Leaving decomb and detelecine on always seemed liked overkill with Blu-ray/HD sources. So I've always turned them off. However, recently I read that it's a good idea to leave them on because it can help with the slight motion blur/streaking that comes with compression.
So I did a test encode with Blu-ray episodes of the show True Detective. Left the RF constant quality setting the same as before and turned on both "d's". The bad - Nearly doubled the encoding time of each episode. The good - Playback was even smoother and footage looked cleaner. Also, file size was reduced by 150mb's. Was shocked to see 60 minutes of solid looking 720p come out @ 675mb's. Now I'd heard that leaving denoise on could lower file size. But, you also lose some detail and sharpness in the process. So I've always left it off. However, I had no idea that leaving those other filters on could drop file sizes that much when the advanced settings, speed setting, and CQ were exactly the same.
+ Reply to Thread
Results 1 to 7 of 7
This must be a pretty odd BluRay. Something sounds screwy here. Per BD spec, HD BluRay isn't telecined, unless somebody played loose with the specs.
You didn't really do a "test encode". You did a "test re-encode". Naughty.
- My sister Ann's brother
Exactly. That's why I always left detelecine off. Also, since HD doesn't usually have any interlacing that is also why I leave decomb off as well. But it resulted in better looking playback at smaller file sizes even though all settings and RF cq were left the same.
I also tried the same thing on 1080p web dl sources for The Blacklist and 24 Live Another Day. Same thing. Smaller file sizes and better looking playback. Weird..
I only use VidCoder for DVD to MKV H.264/AC3 192Kbps backup. I use RipBot for Blu-ray to MKV H.264/AC3 backup. I haven't had that good of performance with VidCoder with BDs. But with BDs and RipBot, I do two pass, set a 7900MB final size and use AC3 640Kbps for audio. With Vidcoder, I use a 19.5 Q setting. I do leave detelicine and decomb on. Both look good to me. I admit I haven't compared a BD>MKV conversion on RipBot to a BD>MKV conversion with VidCoder. Maybe with enough fine tuning the two conversions would have similar results.
No matter what you're re-encoding (even if it's an encode) you are re-encoding the same source with and without the "D"s enabled and comparing the two new encodes? You're not re-encoding an old encode while enabling the "D"s and comparing the new encode to the old one?
What about the output frame rate? Is is constant or variable etc?
I think Handbrake's detelecine filter might drop frames even for progressive video. If two frames are identical, it might drop the second frame and the first will have twice the display duration. That sort of thing. Don't quote me on that but you might want to check the log file. It should give you info on the number of frames for the source, the output frame rate and the number of dropped or duplicated frames etc.
I think the decomb filter does some sort of intelligent de-interlacing where it only de-interlaces if combing is detected, but for progressive content if it still de-interlaces sometimes that'd still blur a tad. I'm guessing a little as I'm not a Handbrake user, but I'm at a loss at to how either of those filters could make progressive video look better. Maybe a little easier to compress if they cause blurring, or maybe save a few bits if the output frame rate is variable (although I'm not sure that'd be anything significant), but better....... anything's possible but I'd be interested in some samples if you're motivated.
If lossy encoding can cause increased motion blur or streaking, which I'm not sure about either (I guess it'd be in respect to how much fine detail is lost which might increase with motion), I really don't know what either of those filters could do about it, given they're applied to the video before it's encoded.
Last edited by hello_hello; 18th Jan 2015 at 13:59. Reason: spelling
hello_hello makes some valid points. To which I feel compelled to add:
http://forum.doom9.org/showthread.php?t=154533 and https://www.videohelp.com/hd#tech.
I'd say it's easy to come up with non-compliance problems on re-encodes concerning GOP limits and other specs. You also mention 720p in your results. To me that could mean one of 2 things: (a) you started with a telecined/interlaced 1080 BluRay that was sloppy production work with combing artifacts, which you downsized to 720p (at undisclosed bitrates and frame rates). Big frames down sampled to smaller frames generally do "look sharper" for several reasons, but sharpness isn't the only factor in imaging. Or (b) Your BD source was 720p in the first place but was badly encoded -- interlace and telecine aren't allowed in 1280x720 BD.
I'm just plain skeptical about your judgments here, mostly because they sound 100% subjective. Lots of consumer reviews on Amazon think something looks "great", but an experienced eye would say the rave reviews are hogwash. Then, again, you've shown us no samples and given very slim technical specs or other details on what you're working with. How are you playing this stuff? On your PC only? Set top player/media server and TV? Granted, some BD releases really do look trashy, but it's also true that many players and TV's don't handle interlace well and are pretty klutzy with pulldown.
So I have to repeat what others proposed: post some kind of proof that all these results are so much better than what you started with. I don't conclude that you completely misfired, but I'm saying that the laws of physics don't support you at this point.
Last edited by LMotlow; 17th Jan 2015 at 09:30.- My sister Ann's brother
iTunes origin (or similar).
If so it's generally quite good quality, but not something that benefits greatly from re-encoding in respect to file size reduction without also reducing the resolution (it's already encoded at a download friendly bitrate, as opposed to a Bluray disc type bitrate). Of course that depends on the CRF value, the individual video and lots of subjectiveness, but that aside, it'd be 1080p (or 720p) so I really don't understand how enabling the "D"s could result in "better looking playback".
Last edited by hello_hello; 18th Jan 2015 at 14:46.