Proudly Presents
Release Name .....: Top.Gear.US.S03.VOB.SAMPLE.DVDRip.x264-DEiMOS
Release Date .....: 28/Apr/2014
TL;DR
I realize most of you don't care about this. You might have looked and seen with your
own eyes that this season was done with enough care and diligence that it's due. Or
else you just might not give two fucks. Either way, here's what you need to know.
There's been suggestion that the episodes were encoded incorrectly, and that IVTC was
needed to remove dupe frames. In fact, using IVTC (or any other dupe-removal method)
results in decimating several key 29.970fps segments present in all the episodes and
5/6 extra segments. Using IVTC where it's not needed is like killing a spider with
a grenade. It'll do something to the spider, but not without a lot of collateral.
In this brilliant late-night simile, the spider are dupe frames and the collateral
is the frames that aren't dupe, but are still killed with the "spider". I swear this
made sense in my head.
We decided that keeping all frames was the correct course of action, since that way
we didn't lose valid video frames. A few dupe frames is worth it to prevent jerky
playback during the more "enjoyable" segments of the show.
If you're curious how we came to this (obvious) conclusion, read on...
Read Me
Top Gear US (Season 3) was released on 4 DVDs (see proof image on any one of the other
17 releases for proof). 4 episodes per disc were provided, plus nominal extras on the
first disc.
All the episodes and bonus features (that we released) seem to have been prepared the
same way by BBC Worldwide Americas (except 1, see below). Footage in or near the cars
where the hosts were present were filmed at 29.970fps. Footage of them standing around
or landscape footage or other shots were filmed at 23.976fps. The DVDs were presented
to me as a 29.970fps progressive source, which is how I prepared them.
SDx264 v1.1 rule 3.6 says that VFR is not allowed, so I couldn't simply drop dupe
frames from the 23.976->29.970 fps segments and call it a day. I either had to drop
dupe frames throughout, or leave some in. I chose the latter.
For reference, I've included a VOB sample from each and every episode and extra video
that we released. You'll note that
Top.Gear.US.S03.Adam.Show.Cadillac.Concept.DVDRip.x264-DEiMOS is 23.976fps. I looked
through it quite a bit and simply couldn't find any of the 29.970fps segments that I
was able to find in ALL the other videos, so I dropped every 5th frame as is required.
Now, proof time!
I've included an MKV encode of the VOB sample for S03E01, done 3 different ways.
sample-top.gear.us.s03e01.dvdrip.x264-deimos-1-in-5.mkv
- A straightforward 1-in-5 decimation, accomplished via: TDecimate(cycleR=1)
sample-top.gear.us.s03e01.dvdrip.x264-deimos-ivtc.mkv
- A straightforward IVTC, accomplished via: tfm(order=1).tdecimate()
sample-top.gear.us.s03e01.dvdrip.x264-deimos-progressive.mkv
- A straightforward encode, nothing fancy:
Feel free to compare them. Here, I'll help. The 1-in-5 sample and the IVTC sample have
963 frames, while the progressive has 1205 frames. 1205 * 0.8 = 964 (so right on target
there). Runtime is the same, as expected. No problem so far. So what happened to the
missing frames?
Now we go to the proof pictures I've painstakingly taken for you. Forgive the slightly
different cut on each, but they are all accurate (check the samples if you doubt).
Frame 1 (IVTC):
Frame 33 from the IVTC sample (I assure you, the 1-in-5 sample is exactly the same in
this regard): proof-tgus.from.e01.ivtc.frame.33.dvdrip.x264-deimos.jpg
Please notice discount Mark Wahlberg's eyes closed, and look at the parked truck on
the right. The front driver-side wheel is completely visible, but not much else from
the rest of the truck.
Frame 2 (IVTC):
Frame 34 from the IVTC sample: proof-tgus.from.e01.ivtc.frame.34.dvdrip.x264-deimos.jpg
Discount Mark Wahlberg has now turned more towards the camera, his eyes are finally
open, and about half the truck is now visible.
So far, this doesn't look that bad. Now though, we'll compare it to the progressively
encoded sample.
Frame 1 (Progressive):
Frame 41 from the Progressive sample (see how I even found the exact frame to compare):
proof-tgus.from.e01.progressive.frame.41.dvdrip.x264-deimos.jpg
This is an exact replica of Frame 33 from the IVTC/1-in-5 sample. No problems here.
Frame 3 (Progressive) (oh no, where'd frame 2 go?):
Frame 43 from the Progressive sample:
proof-tgus.from.e01.progressive.frame.43.dvdrip.x264-deimos.jpg
This is an exact replica of Frame 34 from the IVTC/1-in-5 sample. No problems here.
Frame 2 (Progressive) (ah, there it is):
Frame 42 from the Progressive sample:
proof-tgus.from.e01.progressive.frame.42.dvdrip.x264-deimos.jpg
Now this is where we have a problem. Discount Mark Wahlberg is slightly turned towards
the camera and looks very squinty, almost suspicious at how I found this magic frame.
He knows that if IVTC were applied, I wouldn't see his squinty little eyes. This frame
in fact, which is present in the Progressive sample but NOT present in the IVTC/1-in-5
sample, proves that if I had applied any IVTC methods, I would have dropped valuable
frames. This is something I just can't live with.
Just to prove I'm not being unfair (maybe there is a way to apply IVTC without dropping
frames during the 29.970fps segments, let's look into the sample provided via the nuke,
kindly provided in this VOB.SAMPLE as bad-sample-top.gear.us.s03e01.dvdrip.x264-deimos.mkv.
Assuming that this sample was directly re-encoded from the sample that pred with the
episode, I've also included that sample in this release as:
good-sample-top.gear.us.s03e01.dvdrip.x264-deimos.mkv.
Okay, I'll admit that the first 71 frames look awesome. There is an AMAZING 20%
reduction in dupe frames. My sample takes 89 frames to do the same job this one did in
the 71. A bang up job.
By Frame 76 though, bad-sample has already forgotten about a frame, whereas good-sample
has not lost sight of the bigger picture.
JPG proof time!
Frame 1 (bad-sample):
Frame 75 from the bad sample: proof-tgus.bad.sample.frame.75.dvdrip.x264-deimos.jpg
Please note how the telephone pole on the left of the image is almost entirely visible,
save for the smallest of slivers. Also notice the graffiti tag is still present. This
tag goes completely out of the image by frame 84 of bad-sample.
Frame 2 (bad-sample):
Frame 76 from the bad sample: proof-tgus.bad.sample.frame.76.dvdrip.x264-deimos.jpg
TELEPHONE POLE! WHERE ARE YOU?! That's right, the same telephone pole is now totally
missing. We know I haven't skipped too far ahead because the graffiti tag is still
totally there.
Again, not bad so far. If you didn't know what you were looking for, you'd never know
it was missing. Thankfully, we have:
Frame 1 (good-sample):
Frame 93 from the good sample: proof-tgus.good.sample.frame.93.dvdrip.x264-deimos.jpg
Again, like with my VOB encodes, this is the exact same frame that's present in
bad-sample. Carrying on.
Frame 3 (good-sample) (darn it, those pesky frames always go missing when you need them):
Frame 93 from the good sample: proof-tgus.good.sample.frame.95.dvdrip.x264-deimos.jpg
As expected, this is an exact match to frame 76 from bad-sample. Let's see if we can
find what's in between though.
Frame 2 (good-sample) (found it!):
Frame 93 from the good sample: proof-tgus.good.sample.frame.93.dvdrip.x264-deimos.jpg
Here we go. You can clearly see that there is just a little bit of the telephone pole
in the top left corner. Just enough so we know it hasn't run away with a storm drain.
Now, this simple experiment can be repeated again and again from either bad-sample
or from any of the other 15 episodes or 5 eligible extra features. The fact of the
matter is, removing valid frames so that some segments don't have dupe frames isn't
acceptable to me. Simply put, I'd miss those frames, because the in-car segments
would be jerky and would entirely throw off the enjoyment of the episode. I kept the
frames because I care.
So, in closing, I present what I hope is sufficient proof that this season and its
extras (sans one) could not have been redeemed by the hammer of IVTC, and in fact
should be sufficient as-is, considering the source.
Group Info
Our mission is to provide the scene with decent quality rips of old TV classics.
Greets fly out to ARCHiViST, GERUDO, INQUISITION, RedBlade, REWARD, and everyone
that helped us; you know who you guys are!
Looking for decent affil sites (no rented/colo) and rippers/cappers.
Ask around for us.