The Perfect Testing Clip

General questions or discussion about HandBrake, Video and/or audio transcoding, trends etc.
Post Reply
Honeyko
Posts: 33
Joined: Mon Sep 24, 2007 4:10 pm

The Perfect Testing Clip

Post by Honeyko »

(Sticky request.)

Image

This little bugger is a twelve-second "splash" that plays in front of every WB main film title; it contains all the nasty, codec-breaking elements you'd ever want to test in a neat & tidy 8.08mb package (of which 2.05mb consists of AC3 audio, leaving just 6.03mb of video to mangle your pride):

-- It's anamorphic, so you screwed-up if output not WS.
-- Scattered interlaced frames could cost you 10% file-size.
-- The clouds are continuously in motion; watch for blocking.
-- The logo and its lettering should smooooothly fade in and out.
-- The logo top has shallow-angle edges which may "jag" like hell.
-- The logo slightly shrinks in size, and grows opaque to the clouds.
-- The video fades in and out to black; watch for "snow" and "ghosting".
-- Deep blue sky patches awaiting your grainy pixel-storms.
-- Watch for scumminess over the "W" and the "H".
-- Colors ever so gradually shift and darken.

Download it here.

Because it's so tiny, it's possible to test oodles of variant tweaks. Because it's so short, and yet so "busy", you can watch it loop many times to spot defects quickly.

I threw Handbrake's full-size presets (other than Bedlam) at it, and rate them A to F accordingly:
(Many of these crush the 2mb AC3 into AAC, saving 1.75mb. Others drop AC3 to 160kbs, saving a bit.)

Animation (ACC/1.65mb) C ...Excellent ratio; fade-out ghosting.
Apple TV (ACC/3.78mb) F ...Unacceptable blocking
Classic (ACC/1.68mb) F ...Even more horrific blocking
Constant Quality Rate (AC3/3.88mb) A ...no obvious blemishes.
Deuz Six Quatre (AC3/2.90mb) C ...fade-out blocking.
Film (AC3/3.48mb) D ...fade-out blocking
Normal (AAC/2.39mb) D ...fade-out blocking.
PS3 (AAC/3.82mb) F ...ugly as sin blocking.
Quicktime (AAC/3.06mb) D ...blocking; dark blue sky to left is killer.
Television (AAC/2.06mb) B ...Not bad at all. Excellent ratio.

"Control": Mpeg4, AC3 192(pass-through), MKV container, fast deinterlace, Anamorphic, Chapters, 2-Pass, 5000 Avg Bitrate ....Result: 5.32mb (video size=3.57mb, or 59% of original), A++ ...indistinguishable from Mpeg2 VOB.

==//==

OS/Display: Windows XP, standard 1024x768 LCD laptop screen.
Media Player Classic with Haali Media Splitter and FFdshow video decoder.

==//==

Below, post your H.264 settings & strings that will crush this puppy. (Arf!-*splat*) ... Priorities: 1. Quality (I'm looking for "A"s), 2. Encoding Speed, 3. Size (but your 264 video can't be worse than 3.57mb, or it's worse than Mpeg4).
cvk_b
Veteran User
Posts: 527
Joined: Sun Mar 18, 2007 2:11 am

Post by cvk_b »

Just white/black/yellow/blue? I don't think this is the perfect testing clip. The horse race clip floating around the forum was better (but I didn't really like that one either because I don't like staring at ass).
Honeyko
Posts: 33
Joined: Mon Sep 24, 2007 4:10 pm

Post by Honeyko »

Well, what the short clip is insanely great for is in quickly weeding out experiments that are destined to blow chunks regardless. (The ones that pass muster you then tweak on other video.)
dynaflash
Veteran User
Posts: 3820
Joined: Thu Nov 02, 2006 8:19 pm

Post by dynaflash »

Or the iPod_killer clip rhester uses to stress out just about everything. I believe it is the first 8 minutes of HBO's "From the Earth to The Moon". Not only does it max out the above criteria, but the framerate switches so much in 8 minutes it gets nauseating watching it in the activity window, and it is so grainy that crf creates an awfully big file given its length.

As he said (and I am paraphrasing) "it is , from a technical standpoint, one of the biggest abominations in film making history" or something to that effect.

Honeyko: I ran your source through my preferred crf 70% setting which I use for all of my appleTV encodes, it looks fine though the file size exceeds what you have set as a maximum for your test. So it doesnt count.
Cavalicious
Moderator
Posts: 1804
Joined: Mon Mar 26, 2007 12:07 am

Post by Cavalicious »

I say we use the first 5mins of A Night at the Museum. But not be responsible for device blow up.
dynaflash
Veteran User
Posts: 3820
Joined: Thu Nov 02, 2006 8:19 pm

Post by dynaflash »

Cavalicious wrote:I say we use the first 5mins of A Night at the Museum. But not be responsible for device blow up.
Very true. that has been as problematic as any for me. Very complex slow pans, drives crf bitrate spikes through the roof. solid blue skys (blockiness). Decent choice imho.
redraiderbum
Enlightened
Posts: 132
Joined: Thu Jun 21, 2007 3:53 am

Post by redraiderbum »

I'm not sure if I can post this, but try the beginning of each Planet Earth episode. Interlaced, fast motion, slow motion, huge panoramas, closeups, clouds, water, dust storm, high detail (10,000 birds taking off at once) and low light. It's amazing. Am I allowed to post it?
Cavalicious
Moderator
Posts: 1804
Joined: Mon Mar 26, 2007 12:07 am

Post by Cavalicious »

@ dynaflash

Maybe we should have a Thread where people can post Test Streams. We would have to be very specific on souce format (i.e. VOBs, MPEG PSs, etc.).

This, of course, would have to be linked somewhere not associated with HB.
dynaflash
Veteran User
Posts: 3820
Joined: Thu Nov 02, 2006 8:19 pm

Post by dynaflash »

Well, if its on the HB forums, it should stay something thats in the public domain. Kind of a touchy subject. Know what I mean ?
Honeyko
Posts: 33
Joined: Mon Sep 24, 2007 4:10 pm

Post by Honeyko »

I was kinda hoping somebody might have a string to share.
awk
Enlightened
Posts: 109
Joined: Sat Mar 31, 2007 11:55 pm

Post by awk »

I've not had a chance to try it yet - but portions of 'The War' by Ken Burns are pretty tricky.

The combination of 16mm film (Ken Burns preferred format for Interviews) and it's fairly grainy image, and an interview subject with fairly heavy rouge (red makeup) applied really threw the MPEG-2 compressors that are typically used for compression during broadcast.

The result was that during the first broadcast audio sync was lost on those scenes as the MPEG-2 compressor 'gave up' bandwidth from the audio portions to 'give to' the video portion of the stream.

PBS worked pretty hard to fix the problem in subsequent evenings and I'm pretty sure they went back and fixed up the first nights problems and then resent that episode out to the member stations - but if you were watching during the 'premier' thats why you may have seen a loss in audio sync (some viewers were particularly badly affected since they were at the end of a number of uncompress/recompress sequences due to local redistribution by cable companies and the local member station)

Be interesting to see how H.264 fares with the same content.
Post Reply