migraine. This was the case with all the behind the scenes bonus footage for Cloverfield – andÂ the problem exists on both the Bluray and DVD versions. Â As bad as the shaky camera was on the movie, it pales in comparison to how bad the BTS is. Â It is completely unwatchable, which is a real shame. I noticed the problem months ago when the disc first came out. But this was the first opportunity I’ve had toÂ analyzeÂ the footage and try to figure out WTF is going on.
It appears to be wrong on so many levels that its hard to nail down exactly what the source of the problem is. At first I thought that maybe it was an encoding issue, or just an issue with my system. But seeing that the problem exists both on the BD and the DVD proposed that it was not an encoding issue but rather was mastered to tape with field issues. This raises several questions: How many people did the messed up video pass through before getting in my hands on a Blu-ray disc? The shooter(s), the assistant editor(s), the editor(s), the post-production supervisor, the segment producer, the executive producer, the online editor (if they had one), the dvd video encoder, AND the dvd QC guy? Â How many of these people noticed it was wrong and said “oh well”? Â Or even worse – how many of them honestly couldn’t tell that there was a problem? Â Where did the problem originate?
I set out to reverse engineer this. Â I’ve done post production consulting for many major clients. Maybe I can figure it out. Examining the footage, the fields appear so mangled that I can’t figure out how one would even recreate something this messed up. I’ll start at the beginning. We know that the BTS footage was probably shot with the Panasonic HVX200 P2 camera:
‘HUD’ shooting BTS –
The BTS HUD is shooting –
A closer view of the camera –
The HVX 200 –
In the BTS you can see this camera used a lot and it appears to sometimes be used to shoot the actual movie. But we know from the screenshots above that it was used to shoot at least that one BTS shot (picture 2), and that shot exhibits the problem.
The director using the HVX 200 –
We also know that the problem is widespread, appears to cover most all the footage in the BTS, is not just on certain shots, and therefore spans across multiple shooters. T.J.Miller, who played “HUD” is seen shooting some BTS footage, it appears that so is the director Matt Reeves, plusÂ Richard Oak is credited with the BTS camera work. Â My point is that likely one person set up the BTS cameras for the whole shoot, and picked the filming mode, which might have been 24P advanced mode… but there’s a twist. Â Even the footage borrowed from the movie has the problem – which implies that the issue happened in post production.
A quick lesson in 24p and 24pA: When shooting in 24p there are (at least) two ways to store those 24 progressive frames per second on a tape that runs at 30i. Of course a HDX 200 uses P2 cards but it also has DV tape. I believe it generally still stores 24p in one of these two methods. One way is the regular 24p mode which cine-expands the footage following a traditional 2-3-2-3 field pattern. Progressive frames are split every other line into fields. Â 4 progressive frames become 5 interlaced “telecine’d” frames. The pattern is AA, BB, BC CD DD, so that the BC and CD frames both have mixed fields. Â Then during capture the editing system can combine these fields back together on the fly and discard the repeated fields – thus returning back to the original 4 progressive frames. Â Another important note is that this patten is the one we’re all used to seeing. It has been used since we were all children to convert film to interlaced video. In America NTSC-land that pattern “feels” like film.
In 24pA, or “advanced” mode, every 4 progressive frames are converted to 5 interlaced frames using a different pattern: AA, BB, CC, DD, DA. The supposed advantage of this pattern is that the computer only has to remove one full frame every 5 to get back to the original 24 progressive frames, and doesn’t have to do the work of combining frames back together. A concept which to me has no real advantage since A) the footage looks awful when viewed without “decoding” and B) most editing software has no problem handling regular old 2-3-2-3 telecine’d footage.
AnalyzingÂ the Cloverfield bonus footage is difficult because “ripping” it, importing it into software, or even viewing it on a computer can actually change (or misinterpret) the field order. The easiest place to notice that there is a problem is on a camera pan, or any kind of movement. So I took a series of frames off the DVD and split them out into odds and even frames. Two disclaimers: First is that I’m not showing a clip of the problem on my site. It wouldn’t prove anything because it would be down-converted and re-compressed. Â Honestly, if you want to see the problem I recommend that you go out and rent or buy the movie and see it for yourself. Second is that I did the best I could to make sense of this mess. It is possible that the fields got messed up even worse in my process – but I don’t think so. Having said that, here’s what I found – and it’s a complicated train wreck:
Frame 1 is progressive. Â So both fields are of source frame A. I’ll call this A A. Â Frame 2 has one field matching Frame 1 (source frame A), and one field which is a MERGEDÂ image of source frame A and source frame B. Â I’ll call this A [AB]. Â Frame 3, 4, and 5 repeat the oddity of Frame 2. Â The first field is normal and the second field is merged. Â Then Frame 6 starts the pattern over again with another progressive frame. So the very strange fieldÂ pattern is: Â AA, A [AB], B [BC], C [CD], D [DA]. Â The first field of each frame matches the normal 24pA mode. But the 2nd field is a merged mess of the two fields on either side of it. How in the world could this happen? Â This means that for 4 frames you’re seeing an A frame in some form, which causes the footage to appear to jerk to a quick stop. Then for the next 4 frames you’re seeing frame blended crap. It’s like somewhere in post the “will it blend” guy got ahold of the footage. Check out these sample frames:
A progressive frame –
The next frame is interlaced, click to see it full size –
Field 2 of the same frame – click to enlarge and see the merged mess –
Like I said, who knows how this happened. My guess is that the BTS producer and editor didn’t really know how to handle 24p footage. Possibly they captured it as 30 frame video, which is wrong, and then output it using some equipment that was off by 1 field. I’ll also mention that I found frames that were split over a cut – with one field of shot A and one field of shot B. Likely an output problem. There’s also the possibility that they edited on a system without a real producers monitor that shows both fields. Â In an Avid you only see field one while playing back footage unless you look at a real monitor. Basically, probably more than one technical issue led to the mess.
I hate to publicly slam those who produced and edited these featurettes, but unfortunately it is ultimately the editor’s responsibility to understand how the footage was shot, what the sequence settings should be, and how to master it to tape.Â As a sort-of FAN of this movie, I have to say that it really ruins the disc. Which is a shame because the content of the bonus features were produced well.
One last point I’ll make is that NO ONE else has seemed to notice the problem. Â I haven’t been able to find any mention of it elsewhere on the web. Maybe it’s just no big deal. Maybe no one cares – all the way from J.J. Abrams to joe six pack.
What do YOU think? Did you see this disc? Did you notice the field problems or anything jerky about the bonus features? Did it bother you? Do you think the general audience cares? Let’s get some discussion going…
3 thoughts on “Cloverfield BD and DVD Bonus has major field issues”
Michael — GREAT article. I hope this gets the traction it deserves. // I am confused about the compression artifacts on the progressive shot you show. Looking at the girls foot and then also the horizontal lines in some of the macroblocks… it does appear that this frame might not be completely progressive. What do you make of those two issues?
Michael — GREAT article. I hope this gets the traction it deserves. // I am confused about the compression artifacts on the progressive shot you show. Looking at the girl's foot and then also the horizontal lines in some of the macroblocks… it does appear that this frame might not be completely progressive. What do you make of those two issues?
Thanks. One quick note. The screen shots that towards the top of the article don't show the problem because I specifically grabbed the progressive frames.
Of the 3 pictures you're mentioning with the girls foot, the first one is a good frame. Any artifacts or microblocking you're seeing there is because of the MPEG encoding and/or the way I ripped the frames.
The second frame, if you click through to see the full size, shows what looks like a NORMAL interlaced frame. Interlaced frames are normally found in DVDs. What makes it wrong is what's found in the second (lower) field – which is seen in the 3rd picture… That picture shows that within that field there are two frames that are merged. The result when playing full speed is that the motion is all blended together and then jerks to a stop, then moves all blended again. So it appears to stutter along.
I double-checked this the other night on my Bluray and it's very hard to watch.
Comments are closed.