Import – Export : getting poor quality

[Jerry Hofmann]

I have to inject an old addage: Garbage in, then Garbage out.

It comes down to the quality of the footage you began with. There is a world of difference between the end compressions of say uncompressed SD and DV. (or lower quality than that)… so I think what I’m advocating is to start by NOT shooting any consumer formats and use the pro level stuff from the get go. You’ll be amazed how much better web streams look if they started with DVCPROHD or Uncompressed SD, or better over all the formats you’ve posted about here…

Jerry

Apple Certified Trainer

Author: “Jerry Hofmann on Final Cut Pro 4” Click here

Dual 2 gig G5, AJA Kona SD, AJA Kona 2, Huge Systems Array UL3D

[Adam Taylor]

what are your sequence settings?

when you first drop material onto the timelin, it will ask if you want to match the settings…if your first material is a really crappy file type, then you will be setting your sequence to work at that resolution and you will be converting everything else to match.

I would suggest setting your sequence to one of the uncompressed formats (assuming your hardware can handle it) and then only going down to lower resolution quality at the final stage when you output to flash or whatever you use.

As for your motion artefacts, that does sound like a deinterlacing issue. It is an issue because different video formats play back their images in different ways.
Before progressive scan, screens would show alternating fields of a picture. Thus 1 frame = 2 fields – 1 field = even scan lines 1 field = odd scan lines. A 25 frame (1 sec) clip (i’m in pal land) would actually show 50 images that your eye would blend optically into a smooth motion (persistence of vision).

Because of this, with motion you could see a slightly different image between the first and second fields in a single frame. All well and good if your playback device plays the fields in the correct order….but disastrous if it does not, and this is the root of the problem…..most video formats (but not all) play back fields with the Upper / Odd field first. DV plays back with Lower / Even fields first.

If your sequence is set to one method and your material to another then the mismatch will cause that horrible strobing.
You can either deinterlace, which will lower the quality of your video, or you could use the filter called Shift Field, which will swap the field order .

hope that helps

adam

Editor/Mixer
Character Options Ltd
Oldham, UK

[Ed Dooley]

You say you’ve read posts about interlacing, but you don’t say whether you believe it, because it sounds exactly like interlacing showing up on a computer monitor. You didn’t say whether you’re looking at it on a video or computer monitor (you should *always* use a video monitor for interlaced video). The suggestions you’ve already gotten here (checking your sequence settings, using a high quality sequence, etc) are good advice. I disagree with the suggestion (below) that de-interlacing will lower the quality (it will if you do it wrong, but it doesn’t have to). Your guide should be: what’s the final use of the project and how much interlaced (most DV for example) vs progressive (stills, graphics, etc.) does your project have? If you have a ton of interlaced, set that as your sequence, if you have a majority progressive, de-interlace your video. Do it this way and you won’t lose quality (simply de-interlacing throws away one of the fields- half the quality).

Use Compressor’s Advanced Format Conversion:
into Compressor and in the Settings (Presets) select the Apple/Advanced Format Conversions/8-bit Uncompressed(either PAL OR NTSC).
Select the Duplicate Selected Setting (one of the 3 buttons at the top left-it’s the right one). You can now edit the setting. Double click the setting
to open the Inspector. Right under Description at the top of Inspector are 6 buttons, select Frame Controls. Select Custom as the Frame Controls
(default is Off). Select Best for Resize Filter, Progressive as Output Fields, and Best as Deinterlace. Also select Best as Rate Conversion at bottom (not sure if that’s necessary, but what the hell). One of those 6 buttons at the top of the Inspector is Filters, and one option there is De-interlace,
DON’T use that one, it’s crap, and you’ve already selected all the de-interlacing you need.
Ed

[lightning ad]
Before progressive scan, screens would show alternating fields of a picture. Thus 1 frame = 2 fields – 1 field = even scan lines 1 field = odd scan lines. A 25 frame (1 sec) clip (i’m in pal land) would actually show 50 images that your eye would blend optically into a smooth motion (persistence of vision).

Because of this, with motion you could see a slightly different image between the first and second fields in a single frame. All well and good if your playback device plays the fields in the correct order….but disastrous if it does not, and this is the root of the problem…..most video formats (but not all) play back fields with the Upper / Odd field first. DV plays back with Lower / Even fields first.

If your sequence is set to one method and your material to another then the mismatch will cause that horrible strobing.
You can either deinterlace, which will lower the quality of your video, or you could use the filter called Shift Field, which will swap the field order .
“

[Steve Thomas] “After getting what appeared to be a ‘general state of lower quality’, I decided to do some trial sessions with just MiniDV captures to isolate the problem. Even in this state, I can see movement screendoor effects, and “zipper” noise.

Most of the dialogue on the forums regarding this issue seemed to point to interlacing i.e. are you watching on a TV or a Computer monitor?, The quicktime player setting set to “play high when avail”, and output format settings.
“

[Steve Thomas]

Gentlemen,

Thanks so much for taking the time to chime in. As a newBee, it is very cool to get some direction from the O.G’s.

For clarification, I’m doing everything in the computer and monitoring on my screen (not using an external monitor). Given that my intended output is 100% computer vids (QT, flash, streams), I’m hoping I don’t need to add a monitor ?

Did some experimentation on your notes.

1. On the Garbage in/Garbage out theory, I’m trying to wrap my brain around what garbage gets ADDED in various workflows. I can live with the lo quality nature of some of my input media (in some cases, it is the “home video clips/MTV burst clip” feel I’m going for). Where I get concerned is when the high quality material I add (stills, graphics) seem to get dithered down to the weakest media format in the timeline. As I’m primarily a music producer, the best analogy i can think of would be the use of a lo-fi, junky, vintage vinyl sample. When I start with one of these and then add hi quality instruments and audio around the sample, it lifts the over all fidelity up and the results are big, bright, and professional sounding (hopefully : >).

2. Xport using Ed’s custom compressor settings cured the “zipper/screendoor” problem on my miniDV captures! The quality on the DV material seemed very close to the original minus the motion artifacts. Building on my comments in point 1, it still didn’t export my still graphics very well. I made a quick comparison graphic that shows a screengrab of the original picture juxtaposed a screengrab from the QT movie exported using the custom compressor setting.

D/L here: https://homepage.mac.com/reelsteve/filechute/example.zip

As I am just getting started creating video, I’m probably going to be sticking with miniDV, camera clips, and stuff I can mangle/create within the different apps (Photoshop, AE, FCP) until I graduate to the next level where I will start thinking about getting higher quality footage.

That said, it would be great to have a workflow that cured my quality loss issue described above and give me the best quality mid to high level QT movies for playback on computers.
Not concerned with TV or DVD at the moment.

advice?

thanks!
steve

[Adam Taylor]

Steve, try thinking of it this way – garbage is not being added…quality is being removed.

to continue your music analogy….if you record your crappy lo-fi vinyl sample onto a cassette reel made of cheap low grade iron oxide, then no matter how good the shiny new instruments are, they will also sound crappy on the cassette….but if you record the same files to a cd, then the crap will still sound crap but the good stuff will shine.

Same with video sequence/timeline settings. If your sequence is set to render a low grade image, then everything you put on that timeline will be rendered low grade. Use a good uncompressed setting and things will look much better.

Its worth remembering that if you start with a compressed image, put it on a compressed timeline, add effects and then render it out to a compressed format, you are sending that initially poor image through several more compression stages, each of which has the potential to degrade the image further.

Are your graphics being deinterlaced the same way your video is? How did you create the graphics? some graphics packages will output a progressive file as opposed to an interlaced one. If yours is a progressive file, and you are deinterlacing it, then you will degrade the image, no matter which method you use – You will be removing data and then asking the software to try recreate the bits its removed in an effort to cure a problem that wasn’t there to start with.

One thing you still haven’t done is tell us what your timeline settings are? without knowing those, its very difficult to pinpoint quality issues.

Its completely wrong to believe the idea that digital somehow means no loss of quality, ever. Thats marketing nonsense to help consumer justify spending money on more unnecessary electronics for the home.

adam

Editor/Mixer
Character Options Ltd
Oldham, UK

[Ed Dooley]

If your output is not TV/DVD then you don’t necessarily need a vido monitor (but you need a well calibrated computer monitor at the very least). The tricky thing is, though, that when you have interlaced footage (as you do), the only sure way to work with it is with a video monitor. Otherwise you make posts saying “I’m getting a zipper look, what’s going on?”
In your first post you said you wanted high quality QTs to make Flash video from. You have a few options: You don’t need to go through the de-interlacing of the raw DV footage if you’re going down eventually to a small file, regular de-interlacing (throwing away one field) works OK in that situation and the quality is fine (you often wind up with a frame rate that’s lower than the raw file anyway-15FPS, for example). You can even wait until you’re compressing the file to do the de-interlacing. You can also create Flash video right out of FCP to eliminate the FCP/QT/FLV steps. The ON2VP6 plug-in for FCP is cheap and allows you to export to Flash.
https://www.on2.com/
Ed

[Steve Thomas] “For clarification, I’m doing everything in the computer and monitoring on my screen (not using an external monitor). Given that my intended output is 100% computer vids (QT, flash, streams), I’m hoping I don’t need to add a monitor ?
“