DVFilm maker vs. Joes De-interlacer
 

I own Joes plugins and have been using them exstensively with my footage.After reading some previous threads here I tried the demo of DVFilm maker. I cant tell a difference in a side by side ccomparison is there really any? Why would a stand alone De-interlacer be superior to a plug in type? And both joes and DVFilm handle motion differencing but Joes seems to be able to control it more though I usually turn it off due to mostly all handheld shots.Any comments from users of both or either one?

Joes de-interlace is good - it's certainly no worse than DV Film maker, which is slow and expensive. There's also a de-intleracer in my Film Effects (30p mode) which you could take a look at and compare.

Graeme

Cool I will check it out thanks!

<<<-- Originally posted by Graeme Nattress : Joes de-interlace is good - it's certainly no worse than DV Film maker, which is slow and expensive. -->>>

Sorry I haven't had a chance to look at your Film Effects package yet Graeme. But you got me thinking about this, so I finally had a chance to do a comparison since I own both products.

As far as quality, I see very little difference using each program's default settings. There are some small things that are different, but overall they both seem to do the job fine. But I take issue with the notion that DVFilm Maker is slow. After I read your comments I thought that maybe I had been wasting a lot of time de-interlacing with DVFilm Maker when I could have just stayed in FCP and rendered Joe's de-interlacing. But when I tried using Joe's filter, FCP estimated HOURS to render a clip that I knew would only take MINUTES to do with DVFilm Maker.

So I ran a controlled test with a 30 second clip. It took Joe's de-interlacer 378 seconds to render the clip, while DVFilm Maker only took 128 seconds, a three-fold speed difference. As noted above, both programs used their default settings (Joe's de-interlacer was set to "fast interpolate"... can only guess how long it would take at "normal" speed ;-)

Anyway, you will find specifics and sample frames here

Thanks for the comparison Boyd. It does seem that DVFilm Maker is much faster than I had thought from my brief look at it. However, I can't see much, if any difference between the images you posted in your comparison. Did you notice much difference??

Graeme

No, I didn't notice much difference and that's why I wrote
:-)

Thank you for posting that Boyd. I'm curious what your numbers would be for Joe's De-Interlacer with FCP 4.02. In my own highly unscientific testing, I've found De-Interlacer is more than twice as fast in FCP 4 as it was in FCP3. (and I was happy about the speed in 3!)

I just did a 10 second test at default settings (NTSC DV):

• FCP 3: 217 seconds
• FCP 4: 104 seconds

(867mhz G4 PowerBook, 10.3 with a bunch of other apps running)

Multiplying that difference against your numbers, I'd expect FCP 4 to render De-Interlacer in about 181 seconds. That puts DVFilm Maker at only about 1.4x faster, much better than 3x faster. At any rate, the time savings of working inside FCP and not having to import/export and possible reconnect all your media should also be considered.

The 'Normal De-Interlacing' method uses a very precise field-preservation algorithm but the difference is basically non-existant.

Any FXScript effect is going to be at the mercy of FCP's rendering engine regarding speed. The benefit is that you don't have to manage clips in and out of FCP, the disadvantge is speed and not having access to the unlimited tools available in Carbon or Objective C. Thankfully FCP is only getting better and faster.

Joe Maller
www.joesfilters.com
Joe's Filters for Final Cut Pro and Final Cut Express

Thanks Joe, this also gives me heart that FCP4 will be a worthwhile upgrade (although one that I'll probably wait on for awhile :-) Your points about staying in the FCP environment the whole time are well taken. How is the render time affected by the amount of movement in the source clip? I deliberately used an example with lots of movement where the camera was panning to follow two dancers.

My tests were done just to satisfy my curiousity as to the best approach for deinterlacing a couple large projects I'm working on. I should also point out that I was using the anamorphic setting in FCP3. This wouldn't make any difference to DVFilm Maker (which ignores this flag and treats all footage the same), but I wonder if it was a factor in FCP?

Anyway, I hope you don't think I was trying to find flaws in your excellent filter package. It's an incredibly useful collection of tools that I use regularly, and it should be on everyone's "must have" list!

Yes - FCP 4 does seem to be more optimised than ever before, and it makes use of dual processors, which is becoming increaseingly vital. I don't know exactly about Joe's de-interlacer, but I don't think it's matters what the source footage is, anamorphic or otherwise, or what the level of movement is - at least that's how Film Effects and it's de-interlacer behaves.

Joe's also right about workflow - I'm doing more and more sfx inside FCP rather than go into After Effects because the FCP workflow is much quicker.

Another advantage of the FCP workflow is that you can stop a render half-way through, and come back to it later. Stopping a render in DVFilm Maker means starting again...

Graeme

The De-Interlacer uses differences between the current, previous and next frames to generate a mask highlighting only the changed pixels. The term 'adaptive' is probably misleading, Areas of high motion don't take any more or less time to render, they just make it more difficult to pull a good motion mask. I added threshold and soften controls to allow for precise adjustments which balance the needs of a largely non-moving scene with something like the bicycle example from the De-Interlacer page.Anamorphic is mostly just a display preference for FCP. The video itself is the same as standard frame; 720x480 with 60 interlaced fields. Not at all. Craft is extremely important to me but there's always room for improvement and there's always something I didn't think of or notice. This filter specifically benefitted from a mediocre review of the previous version. After reading that, I was determined to rewrite the filter so it would compete against the best options available. There are still a few things I could do though...

Joe Maller
www.joesfilters.com
Joe's Filters for Final Cut Pro and Final Cut Express

This thread has become a pretty darn good fly-over reference to some of the basic issues concerning one of the most influential technical aspects of giving a non-video look to video: deinterlacing.

Thanks very much for you remarks, Joe. I have your filters and use several of them routinely. They are an excellent value and very well crafted.