TDeint and TIVTC

[Wilbert]

There is one havily compressed image as an attachment.

You should be able to post attachments again. But I don't see any?

[tritical]

@xappy
Hm, afaik it should compile fine as long as you have malloc.h included, and since it didn't say it couldn't find it I'm not sure, but I'm no expert. You could go ahead and change the _aligned_malloc and _aligned_free to just plain malloc and free, won't make much difference, or use new and delete instead like I probably should . If you know some obvious improvements to make I'd be grateful if you would share them... the few things I've attempted to use as a post-processing to clean up left over bob artifacts do work, but also end up messing with good frames more then i would like. Either setting an ap threshold as in dgbob or testing the output frame for combing and then dumbbob deinterlacing it or storing the previous output frame from TDeint and using that to do motion adaptive deinterlacing from.

@Mug Funky
I'll still try to grab the sample if you don't mind... problem is I'm never at my own comp when you seem to have yours up and running.

[Mug Funky]

it's up now. i'm installing some software and have had to reboot a few times, but i'll refrain from that for a while.

think i'll leave this box on for a good day or so, hopefully that will give you time to d/l it.

[xappy]

@Wilbert
I tried to send the image again, but got the following error:

You have already attached this image to a post. Please refrain from uploading duplicate images. If you need to refer to this image it can be found in this thread: Another motion adaptive deinterlacer and smart bobber

@tritical
_aligned_malloc and _aligned_free do not exist in my malloc.h. Should I install some SDK or some other package? I'm using MS VC++ 6.0, but I'm totally newbie with it. However, I managed to compile some other filters...

Can I send you some (small ~10kB) pictures about those artefacts I mentioned before to your email seen in your home page?

[tritical]

@Mug Funky
Finally getting that sample, thanks .

@xappy
An email is fine... though I can see the attachment now. It would actually be easier if I could get a small sample of the video. Just looking at the pictures leaves a lot of guessing, but I'm fairly certain what caused the artifacts in the attached image. With VC++ 6.0 I think you need the processor pack to use _aligned_malloc if you don't have it. Should have mentioned that before, but I always forget that not everyone uses the same compiler I do .

[lordreign]

Hey Mug Funky, I'm just curious as to settings u use with TDeint... so I could perhaps see if they work for some of my anime. Bad field blended PAL anime that is.

Cheers.

[Mug Funky]

sure thing... i rarely do anything other than Tdeint(1,1) for top-first bobbing.

i'll only really tweak it when i get too much coming left over - i'll set Cthresh to 5 or 4 (default is 6), and/or MI to about 20 (default 32). this is only when i get combs left over, as it also risks getting shimmer on good frames.

[edit]

the last few anime encodes i did were disc 1 and 2 of Azumanga Daioh. i just left it bobbed and at 50fps, and let Xvid sort it out.

512x384, lanczos4resize, limited removedirt (just to stop removedirt from removing actual motion - i'll post the script in a jiffy.. it's probably sub-optimal, but doesn't hurt things) and encoded with CQ 3, and 4 b-frames instead of 2 (i figure 50fps has twice the redundancy, so i could get away with it). all other settings are at default.

episodes are coming out at around the 110MB mark (sans sound) which is quite acceptable to me.

[offtopic]
by the way, speaking of Azumanga Daioh, you should see the rating disc 1 gets... the stupid australian office of film and literature classification gave it an MA 15+ rating... disc 2 is a G. that puts a _very_ tame anime on the same rating as Kill Bill 2, which is utterly stupid. however, it costs Madman several thousand dollars to have these discs re-assessed.
[/offtopic]

[Chainmax]

tritical: I finally was able to encode the clip I promised in my "need help IVTCing..." thread. You can download it here. I'd appreciate it if you could comment on the image quality of it besides assessing TDeint's work as a fieldmatcher.

Have you considered posting this thread in Avisynth usage? I bet more people would see it in there than in here.

[tritical]

I'd appreciate it if you could comment on the image quality of it besides assessing TDeint's work as a fieldmatcher.

I'm not sure exactly what you mean, but after looking at the clip it seemed like all the fields had been matched correctly where possible and the ones that couldn't be were deinterlaced (the fades). On the subjective side, I thought it looked pretty good considering the source .

As for posting it in usage, I plan too, but I still haven't been able to find a good method of post-processing to remove left over artifacts as in xappy's picture. In the end, I think I may simply add an optional artifact threshold as in dgbob and be done with it. The major problem is that those type of artifacts are inherent to motion adaptive deinterlacing/bobbing, and detecting them will also detect lines/edges pretty much mitigating the benefits of motion adaptation. I've also been working on a new method of edge-directed interpolation that is considerably better then TDeint's current method, but haven't gotten all the kinks worked out just yet. I would like to get the above two items finished and added before posting it in usage.

On a side note about field matching, the blind field matching of TDeint is actually better then that of Telecide, especially when it comes to the small movement problem (flapping mouths, etc...). TDeint's field matching is currently lacking one improvement that would optimize it for field matching of telecined sources vs finding the overall most similar field for deinterlacing as it does now. For that reason, and that I was a little tired of working on TDeint , I started working on a stand alone field matcher called TFM a day or two ago. The only other field matcher I've seen that can match small flapping mouths in anime as well as it is bruteivtc. Its main advantage over bruteivtc is that is it about 9-10 times faster and supports user overrides like telecide. Anyways, that's probably more then you wanted to know... but I wanted to show that I'm still working on this filter.

[Mug Funky]

your efforts are very much appreciated

if you need some problem samples (i've got some flapping mouth stuff that even bruteIVTC baulks at), you know where to get them. i'll just have to remember to leave my computer on longer (maybe i'll start some encodes )

[Chainmax]

Originally posted by tritical
I'm not sure exactly what you mean, but after looking at the clip it seemed like all the fields had been matched correctly where possible and the ones that couldn't be were deinterlaced (the fades). On the subjective side, I thought it looked pretty good considering the source .

As for posting it in usage, I plan too, but I still haven't been able to find a good method of post-processing to remove left over artifacts as in xappy's picture. In the end, I think I may simply add an optional artifact threshold as in dgbob and be done with it. The major problem is that those type of artifacts are inherent to motion adaptive deinterlacing/bobbing, and detecting them will also detect lines/edges pretty much mitigating the benefits of motion adaptation. I've also been working on a new method of edge-directed interpolation that is considerably better then TDeint's current method, but haven't gotten all the kinks worked out just yet. I would like to get the above two items finished and added before posting it in usage.

On a side note about field matching, the blind field matching of TDeint is actually better then that of Telecide, especially when it comes to the small movement problem (flapping mouths, etc...). TDeint's field matching is currently lacking one improvement that would optimize it for field matching of telecined sources vs finding the overall most similar field for deinterlacing as it does now. For that reason, and that I was a little tired of working on TDeint , I started working on a stand alone field matcher called TFM a day or two ago. The only other field matcher I've seen that can match small flapping mouths in anime as well as it is bruteivtc. Its main advantage over bruteivtc is that is it about 9-10 times faster and supports user overrides like telecide. Anyways, that's probably more then you wanted to know... but I wanted to show that I'm still working on this filter.

Well, you wanted the clip in order to see how the clip looked with TDeint+Decimate. I wanted you to comment on the IQ of the clip as well which you did, that's all. Anyway, if you release TFM what will become of TDeint? Will future releases of it be composed just of its bobbing functionality? Also, could TFM have decimation capabilities? It would be nice to finally have an alternative to Decomb and SmartDecimate for automatic IVTCing...

[tritical]

@Chainmax
I'm not sure about adding decimation mainly cause I don't know of any real improvements to make over decimate, aside from maybe using 3 sets of offset blocks at +16x, +16y, and +16x+16y to better catch small movement (but it would make it slower) and making a small change to mode=2 decimation to fix an inherent limitation that can cause jerkiness when going from panning to still and still to panning scenes. I don't think TFM will change TDeint much at all, it will still have the option to weave fields, but I will probably turn it off by default and make the defaults of TDeint better for true interlaced material as opposed to how they are atm with the defaults being a compromise between settings for several different types of material.

@Mug Funky
Would always take any problem samples your willing to send .

[Chainmax]

If you can find the time to do it I'm sure it would be appreciated. Some time ago, I saw some VHS captures that were IVTCed via Decomb, SmartDecimate, TDeint (old version, 0.9.2 IIRC)+BlendBob and TMPG manual IVTC. The manual IVTC looked showed much more fluid motion than the other options. I know it's pretty much impossible that an automatic method yields equal results than a manual one, but any improvement is welcome in my opinion.

[tritical]

TFM v0.9[link removed, new version is linked farther into thread], its a field matcher like uncomb/bruteivtc/telecide... It supports 2-way, 3-way, 2-way + 3rd on combed, 2-way + 3rd on combed + 4th/5th on combed, and 3-way + 4th/5th on combed matching strategies. It doesn't atm do any post-processing so if no good match exists it will output combed frames. Also, there is no included decimation function atm either so you'll need to follow it up with decimate() or something similar to achieve an ivtc. It does support quite extensive manual overrides (specific frames, patterns over frame ranges, changing field/order/mode for ranges of frames, etc...). There are really only two parameters that require setting, order (the field order) and mode (sets the matching strategy). Anyways, its just a start... and TDeint development isn't dead yet either. Also, if anyone has some hard to correctly field match clips (mouth problems, small movement, etc...) that TFM doesn't correctly handle I'd definitely appreciate a small sample.

[Mug Funky]

hmm.. seems to work good on my initial test.

but i inadvertently found a bit that may not even be possible to match... i'll send it if i can't find a good setting for it. as you might have guessed, it's from the same show as the other samples i have difficulty with (this is like Hell's Impossible To Backup DVD Of Doom, and i still haven't even been able to do a straight encode without a crash ).

also, is it possible to get double-rate output? this is useful for standards-conversion for PAL DVD (best way to maintain maximum vertical detail on hybrids). i guess i could use the same trick as with telecide for now.

thanks for this plugin

[edit]

btw, maybe this should have it's own thread?

[edit 2]
hmmm... that awful sample i mentioned seems to be a result of poor compositing - the sparkly bits look like they were added at editing time, but were moved up 1 pixel, reversing their field order with respect to the film underneath. ouch. it might still prove useful for Tdeint, though.

[Chainmax]

Thanks for this release Tritical . Unfortunately I don't have any difficult stuff to try it on other than the Spawn DVD and seeing what a POS it is, probably TDeint bobbing+BlendBob is the only way to go.

I concur with Mug Funky about TFM needing its own thread. I see that you are considering adding a decimation function, will that be similar to vanilla Decimate or FDecimate?

[Mug Funky]

Chainmax - if you can find the R1 release of Astroboy from the '80s, that'll give you IVTC nightmares. you'll think Spawn was a pleasant walk in the park on a mild summer's day. my advice for discs like these is to not buy them... (the R4 one is okay though, and the R2 one is the best).

[Chainmax]

There's a DVD release of the 80s Astroboy? Any ideas where can I find the R2 version?

[Mug Funky]

huh. i can't find it through the normal means, but i know it exists. i'll ask around for you

[tritical]

@Mug Funky
Double rate output can be added pretty easily so I'll go ahead and do that. The other options that TFM is currently missing and that will be added at some point are post-processing of combed frames and an exclusion band for ignoring subtitles or a logo.

@Chainmax
My original idea was for something similar in operation to fdecimate/changefps with a few modifications (mainly dealing with the radius around the calculated frame in which it will look for the frame to return). The nice thing about it is it is simple, pretty fast, and allows decimation to any framerate. However, there are downsides of such a method compared to 1-in-N decimation when you can achieve the desired framerate with 1-in-N decimation. Namely, 1-in-N decimation doesn't require setting a duplicate threshold or even trying to find duplicates if simply removing the most similar frame in each N frame cycle. This type of method usually works well with normal movies, but not anime/cartoons where for correct decimation you do need to detect duplicates. Also, when doing 1-in-N decimation and tracking duplicates (requires a duplicate threshold like the first method) you can correctly handle scenerios like this: http://neuron2.net/ipw-web/bulletin/...cimate+pattern which is not possible with a changefps/fdecimate type of method or with simply removing the most similar frame in a cycle. User overrides of the decimation would be easier to support with 1-in-N decimation then the other method, as well. So after thinking about it, I am planning on adding both an arbitrary frame rate decimation mode and a 1-in-N decimation mode since they each have advantages. I am also planning a two-pass mode for use with hybrid material (film, 60i, and 30fps progressive in one clip) and mkv's vfr support. If anyone else has ideas for a decimation filter don't be shy .

A new thread might be useful, but thats more work then simply posting in this one . Once TFM gets double rate output, post-processing, and an exclusion band I will definitely start a new one.

EDIT: Forgot to mention this, but there is also the possibility of having an N-in-M decimation mode, where like the 1-in-N method (and unlike the fdecimate/changefps method) you consider full cycles at a time. It would offer some advantages (maybe the best of both worlds?), but also has new problems and I haven't had time to think about it enough.